The Ivy Institute | College Admissions Research | November 2026
Executive Summary
College applications rarely become “stuck” at only one dramatic moment. They more often slow through a sequence of smaller transitions: choosing a viable story, turning notes into a draft, matching that draft to a live prompt, incorporating feedback, satisfying a word limit, coordinating outside materials, entering finished work into the application, and reviewing the assembled submission. This study examines those transitions in a deliberately narrow internal sample rather than trying to infer national behavior.
The Ivy Institute studied application work undertaken by 10 seniors across two application cycles. The final analytic period ran from June 2 through November 5, 2026, and from August 7 through December 26, 2025. The unit of analysis was a stage-specific task episode: a documented application task with a dated starting state and an observable next step. Tasks introduced at the last meeting could be retained with no follow-up. A student, application, essay, meeting, or calendar day was not itself an episode. After auditing 42 preliminary rows, two overlapping rows were removed so that an alternate concept branch and an overlapping handoff did not count as independent bottlenecks. The final sample contained 40 episodes.
Students developed topics and drafts in 16 of 40 episodes (40%). Assembly and management of the college list accounted for 12 more (30%). Verification, review and coordination accounted for the other 12 (30%). A completed essay could still need a check against the live prompt, a response from a reader, or entry into the college application. The pattern describes where work remained. The evidence supports counts of pending tasks. Ranking actual delays would require a comparable start, due date and submission date for each task. The Ivy Institute, internal materials, data, and information.
Two dated observations established an identifiable milestone for nine episodes. In the other 31, follow-up ended, work was partial, or the evidence left the next step uncertain. Some later conversations covered other tasks. Their dates tell us when contact took place but cannot establish how long an earlier task stayed open.
For the nine episodes with both a defensible start and an observed milestone, the median elapsed time was 12 calendar days. The inclusive interquartile range was 7 to 35 days, and the full range was 1 to 75 days. These values are descriptive only. They combine different endpoints, including draft development, review handoff, near-final status, and submission. They measure elapsed calendar time between documented states, not active work time, effort, procrastination, or counselor response time. The records do not support a universal claim that an application task “takes 12 days.” They support the more modest conclusion that, in this sample, documented transitions frequently spanned more than a single work session and varied widely.
Drafting created the largest number of observed task episodes, but assembly, verification, and coordination together produced 24 of 40 episodes. A family that plans only for writing may underestimate the work required after the writing exists. A more realistic plan assigns separate milestones for content creation, prompt verification, revision, word-limit compliance, application entry, full-preview review, and submission.
This report does not estimate national prevalence, evaluate admissions outcomes, rank students, assign blame, or test whether advising caused faster completion. The sample is small, selected, and drawn from the records of a commercial advising organization that conducted and is publishing the study. Students who work with The Ivy Institute may differ from applicants generally, and the available documentation was created for advising rather than research. The analysis is best read as a transparent operational case series that identifies where planning buffers may be useful.
Practical headline: families should treat “the essay is written” as a midpoint, not the finish line. A backward plan should reserve distinct time for verification, coordination, assembly, and a complete application review. The safest workflow measures status through observable deliverables—such as a prompt-checked draft, a word-compliant version, a complete preview, or a confirmed submission—rather than relying on broad labels like “almost done.”
Key findings at a glance
- Study population: 10 seniors across the 2025 and 2026 application cycles.
- Final analytic sample: 40 stage-specific task episodes after two overlapping preliminary rows were removed.
- Narrative development: 16 episodes (40.0%).
- Assembly and portfolio management: 12 episodes (30.0%).
- Verification, review and coordination: 12 episodes (30%).
- Observed milestone: 9 episodes (22.5%).
- Without a verified milestone: 31 episodes (77.5%).
- Reliable elapsed-time subset: 9 episodes; median 12 calendar days; inclusive interquartile range 7–35; range 1–75.
- Interpretation: elapsed days are not active hours, right-censoring is not failure, and these internal results are not nationally representative.
1. research questions
This study addressed five questions.
- At which stages did application tasks most often show documented friction in this internal sample?
- How much of the observed friction occurred during narrative development, and how much appeared after a draft or substantive content already existed?
- What status was visible at the last dated observation: an observed milestone, an open right-censored task, or an intermediate or ambiguous state?
- Where both endpoints were reliable, how much calendar time elapsed between the first documented state and the observed milestone?
- What planning practices follow from the observed process without claiming that advising caused completion or improved admission outcomes?
The questions are intentionally operational. They do not ask whether a student was admitted, whether a particular essay was “good,” whether a family worked hard enough, or whether a counselor’s intervention changed an admissions decision. The available evidence can describe workflow states. It cannot identify causal effects on admission.
2. background: why applications are a pipeline
The phrase “college application” sounds singular, but a submission is an assembled product. The Common Application’s official first-year guide separates the process into gathering materials, creating an account, adding colleges, engaging supporters, understanding requirements, planning essays, and submitting. It also explains that colleges can set different requirements for deadlines, essays, courses and grades, testing, portfolios, writing supplements, and recommendations. Submission itself includes review, payment where applicable, and the act of submitting. In other words, the platform’s own guidance describes a multi-stage process rather than a single writing assignment. [Source]
This structure matters because progress at one stage does not guarantee readiness at the next. A personal statement may be polished while activities remain unentered. Supplemental content may be strong but exceed the permitted length. A school-specific response may be vivid yet answer an outdated or misread prompt. A nearly complete application may still need official materials, a full preview, payment, or an electronic confirmation. A family may reasonably say that the application is “almost done” while several different readiness conditions remain unresolved.
Application volume makes these handoffs more consequential. Common App reported 1,527,328 distinct first-year applicants and 10,786,782 applications in the 2025–2026 season, an average of 7.06 applications per applicant. Through January 1, 2026, the organization reported 1,281,012 applicants and 7,607,278 applications. These national totals do not describe the Ivy Institute sample, but they show the scale of a system in which applicants frequently manage multiple institution-specific requirements. [Source] [Source]
Research on common applications also suggests that reducing duplicated form-filling is not the same as removing every application barrier. Delaney and Odle’s 2025 study of state-level common-application policies describes the number, diversity, and complexity of applications as potential barriers, while finding no positive or negative effect on full-time-equivalent public enrollment in their design. Their result is a useful warning against an overly simple causal story: a common platform can reduce some administrative repetition without necessarily changing enrollment outcomes by itself. [Source]
RNL’s 2024 survey of high school students reported both cost concerns and application difficulty among reasons for leaving applications unfinished. The survey asked about students’ decisions. The current study used observations about specific tasks. A pending task cannot establish a student’s reason for stopping an application.
Task-timing research offers another useful but limited lens. Buehler, Griffin, and Ross’s 1994 studies found that people often underestimated their own task-completion times and focused on plan-based scenarios more than relevant past experience. The present study did not measure predictions, optimism, or psychological mechanisms, so it cannot diagnose a planning fallacy in any student. The research is relevant only to the practical recommendation that future plans should use observed prior timelines and explicit buffers rather than best-case intuition alone. [Source]
The distinction between dependence and blockage is equally important. Common App states that students may submit an application before counselors or teachers submit their forms, while advising them to verify that recommenders can meet the institution’s deadline. A recommendation that is still pending is therefore a coordination item, but it may not block the student-controlled submission. Treating every outstanding dependency as a stop condition can create unnecessary delay; ignoring a true institution-specific dependency can create a different risk. The planning task is to determine which requirements are blocking, which can proceed in parallel, and which need monitoring. [Source]
3. methods
3.1 Study design
This was a retrospective descriptive case series of application work among Ivy Institute students. The organization conducted, funded, and is publishing the study. The design was chosen to describe observable workflow states in a small advising sample. It was not preregistered, randomized, prospective, or independently audited.
The study used a two-cycle window because sufficiently detailed, ordered observations were available in both periods. The 2026 observations ran from June 2 through November 5. The 2025 observations ran from August 7 through December 26. Dates indicate the first and last relevant observations included in this analysis, not the beginning or end of each student’s entire advising relationship.
3.2 Study population
The final sample included 10 seniors across the two cycles. Inclusion required enough dated information to reconstruct more than one application-period state and to distinguish at least one identifiable task from the student’s overall workload. Students were not selected to represent all Ivy Institute clients, all applicants, or all application types. The sample overrepresents students for whom the internal record was sufficiently detailed to support ordered coding.
The student was not the unit used for the main counts. A single student can work on several applications and can encounter different stages within one application. Conversely, one application can contain multiple essays, administrative sections, and external requirements. Counting only students would hide the process; counting every mention would inflate repeated discussion. The study therefore used a stage-specific task episode.
3.3 Unit of analysis: the stage-specific task episode
A task episode was defined as a bounded application task with:
- a recognizable object of work, such as a personal statement, a supplemental set, a school-research step, a recommendation dependency, an activities section, or a full application assembly;
- a dated first state that described what remained or what stage had begun; and
- at least one dated later observation, unless the task first appeared at the final observation and was explicitly retained as a zero-day right-censored episode.
The episode is a process unit, not an allegation that someone delayed. It identifies a transition the records allowed the study to observe. When the same application moved from drafting to trimming and then to full assembly, those could be distinct episodes because they were different stages with different completion conditions. When two preliminary rows described the same stage of the same underlying task over overlapping dates, they were merged or one was removed.
The preliminary inventory contained 42 rows. During final adjudication, two overlapping rows were removed. One described an alternate concept branch within the same personal-statement development process; counting it separately would have treated exploration of two concepts as two independent bottlenecks. The other described an overlapping handoff and review for the same main essay during the same short interval; the clearer review episode was retained. The final denominator was 40 episodes.
3.4 Stage coding
Each episode received one primary stage code. Granular codes were consolidated into three public reporting groups to reduce sparse cells and protect confidentiality.
Narrative development included topic selection, brainstorming, outlining, drafting, and development of unfinished essay sets.
Tailoring and verification included prompt alignment, word-limit compliance, and school-specific research. These tasks answer the question: does this content fit the actual requirement for this institution and this cycle?
Review and coordination included revision or feedback cycles and external dependencies. External dependencies were coded only when the record identified an outside action or material relevant to the task. The code does not imply that the outside item legally or technically prevented submission.
Assembly and portfolio management included activities or other application-section completion, transfer of content into the platform, full-preview review, submission preparation, and management of multiple applications or remaining schools.
The coding hierarchy prioritized the stage visible in the episode rather than the form of the underlying document. For example, an essay that already existed but needed to be cut to a limit was coded as tailoring and verification, not narrative development. An application whose essays were nearly complete but not entered or fully reviewed was coded as assembly.
3.5 Outcome coding and right-censoring
Every episode was classified at the last dated observation into a detailed status, combined into two public reporting groups.
Observed milestone meant the record documented a defined transition such as a developed draft, a handoff for review, a near-final deliverable, or a submission. The exact milestone varied by episode. This group should not be read as “all work complete.”
Follow-up was right-censored at the last observation that actually established the task was open. A later conversation that omitted the task could not extend that window. We group open tasks with partial progress and unclear endpoints in the public status table.
Intermediate or ambiguous status included partial completion, a documented need for rework, or a state that could not be classified more precisely. These episodes were kept rather than forced into a success/failure binary.
3.6 Elapsed time
Elapsed calendar days were calculated only when both a first state and a defined later milestone were reliably dated. Nine episodes met that stricter standard after overlap adjudication. The calculation was the difference between the two calendar dates.
Elapsed time is not active work time. The study did not continuously observe students, count hours, measure pauses, or separate weekends, school days, family obligations, counselor review time, or time spent waiting for an outside party. The endpoint also varied. Consequently, the nine values are presented as a descriptive distribution rather than pooled into an expected completion time.
For unresolved tasks, a later conversation about a different topic provided no evidence of the earlier task’s status on that date. Such episodes were retained in the stage counts and excluded from the timing analysis.
3.7 Missingness and exclusions
The study did not assume that a task was complete because it disappeared from later discussion. Absence of a later mention could mean completion, reprioritization, an unrecorded event, or simply the end of the available record. Such cases remained right-censored or indeterminate.
The study did not infer active work from timestamps, infer family motivation, or infer financial circumstances. It did not convert words such as “nearly,” “mostly,” or “95%” into numerical completion except when reproducing the status category in aggregate. It did not infer that a planned submission occurred. It required an observed milestone for that classification.
3.8 Privacy and suppression
Public tables combine verification, review and coordination across students. The public timing results use summary statistics. Detailed task and student information stays in the restricted ledger.
3.9 Organization role and conflict disclosure
The Ivy Institute is a commercial college-admissions advising organization. It selected the question, supplied the internal evidence, conducted the coding, interpreted the results, and is publishing the report. The organization has an interest in presenting its work as useful. These conditions create a risk of selection, interpretation, and publication bias. Readers should therefore treat the study as an internally conducted operational analysis, not independent verification of service quality.
4. results
4.1 Sample and episode inventory
The final analytic sample contained 40 episodes across 10 seniors. Each student contributed more than one episode because applications contain multiple stages and the study required enough ordered information to observe process transitions.
The records covered personal statements, supplemental sets, school-specific research, prompt and length requirements, recommendations and other dependencies, activities and application sections, full-application review, and multi-school portfolio management. The public report intentionally does not publish the number of episodes for individual students. The study’s purpose is to describe the process, not compare clients.
Table 1. Task episodes by public stage family
Developing the narrative · 16 of 40 (40%)
Assembly and portfolio management · 12 of 40 (30%)
Verification, review and coordination · 12 of 40 (30%)
| Stage family | Episodes | Share of 40 |
|---|---|---|
| Narrative development | 16 | 40.0% |
| Assembly and portfolio management | 12 | 30.0% |
| Verification, review and coordination | 12 | 30.0% |
Source: The Ivy Institute, internal materials, data, and information.
4.2 Narrative development was the largest single stage family
Sixteen of 40 episodes involved narrative development. The category included tasks that began with rough ideas, a need to choose a topic, an unfinished draft, or a set of responses still being developed. This finding is consistent with the intuitive view that writing is a major part of selective-college applications, but the coding adds two qualifications.
First, “drafting” was not one uniform state. Some episodes began before a topic was selected. Others began after brainstorming but before prose. Others involved a set in which some responses existed and others did not. A progress label such as “working on essays” therefore conceals meaningful differences. The next observable deliverable should be stated: topic approved, outline complete, first full draft, revision ready, or final word-compliant version.
Second, the category does not indicate that the task was continuously active for the entire interval. A 35-day elapsed period may include school, work, other applications, waiting for feedback, or no recorded activity. The data support a statement about dated states, not intensity.
Narrative development appeared across multiple students. To account for differences in the number of episodes each student contributed, we repeated the calculation giving every student equal weight. The sensitivity analysis below reports both versions.
4.3 Assembly and portfolio management accounted for 30% of episodes
Twelve episodes involved assembly and portfolio management. This was tied for the second-largest reporting group and included tasks such as finalizing activities, entering content, preparing a complete preview, moving a group of applications toward submission, or tracking remaining schools.
The distinction between content readiness and submission readiness was repeatedly visible. In several episodes, substantive essays were complete or nearly complete while platform entry, final checks, or another section remained. In others, a complete application review was planned but not yet observed. These are not clerical footnotes. A wrong field, omitted section, stale prompt, missing preview, or unconfirmed submission can matter even when the writing is strong.
Assembly or portfolio-management episodes appeared in all 10 students’ records. That does not mean all students experienced a severe bottleneck. It means every selected case contained at least one observable transition beyond drafting. The result supports a planning model that allocates explicit time to assembly rather than treating it as instantaneous.
Portfolio management is especially important when work is distributed across institutions. The same personal material may be reusable, but prompts, lengths, required sections, and deadlines differ. Reuse can reduce writing load while increasing the need for verification. A response adapted from another school may be efficient only if names, programs, claims, and prompt fit are checked again.
4.4 Review and coordination
Review and coordination included feedback on writing and follow-up about outside materials. Review episodes often ended at a handoff rather than at submission. This is why the study calls the endpoint a milestone. A draft ready for counselor review is meaningfully different from an unfinished draft, but it is not necessarily ready to submit.
Coordination episodes included outside materials whose status was relevant to the application. The records sometimes showed one element complete while another remained unconfirmed. The proper operational question is not simply “is the recommendation done?” It is: who owns the next action, what does the institution require, can the student submit in parallel, and what evidence will confirm receipt?
Common App’s official support guidance is useful here. It states that a student can submit before counselor or teacher forms are submitted, while recommending follow-up so the school’s deadline is met. A workflow that marks every recommendation as a hard blocker may delay student-controlled work unnecessarily. A workflow that ignores recommendations may miss a real requirement. The process map must distinguish parallel dependencies from sequential blockers.
4.5 Tailoring and verification
Some tasks involved prompt fit, word limits, or school-specific research. A draft can be well written and still be unusable if it answers the wrong question, exceeds the limit, or relies on generic or inaccurate school information.
The category also illustrates why “percent complete” can mislead. A response that is 95% written but needs a prompt-level reframing may require more than 5% of the intellectual work. A 250-word response at 340 words may contain all the desired substance but still need difficult decisions. Research may be complete in volume but weak in relevance. Verification requires binary checks tied to the live requirement, not a global feeling of progress.
4.6 What could be verified at follow-up
Table 2. Status at the last dated observation
| Status | Episodes | Share of 40 |
|---|---|---|
| Observed milestone | 9 | 22.5% |
| Without a verified milestone | 31 | 77.5% |
Source: The Ivy Institute, internal materials, data, and information.
For most tasks, the study ended without evidence of the next defined milestone. An applicant could have finished independently, reconsidered an application, or simply moved on to a different topic at the next meeting. The study cannot choose among those explanations.
Nine episodes reached an observed milestone. The milestone could be development of a strong draft, a review handoff, a near-final application state, or a verified submission. Because the endpoints differ, the group should not be translated into a 22.5% “completion rate.” It is a count of episodes for which the next coded state was visible.
Partial progress, requests for rework, and incomplete follow-up all contributed to the group without a verified milestone. Including these tasks protects the stage distribution from the bias that would result from selecting only work with visible endpoints.
4.7 Elapsed calendar days varied widely
Nine pairs of dated observations documented a transition. The median interval was 12 calendar days, with an inclusive interquartile range of 7 to 35 days. A student could have begun before the first observation or reached the milestone before the second. These figures describe observation intervals.
Table 3. Elapsed calendar days for the reliable milestone subset
| Statistic | Value |
|---|---|
| Episodes with both endpoints | 9 |
| Median | 12 days |
| Inclusive interquartile range | 7–35 days |
| Range | 1–75 days |
| Mean | 22.8 days |
Source: The Ivy Institute, internal materials, data, and information.
The values are deliberately not broken out by student or school. The study also does not compare stage-specific medians because the resulting cells would be small and the milestone definitions differ.
For 31 episodes we could not verify a dated milestone that met the study definition. In some cases the last task-specific observation still described work underway. In others the task disappeared from subsequent discussions or its goal changed. These different forms of missing follow-up preclude an estimate of typical completion time across the full group.
4.8 The apparent bottleneck changes as the application advances
For planning, verification and review can be considered separately:
- Narrative development: decide what to say and produce substantive content.
- Tailoring and verification: ensure the content answers the live requirement and fits the school, prompt, and limit.
- Review and coordination: incorporate feedback and manage parallel outside materials.
- Assembly and portfolio management: enter, preview, reconcile, and submit the complete package while tracking other applications.
The observed distribution was not evenly spread across these stages. Narrative development contained the largest number of episodes, but the work outside narrative development together accounted for 24 of 40 episodes (60.0%). This is the report’s most actionable pattern. If a plan reserves nearly all available time for brainstorming and drafting, it leaves too little room for the majority of observed stage episodes that occurred elsewhere in the pipeline.
This does not mean later stages require more hours. The study did not measure hours. It means later stages generated more than half of the identifiable episode count and therefore deserve explicit milestones and buffers.
4.9 Checking whether the result depends on a single student
A student who attended more meetings could contribute more observable work. To test how much that mattered, we divided each student’s episodes among the three stages. We then averaged those ten personal distributions, giving each student one tenth of the weight. This asks how a typical share changes when the student becomes the unit of weighting. We also removed each student in turn and recalculated the episode percentages for the remaining nine.
| Developing the narrative | 40.0% | 39.8% | 35.3–45.5% |
| Assembly and portfolio management | 30.0% | 34.9% | 27.0–33.3% |
| Verification, review and coordination | 30.0% | 25.3% | 21.2–34.3% |
Source: The Ivy Institute, internal materials, data, and information.
The endpoints of the ranges describe ten deliberately chosen recalculations. They quantify sensitivity to the participating students. They do not provide margins of error for a population estimate. Repeating the exercise with the same ten clients cannot correct the original selection for richly documented cases.
The choice of denominator made the greatest practical difference for assembly and portfolio management. Those tasks accounted for 30.0% of episodes when each episode counted once and 34.9% when the shares for each student received equal weight. Every participating student contributed some work in this group. A student with only a few documented tasks can therefore contribute a comparatively large personal share to assembly. Equal weighting gives that student’s experience as much influence as the experience of a student whose work was discussed in greater detail. Both percentages describe these students. They answer different questions about where attention went.
For verification, review and coordination, the change went in the other direction. The share fell from 30.0% to 25.3% under equal student weighting. It follows that a counselor describing their total queue of pending tasks could arrive at a different picture from a counselor describing the average distribution across the students they serve. A queue can contain several successive requests from one student. A family making a plan for a single applicant has more reason to look at the second calculation. Neither calculation measures how much support a family needs.
The narrative-development share moved by just two tenths of a percentage point, from 40.0% to 39.8%. Across the ten leave-one-out calculations, its lowest value was 35.3%. Even that value exceeded the highest value for assembly or for verification, review and coordination in those calculations. Within this sample, the ordering of the largest group therefore survives the removal of any one student. The test is possible because the underlying calculations retain which episodes belong together. Treating all 40 observations as independent would throw away that information.
This result has a narrower interpretation than it might appear to have. The participating students share an advising setting in which writing is discussed explicitly and often. A task that attracts a request for feedback leaves a visible trace. Other tasks, including decisions made at home and routine entries completed without discussion, may be less visible. Removing one student in turn leaves the shared setting intact. It cannot tell us whether whatever brought a task into view differed across stages. The stage shares therefore describe selected documented work, even where the ordering is stable.
There are also decisions about the size of a task. Several supplemental responses discussed as a set may enter the count once. A personal statement can generate a development episode followed by a separate review episode. A single observation can record progress on more than one clearly distinguishable deliverable. The family deciding how much writing remains might count individual responses. The adviser deciding which feedback to return next might count review requests. This analysis counts separately identifiable transitions in the advising process. Its denominator should stay attached whenever the result is quoted.
For example, a journalist can accurately report that 16 of the 40 selected task episodes involved developing a narrative and 12 involved assembling applications or managing the list. Saying that four in ten applicants experienced a writing delay would change both the unit and the event being measured. Saying that applicants spent three tenths of their time on assembly would require a time-use study. The distinction matters particularly when an observation interval spans several weeks and other tasks could have advanced during the same period.
The same issue affects comparisons between years. The autumn observations stopped at different points in the two years. An applicant seen later in the season has had more opportunities to report a handoff or a submission. A comparison of the fraction with visible milestones by year would mix documentation and opportunity with any real difference in workflow. Pooling the two observation periods provides a broader range of tasks for the descriptive inventory. The public estimates stay at that pooled level.
5. a process map for application readiness
The following map translates the findings into observable gates. A task moves forward only when the evidence for the current gate exists.
Gate 1: Scope confirmed
- Live prompt copied from the current application or official source.
- Required, optional, or conditional status identified.
- Word or character limit confirmed.
- Intended deadline and internal target recorded.
- Owner of each dependency identified.
Gate 2: Content direction confirmed
- Purpose of the response stated in one sentence.
- Candidate story or evidence selected.
- Relationship to the prompt is explicit.
- Overlap with other application components reviewed.
Gate 3: Full draft exists
- A complete beginning, middle, and ending are present where appropriate.
- Placeholders and research notes are visible rather than hidden.
- The draft is ready for substantive, not merely line-level, feedback.
Gate 4: Tailoring complete
- The draft answers the exact live prompt.
- Institution and program references are verified.
- The response is within the platform’s actual limit.
- Reused content has been checked for stale names and mismatched claims.
Gate 5: Review complete
- Major comments have been accepted, rejected, or resolved deliberately.
- Student voice and factual accuracy have been preserved.
- The current version is clearly identified.
- Any outside dependency has an owner, deadline, and follow-up plan.
Gate 6: Assembly complete
- Final text is entered in the application.
- Activities, honors, courses, testing, and other relevant sections are complete.
- Formatting survived copy and paste.
- A full preview has been reviewed as the college will receive it.
Gate 7: Submission confirmed
- Required agreement and payment steps are complete.
- Submission confirmation is recorded.
- Portal follow-up is scheduled.
- Separate recommendations, transcripts, portfolios, or test reports are monitored according to the institution’s instructions.
The gates reduce ambiguity because each is tied to evidence. “Essay almost done” becomes “full draft exists but prompt verification and word-limit work remain.” “Application complete” becomes “full preview reviewed, but submission not yet confirmed.” That level of specificity is more useful for planning and less likely to assign blame.
6. backward-planning framework
The study does not identify a universal number of weeks for every application. The reliable elapsed-time subset is too small and heterogeneous for that claim. The following framework is therefore a planning tool, not an empirical completion-time estimate.
Eight or more weeks before the deadline
- Confirm the list and the current official requirements.
- Identify conditional prompts, honors requirements, portfolios, and separate programs.
- Inventory reusable material without assuming prompts are interchangeable.
- Start high-uncertainty narrative tasks: personal statement, identity questions, difficult academic-interest essays, and any response requiring substantial reflection.
- Request recommendations according to school policy and institution requirements.
Six weeks before
- Select working topics for priority applications.
- Produce complete drafts for the hardest responses.
- Record unresolved research questions.
- Confirm whether outside materials can arrive after the student submission.
- Review whether the application list is realistic given the remaining workload.
Four weeks before
- Move priority drafts through substantive review.
- Verify prompt alignment and school-specific facts.
- Begin entering stable application sections.
- Resolve activity descriptions and ordering.
- Schedule a complete-preview review date rather than leaving it implicit.
Two weeks before
- Bring all required responses within the actual limits.
- Enter final text and test formatting.
- Review recommendations and other dependencies without assuming they block submission.
- Generate or inspect the full application preview.
- Compare the preview against a requirement checklist.
One week before
- Reserve time for factual corrections and platform issues.
- Stop adding optional material merely because space remains.
- Recheck names, programs, dates, and reused language.
- Complete payment or waiver steps.
- Submit before the deadline when feasible and retain confirmation.
After submission
- Create or monitor the applicant portal.
- Check that institution-specific materials are received.
- Track interviews, financial-aid forms, and later updates separately.
- Preserve the exact submitted version for future reference.
The buffer matters because the median observed transition in the strict subset was 12 calendar days, and half of those observations lay between 7 and 35 days. Those figures cannot predict an individual task, but they argue against a plan that assumes every meaningful transition can occur overnight.
7. practical diagnostic: what kind of “stuck” is this?
Families often respond to every delay with the same instruction: “work on the application.” The findings suggest a more precise diagnostic.
If the student cannot choose a topic, the problem is narrative uncertainty. The next action is not line editing. It is structured comparison of candidate stories against the prompt, application role, and evidence available.
If a complete draft exists but feels generic, the problem may be tailoring. The next action is to identify which claims could be sent to any school and replace only those that need a verified connection. More research is not automatically better; relevance matters.
If a draft is strong but too long, the problem is prioritization under constraint. The next action is to identify the response’s central claim, protect the evidence necessary to support it, and remove material that performs a duplicate function.
If feedback exists but revision has not moved, the problem may be decision load. The next action is to sort comments into required factual changes, structural choices, optional style preferences, and comments the student rejects to preserve voice.
If content is ready but the application is not, the problem is assembly. The next action is to enter stable material, produce a preview, and reconcile missing sections rather than continue polishing one essay indefinitely.
If another person’s material is outstanding, the problem is coordination. The next action is to verify the actual rule, identify who owns the follow-up, and determine whether student-controlled work can proceed in parallel.
If several applications are nearly complete but none is ready, the problem is portfolio congestion. The next action is to prioritize by deadline, importance, and remaining uncertainty, then finish one submission-ready package before spreading effort across every open school.
8. role-specific takeaways
8.1 For students
Use verbs and evidence in progress updates. “I worked on it” is less actionable than “I selected the topic, wrote 420 of 650 words, and marked two factual questions.” The goal is not surveillance; it is reducing the time needed to determine the next step.
Separate content work from compliance work. A creative drafting session and a word-limit session require different thinking. Trying to invent, research, cut, fact-check, and proofread simultaneously can make progress harder to see.
Keep a current version. When several copies circulate, review can become a new bottleneck. Use one clearly designated working version and preserve older versions without allowing them to compete for “final” status.
Do not confuse feedback volume with required change. Some comments identify risks; others offer alternatives. The final application should remain factually accurate and recognizably yours.
Enter stable sections early. Platform entry can reveal unexpected questions, formatting problems, or missing details. Discovering those items before the final night creates options.
8.2 For families
Ask stage-specific questions. “Are you done?” invites a yes/no answer that may conceal several remaining gates. Ask which gate is complete and what evidence will close the next one.
Help with logistics without taking over authorship. Families can support schedules, requirement tracking, records, payment steps, and reminders while keeping the student responsible for personal representation and final approval.
Treat a disappearing task cautiously. If an item is no longer discussed, confirm whether it was completed, dropped, or deferred. This study treated missing follow-up as censoring rather than assuming an outcome.
Build slack around school and family obligations. Calendar time includes everything else in a senior’s life. A plan that uses every available day as if it were an uninterrupted application day is fragile.
8.3 For counselors and advisers
Define the endpoint before assigning the task. “Work on the essay” can mean brainstorm, outline, draft, revise, or cut. A defined deliverable makes the next observation interpretable.
Separate advisory deadlines from official deadlines. The student should know which date is an internal buffer and which is the institution’s actual requirement.
Document uncertainty. If a prompt, policy, or submission rule needs verification, mark it as unresolved rather than presenting a guess as a requirement.
Close the loop. When a student sends a revision, record whether the milestone is receipt, review, approval, application entry, or submission. A handoff is progress, but it is not automatically completion.
Avoid assigning blame from timestamps. A long interval can reflect reprioritization, school workload, other applications, waiting, or missing observations. Discuss the next controllable step instead.
8.4 Applying the findings to a week with competing deadlines
The following exercises use fictional situations constructed for planning. They contain no client examples. A family can use them with a school counselor, a private adviser, or independently. The sequence begins with a decision the student needs to make. It ends with something another person could inspect and understand.
Exercise one. Decide which question needs answering first
Imagine a student who has an evening available and three requests for attention. One response needs a different ending. A college asks whether the student wishes to be considered for a particular program. A teacher has agreed to write a letter. Each item sounds urgent when described in isolation. The program decision may determine whether a further response is required. The teacher controls the letter. The student controls the ending. Before allocating the evening, the student opens the current application, checks what changes when the intended program is selected, and identifies who needs to confirm the recommendation arrangements.
What would count as progress after thirty minutes? The student can state the requirements for their chosen program, identify any still-open question about the letter, and choose the writing task with the most immediate dependency. It is possible to leave the essay untouched for those thirty minutes and make real progress toward a submission. The time spent clarifying the scope prevents polishing an answer to a question the student will never submit.
An adviser using this exercise can ask the student to name the observation that would change their plan. If choosing the program adds a required answer, the student needs to reserve a drafting period. If the college confirms that the program uses a separate, later application, the same period can go back to the main response. Stating the decision this way makes room for uncertainty without turning it into an indefinite hold.
Exercise two. Choose a handoff that gives the reader a real decision
A student sends a long response and asks a reader to look it over. The reader might comment on punctuation, suggest another topic, cut a paragraph, or explain that the response never addresses the second half of the prompt. All four replies could be thoughtful. Their usefulness depends on the timing. If the student is deciding whether the topic fits, line edits can consume attention before the larger decision has been made. If the student has twenty minutes left before their own planned submission, a suggestion to rethink the entire topic requires a clear explanation of what is at stake.
For a first substantive review, give the reader the live question and ask them to identify the sentence or passage that answers it. For a later review, give them the word-compliant version and ask whether a cut removed information needed to understand the sequence. Ask the reader to flag any statement whose accuracy depends on knowledge they do not have. The student can verify that statement separately. Returning a version with two questions answered provides a clearer milestone than returning one with twenty comments whose purpose is unclear.
Students also need an explicit way to disagree. Suppose a reader replaces a modest description with a stronger sounding claim. The student knows the second description extends beyond what they actually did. Recording that the original wording stays resolves the comment. Asking a family to clear every suggestion by accepting it creates a risk to authorship and factual accuracy. Asking the student to decide what to do with each substantive question gives the review a definite endpoint.
Exercise three. Test whether reused material still does its job
Consider two hypothetical prompts. One asks about a community that shaped the applicant. The other asks about a contribution the applicant expects to make. The same experience could furnish evidence for both responses. The first reader needs to learn what the community did and how the student changed. The second reader needs to understand what the student learned to do and how they might apply that experience. Swapping the institution’s name leaves that difference unresolved.
Ask the student to summarize the job of the intended response in one sentence before pasting in the earlier draft. Next, underline the evidence from the earlier response that helps answer this question. A scene can survive intact while its introduction or ending needs substantial work. Conversely, the ending may fit perfectly while most of the scene is unnecessary. This exercise makes reuse a decision about evidence and emphasis. It also gives the student a way to abandon a reuse attempt early when adapting it would take longer than writing a fresh response.
The practical stopping point is a response the student can defend against the exact current question. A reader should be able to tell which aspect of the experience answers the prompt and why it deserves space. If the student cannot yet say what they would remove first when the response is over the limit, the priority among its ideas may still need attention.
Exercise four. Give an outside dependency a specific owner
A family sees an outstanding school form and assumes everyone must wait. Ask who can take the next action. The student may be able to assign a recommender to the correct college. The high school may control transmission. The college may control when an item appears as received. Those actions involve different people and different evidence. A reminder sent by the student confirms only that the reminder was sent.
Start with the institution’s instructions and the school counselor’s process. Record the date by which the student must act and any separate materials deadline. Identify the least intrusive follow-up that will answer the open question. A single clear inquiry from the designated person avoids putting a teacher in the position of reconciling separate reminders from several family members. The family can decide when it will next check the application status and what discrepancy would warrant further contact.
If the family still has no direct confirmation, its own progress description can say exactly that. The student may nevertheless be able to complete work on their side. Common App’s applicant guidance allows separate submission of school forms, subject to the college’s requirements. In this exercise the decision turns on the specific college’s instructions. The pending item by itself cannot reveal whether it is holding up the student’s submission.
Exercise five. Examine the version the recipient will see
An essay that looks right in the student’s writing environment can look different in an application preview. The student and a reader can divide the review into two passes. In the first, check visible facts against the student’s own information. Do the activity descriptions agree with their titles? Are grade levels consistent across sections? Does the chosen academic interest match the college-specific response? The goal is to find discrepancies the student can resolve. If a detail is uncertain, identify the source that can settle it and the person who will consult it.
In the second pass, read as someone encountering the assembled application for the first time. Has a paragraph break disappeared? Is a placeholder still visible? Does an acronym appear before it is explained? Is the same paragraph printed twice across sections because it was transferred to a new field and never removed from the old one? Correct the original field and regenerate the preview. An annotation on a preview does not change the submitted answer.
There is also a judgment about when to stop editing. A correction to a date has a clear purpose. A suggestion to replace a familiar word with a more elaborate one may have little value in a final review, especially when the student prefers the original. Agreeing on the scope of the preview review helps avoid reopening settled decisions at the point where the family most needs a dependable version.
Exercise six. Distinguish a change of plan from a missed target
A student can decide after reading a course catalog or discussing cost that a college should leave the list. In that situation an unfinished application may reflect a deliberate decision. A process that automatically treats every unsubmitted application as a shortfall would reward sending an application the family no longer wants. A more informative question is whether the decision was recorded in time for the student to use the freed capacity elsewhere.
Consider two endings to the same fictional planning discussion. In one, the student decides to skip an optional response because the rest of the application already conveys the relevant information. In the other, the student continues to intend to submit the response but cannot choose an example. The empty field looks identical. The next steps differ. One student can move to a final review. The other could benefit from a focused discussion about content. A reviewer needs the student’s decision to tell these situations apart.
The same distinction applies retrospectively to this study. When a task disappears from later conversations, there is no basis for attributing a deliberate change of plan. There is also no basis for attributing missed follow-through. A prospective study could invite students to record a final status in their own words and allow them to decline to give a reason. That would improve interpretation while preserving control over private family decisions.
8.5 How to use a backward plan when the calendar changes
Work backward from the earliest verified deadline that actually applies to the student. Ask what must be true at the moment of submission. The required fields must be complete, the text must be the approved version, and the person submitting must be able to carry out any required final steps. Now move back one decision. What must be available for a complete preview? Stable content, the correct academic choices, and resolved questions about required sections are plausible prerequisites. Some preparations can run alongside these decisions.
Next, choose a time when the people needed for the preview can realistically participate. Family travel, a school production, an examination, or a counselor’s availability can eliminate an apparently free evening. The illustrative week counts in this report can be shifted to reflect those constraints. A family starting close to a deadline should first identify required work, verify what has already been done, and negotiate any review it can actually receive. Treating an eight-week illustration as a requirement would misapply the evidence.
When a new task appears, ask which existing commitment it displaces. Suppose a college-specific response becomes necessary after the student selects a different program. The family can postpone an optional revision, move a later-deadline application down the queue, or reconsider the academic choice on its merits. It cannot add unlimited work to the same evening. Naming the displaced task makes the cost visible without assigning fault to the student who discovered the requirement.
A brief weekly review can consist of three decisions. Which submission has the least predictable remaining requirement? Which upcoming handoff needs another person? Which item can be marked complete with evidence already in hand? A family can answer those questions aloud, on paper, or with a school counselor. The value comes from identifying real decisions while they are still movable.
8.6 Questions for a future prospective study
A stronger follow-up would recruit consecutive eligible seniors across a stated period and invite participation before work begins. Each student could select a small, comparable set of tasks to follow, such as a first complete personal statement, one designated supplemental response, and a first full application preview. Defining those endpoints in advance would allow comparisons within a type of task. Participation and the amount of information collected would need appropriate authorization.
The study could ask for the date the student first intended to begin, the first actual work date if known, the date they requested a review, the date feedback arrived, and the date they considered the deliverable ready. A separate field could record the day the research team learned about each event. Those last two dates can differ. Even a conscientious student may only tell a counselor about an event at the next meeting. Collecting a confidence indicator would allow the analysis to compare exact dates with approximate ones.
Some tasks would still have uncertain endings. A student might change the college list, finish between check-ins, or prefer to stop reporting. The study could count the reasons students actually volunteer and explicitly describe the rest as unknown. Asking every family for more detail might improve measurement while also making participation harder. A sensible design would pilot the minimum number of questions needed, assess how often people answer them, and explain how incomplete observations affect any summary.
Finally, a prospective design should decide whose question it aims to answer. A student planning an evening needs to know which next step they control. A school counselor planning staffing needs to know when requests arrive and how many require a reply. An evaluator testing an intervention needs a comparison group and a prespecified outcome. The retrospective figures in this report answer a narrower question about where pending work appeared in this sample. Each broader question requires its own denominator and an appropriate design.
9. comparison with external research
The internal episode distribution should not be compared directly with national applicant percentages. The denominators differ. This study counted stage-specific tasks among 10 selected advising clients. Common App counts applicants and applications. RNL surveyed students about perceptions and unfinished applications. Delaney and Odle studied state policy and enrollment. Buehler and colleagues studied predicted task-completion times in experimental and naturalistic settings.
The most defensible comparison is conceptual.
Common App’s official guide supports the pipeline model. It separates gathering information, engaging supporters, understanding varying requirements, planning essays, and submission. That structure resembles the stages described here, but it does not verify their internal frequencies.
Common App’s national volume reports show why portfolio management matters. Applicants submitted more than 10.7 million applications in the 2025–2026 season, averaging 7.06 institutions per applicant. That scale does not prove that Ivy Institute students experienced more friction, but it establishes that multi-application management is a normal feature of the broader system.
Delaney and Odle’s study supports caution about simple solutions. A common platform may reduce duplicated form work, yet their policy analysis did not find enrollment effects. The lesson is not that common applications are ineffective. It is that reducing one administrative barrier does not automatically resolve information, preparation, cost, or decision barriers.
The 2024 RNL report’s survey results show that unfinished applications can relate to cost as well as length and difficulty. The Ivy Institute study did not code financial reasons unless they were directly part of the task, and it did not attempt a national taxonomy of abandonment. The external survey therefore broadens the context but cannot explain the internal episode distribution.
Buehler, Griffin, and Ross provide a reason to use prior observed timelines when planning. Their research does not permit this report to label any student a procrastinator or to infer why an episode remained open. It supports a planning practice: compare the new task with the actual distribution of similar past transitions and add a buffer.
One planned external citation was deliberately excluded. The 2002 article “Procrastination, Deadlines, and Performance” was retracted by Psychological Science on September 2, 2026 after the journal stated that it could no longer attest to the reliability of the findings. This report does not rely on that article. The exclusion illustrates why source verification matters, especially for a study published in a rapidly changing evidence environment.
10. limitations
Small, selected sample. Ten seniors cannot represent the national applicant population, all Ivy Institute clients, or all advising models. Students were included because the evidence was sufficiently detailed for ordered coding, which may select for more documented or more complex cases.
Internal, commercially interested research. The Ivy Institute conducted and is publishing the study. It has an interest in the interpretation. There was no independent coding team, preregistration, or external audit.
Documentation created for practice. The underlying information was produced to support advising, not a research protocol. The frequency and detail of observations vary. A task that generated more discussion may be easier to detect than a task completed quietly.
Episode counts are not hours. The study measured stage-specific episodes, not effort, cognitive load, meeting time, or total work. A five-minute verification issue and a multi-week drafting process each count as one episode if they meet the definition.
Unequal contribution. Students could contribute different numbers of episodes. The aggregate episode distribution therefore gives more weight to cases with more observable tasks. The study avoids publishing student-level comparisons because the sample is small and the documentation intensity differs.
Heterogeneous milestones. The nine elapsed-time episodes did not share one endpoint. Median elapsed time is a description of the observed subset, not an estimate of the time required for a college essay or application.
Missing endpoints. The analysis could verify a dated transition for nine of 40 tasks. The other tasks remain in the denominator for the analysis of stages.
Unobserved work. Students may have worked outside the documented periods. The absence of an observation is not the absence of activity.
No causal inference. The study cannot determine whether advising sped, slowed, or caused any transition. It cannot determine whether a bottleneck affected admission.
No admissions-outcome analysis. Admission decisions were outside the research question and were not used to judge whether a workflow was successful.
Potential coding judgment. Categorizing a real task into one primary stage requires judgment. Consolidating granular codes improves privacy and readability but can hide nuance.
External sources are contextual. Common App reports, survey research, policy studies, and task-timing research do not validate the internal calculations. Differences can reflect population, method, year, or unit of analysis.
11. conclusion
In this internal case series, the most common documented application bottleneck family was narrative development, with 16 of 40 episodes. Yet the majority of episodes occurred outside that stage. Assembly and portfolio management, review and coordination, and tailoring and verification together accounted for 24 episodes. The practical implication is not that students should write faster. It is that application plans should recognize the entire production pipeline.
Nine tasks had documented transitions. The median interval between observations for those nine was 12 calendar days. The observations establish how much time separated two descriptions. A milestone may have occurred before the meeting at which it was first reported.
The strongest operational response is to replace broad progress labels with observable gates. Confirm the live requirement. Select the content direction. Produce a full draft. Verify prompt fit, facts, and limits. Resolve review and dependencies. Assemble the complete application. Review the full preview. Confirm submission. Each gate has a different failure mode and a different next action.
College applications do not get stuck only because students have not written enough. They also get stuck because content has not yet been tailored, feedback has not been resolved, dependencies have not been managed, or the finished pieces have not been assembled into a verified submission. A realistic plan gives each of those transitions its own deadline, owner, and evidence of completion.
12. references
Buehler, R., Griffin, D., & Ross, M. (1994). Exploring the “planning fallacy”: Why people underestimate their task completion times. Journal of Personality and Social Psychology, 67(3), 366–381. https://bear.warrington.ufl.edu/brenner/mar7588/Papers/buehler-et-al-1994.pdf
Common Application. (2026). Application guide for first-year students. https://www.commonapp.org/apply/first-year-students/
Common Application. (2026, January 15). Deadline update, 2025–2026: First-year application trends through January 1. https://www.commonapp.org/reports/january-2026-deadline-update/
Common Application. (2026, August 20). End-of-season report, 2025–2026: First-year application trends. https://www.commonapp.org/reports/end-season-report-2025-2026-first-year-application-trends/
Common Application Applicant Support. (n.d.). Can I submit my application before my recommenders submit their forms? https://appsupport.commonapp.org/applicantsupport/articles/Knowledge/Can-I-submit-my-application-before-my-recommenders-submit-their-forms
Delaney, J. A., & Odle, T. K. (2025). State-level common application policies and college enrollment. Education Finance and Policy, 20(1), 164–171. https://doi.org/10.1162/edfp_a_00439
RNL, Ardeo Education Solutions, & Halda. (2024). 2024 high school students’ perceptions of college financing. https://files.eric.ed.gov/fulltext/ED675523.pdf
SAGE Publications. (2026, September 2). Retraction: Procrastination, deadlines, and performance: Self-control by precommitment. Psychological Science, 37(9), 665. https://doi.org/10.1177/09567976261488042
The Ivy Institute, internal materials, data, and information.
Appendix a. public data tables
Stage-family chart data
| Stage family | Episodes | Percent |
|---|---|---|
| Narrative development | 16 | 40.0 |
| Assembly and portfolio management | 12 | 30.0 |
| Verification, review and coordination | 12 | 30.0% |
Outcome chart data
| Status | Episodes | Percent |
|---|---|---|
| Observed milestone | 9 | 22.5 |
| Without a verified milestone | 31 | 77.5% |
Elapsed-day chart data
Nine paired observations. Median 12 calendar days. Middle half 7 to 35 days.
Appendix b. data dictionary
Student: one senior represented in the authorized internal sample. The public report does not identify students.
Task episode: a bounded, stage-specific unit with a dated starting state and a later observation, or a task first appearing at the final observation and retained as right-censored.
Narrative development: topic selection, brainstorming, outlining, drafting, or development of unfinished responses.
Tailoring and verification: prompt alignment, word-limit compliance, or school-specific research.
Review and coordination: feedback, revision, or management of an outside dependency.
Assembly and portfolio management: completion and entry of application sections, full-preview review, submission preparation, or management of several applications.
Observed milestone: a defined transition visible in the evidence. It is not necessarily final submission.
Right-censored: the next milestone was not observed before the retained record ended.
Intermediate or ambiguous: partial completion, rework, or a status that could not be coded more precisely.
Elapsed calendar days: the difference between two dated states. It is not active work time.
Observation window: the dated interval during which an open or ambiguous episode was visible. It is not completion time.
