Executive summary
To evaluate an admissions consultant's acceptance rate, ask what is being counted, who is included, which period and colleges are covered, and what comparison supports the claim. An offer count, a percentage of applications admitted, and a percentage of students receiving at least one offer describe different outcomes. None should be silently substituted for another.
A reported result can be accurate and still tell a family little about the additional benefit of hiring the firm. Clients may differ from the general applicant population before counseling starts. They may apply to different colleges, use different admissions rounds, or receive different levels of service. A college's overall rate is therefore a benchmark, not automatically the chance those clients would have faced without help.
This report develops a practical verification method using institutional reporting definitions, a small set of original advising studies, and transparent hypothetical examples. It also examines what selected Ivy Institute operational records can support. The aim is to help families distinguish a documented result, a useful comparison, and evidence that a service caused an improvement. Those are progressively stronger claims.
Methods, scope, and the publisher's interest
This is a company-published evidence review and original measurement guide. Sources were checked on November 10, 2026. The public evidence includes the Common Data Set initiative, an institutional submission, federal guidance about testimonials, and three academic evaluations of advising programs. These sources were chosen to clarify specific measurement problems; the review is not a census of all research or all consulting firms.
All numerical firm examples below are invented unless an institution or study is explicitly named. They demonstrate how a statistic behaves when its definition changes. They do not describe any actual company's clients. No new experiment, matched-client study, or independent audit of commercial admissions outcomes was conducted.
The Ivy Institute sells admissions consulting and publishes this report. That commercial relationship is relevant to how readers assess the discussion. The same verification questions should be applied to its own claims. Selected internal records are discussed only in deidentified form, with their limitations. This article has not undergone independent academic peer review.
The related public comparison of consulting outcome claims examines particular company disclosures. The present report supplies a reusable method for reading a claim from any provider, including one not covered in that earlier review.
Begin by defining the outcome
The Common Data Set develops shared definitions and cohorts for institutional reporting. It is a standards initiative rather than a single comprehensive outcomes database. [E31] The underlying lesson is useful for families: a percentage becomes interpretable only after the event and the population are defined.
| Measure | Numerator | Denominator | Question it answers |
|---|---|---|---|
| Offer count | Admission offers | None | How many positive decisions were reported? |
| Application acceptance rate | Applications receiving offers | Eligible applications submitted | What share of these applications succeeded? |
| Student portfolio success | Students with at least one qualifying offer | Eligible students applying within the defined group | How many people obtained at least one such option? |
| First-choice admission | Students admitted to a prospectively identified first choice | Eligible students applying to their identified first choice | How often did a particular preference result occur? |
| Enrollment | Students who actually enrolled | The explicitly stated eligible group | Which students attended, rather than merely received offers? |
Each can be useful. A family wanting several viable options may care most about portfolio success. A student interested in one specific program may need a school- and program-specific rate. A family focused on affordability may care whether an admitted student could actually attend. A strong result on one measure cannot answer all three questions.
“Top college” also needs a definition. It might mean a fixed named set, a ranking category, or a student's personal preferences. Ask which list was used and when it was fixed. A list chosen after the results are known can change the apparent success rate without changing a single admission decision.
One hypothetical cohort can produce several truthful percentages
Imagine 40 students who each submit eight applications within a defined group of colleges. The resulting 320 applications produce 64 offers. Thirty-two students receive at least one offer. Twenty-four receive an offer from a college they had previously identified among their top three choices. Ten are admitted to their stated first choice.
| Statement | Calculation | Result |
|---|---|---|
| Total offers | Count all positive decisions | 64 offers |
| Application acceptance rate | 64 offers ÷ 320 applications | 20% |
| At least one qualifying offer | 32 students ÷ 40 students | 80% |
| At least one top-three offer | 24 students ÷ 40 students | 60% |
| First-choice offer | 10 students ÷ 40 students | 25% |
There is no contradiction between 20% and 80%. The first follows applications; the second follows people across several attempts. A headline calling both an “acceptance rate” would hide the distinction. The same 64 offers also do not imply that 64 different students were admitted.
Application breadth matters. A portfolio result across eight colleges is not the probability of admission to each one. Nor can a family calculate its own portfolio probability by treating colleges as identical, independent coin flips. The applications share a student's record, constraints, interests, and choices. A reported portfolio rate should therefore identify how many colleges students generally applied to, as well as which colleges counted.
This example shows why the first request should be for the original counts. A provider need not disclose student identities to explain that a statistic means 32 of 40 students in a particular cohort. Counts make the definition concrete and allow straightforward arithmetic checks.
Check the institutional benchmark independently
MIT's 2025–26 Common Data Set reports 29,281 first-year applicants, 1,334 admitted applicants, and 1,152 enrolled students for fall 2025. Dividing admitted applicants by applicants gives approximately 4.56%. Its instructions include subsequently admitted waitlisted students. [E32] This is a dated institutional measure with a defined population.
If a consultant compares a client percentage with that benchmark, confirm that both refer to the same college, entry year, applicant category, and relevant decision scope. A transfer percentage does not belong beside a first-year benchmark. A rate for one undergraduate division should not be compared with a broader institution unless the populations align. A result reported before waitlist decisions may differ from a final-cycle result.
A comparison can be imperfect and still informative when its limits are visible. The problem is treating a mismatch as if it disappeared because both numbers contain a percent sign. Ask the provider to identify the institutional source and label any unavoidable difference.
Most importantly, the overall institutional denominator does not identify which applicants used private guidance. It cannot automatically serve as a group of otherwise comparable nonclients. Even a perfectly copied institutional figure does not solve that problem. Verifying the benchmark's arithmetic and verifying the consulting effect are separate tasks.
Why higher client rates do not automatically measure improvement
Consider an invented comparison in which a firm's clients have a 20% admission rate and the college's overall rate is 4%. Dividing 20 by 4 gives a fivefold rate ratio. The arithmetic is correct. Calling it a fivefold increase caused by the consultant requires additional evidence about how those clients would have performed without the service.
That missing outcome is the central research problem. A student cannot submit the same application in the same circumstances both with and without counseling and experience both admissions decisions. Researchers therefore need a credible comparison design, not just a large difference between two published percentages.
One source of difference is selection into the service. Families who hire a particular firm may bring different academic records, school resources, financial circumstances, or application goals. A firm may accept only some prospective clients. Another may specialize in unusual or difficult cases. These possibilities can move observed rates in either direction; a higher rate is not automatically evidence of greater skill, and a lower rate is not automatically evidence of worse advice.
| Starting group | Firm A clients and offers | Firm B clients and offers | Within-group rate at both firms |
|---|---|---|---|
| Group H, defined before advising | 80 clients; 48 offers | 20 clients; 12 offers | 60% |
| Group L, defined before advising | 20 clients; 4 offers | 80 clients; 16 offers | 20% |
| Total | 100 clients; 52 offers | 100 clients; 28 offers | A:52%; B:28% |
The example gives both firms identical rates within each starting group. Their overall percentages differ because their client mixes differ. The letters are abstract categories, not labels for actual students, and the table does not claim that a few background variables fully describe admissions readiness.
Adjusting for measured differences can improve a comparison, but the adjustment must be explained. Matching students on a test score alone does not establish equivalence in courses, recommendations, family circumstances, or intended programs. A convincing report states which starting characteristics were measured, when they were measured, and what important differences may remain.
What original advising studies add to the discussion
Academic research demonstrates why study design matters and why outcomes should remain specific. Barr and Castleman's randomized Bottom Line evaluation found that intensive high-school and college advising increased bachelor's degree attainment among lower-income students. [E27]
Phillips and Reber's randomized virtual-advising study found broader four-year application behavior and greater perceived support, but no increase in admission or enrollment. [E28] Gurantz and colleagues found that a virtual-advising offer increased attendance at high-graduation-rate colleges by 2.7 percentage points, without raising overall enrollment. [E29]
These findings concern particular interventions, populations, and endpoints. They should not be converted into a commercial firm's Ivy admission multiplier. The studies also prevent a sweeping claim that advising never helps. The useful question is what kind of support, for which students, changes which outcome under a credible research design.
For a commercial provider, a stronger causal claim would require evidence appropriate to that service and client population. A randomized evaluation is one possible design; carefully designed observational comparisons may also contribute. Any such study still needs transparent measurement and limitations. This review does not establish that no relevant study could exist elsewhere; it identifies what the sources examined can support.
Ask who entered, who disappeared, and when the count stopped
A cohort should be defined before results are calculated. Does it include every student who purchased the relevant service, only those who completed the program, or only those who reported decisions? Each denominator describes a different population. A firm can report more than one, but should identify them.
Imagine 50 eligible students, 35 known successes, five known students without a qualifying offer, and ten with unknown results. The confirmed success share among all 50 is 70%. Among the 40 with known outcomes, it is 87.5%. If every unknown student succeeded, the overall rate would reach 90%. The available records do not determine which outcome applies to the unknown group.
Publishing only 87.5% would answer a narrower question than many families might assume. The remedy is not automatically to count all missing students as failures. It is to disclose the unknowns and show what can and cannot be calculated. This makes a limitation visible without inventing outcomes.
Ask about students who stop working with the firm, change their lists, apply late, decline recommended advice, or purchase only one review. Their treatment in the statistics should follow a stated rule. Rules written after difficult results arrive can selectively improve the number. A prospective family benefits from knowing the rule before joining the cohort.
The final reporting date matters too. An early decision deferral is not an admission, and an unresolved waitlist is not a final rejection. Record the status at the stated cutoff and update it consistently. Historical figures should remain identifiable rather than being silently replaced by a new mixture of cycles.
Small cohorts, pooled years, and impressive multipliers
With ten applications, one additional offer changes the rate by ten percentage points. With 100 applications, one offer changes it by one percentage point. The same percentage can therefore rest on very different amounts of information. Request counts before treating a small numerical difference between firms as meaningful.
Pooling several years increases the count but also combines different contexts. It can be a reasonable description of a provider's history, provided the cycles, services, and inclusion rules are stated. It is less useful as a description of the exact team and process a new client will receive if those have changed substantially.
Ask how a multi-school average was calculated. A firm might give every college equal weight, combine all applications, or average school-level multipliers. These operations answer different questions. In a hypothetical two-school example, rates of 50% on two applications and 10% on 100 applications produce an unweighted school average of 30%, but an overall application rate of 11 out of 102, about 10.8%.
Neither calculation is inherently dishonest when labeled correctly. The equal-school average describes the typical entry in a two-school table; the pooled rate describes the applications actually submitted. A family should not compare one firm's equal-school average with another firm's pooled rate as though they were the same measure.
Finally, an advertised maximum needs its own label. “Up to” can refer to one school, subgroup, or historical year. It does not establish the typical result across clients. The useful follow-up asks which observation produced that maximum and what the rest of the distribution looked like.
A proportionate verification checklist
| Question | Useful supporting material |
|---|---|
| What event counts as success? | A precise definition of offer, portfolio success, enrollment, or another endpoint. |
| Who is included? | Service type, entry rules, exclusions, and the count of eligible clients or applications. |
| What period is covered? | Entry cycle, admissions rounds, and the final reporting date. |
| What are the original counts? | Numerators, denominators, and missing outcomes. |
| What colleges or programs count? | A named list and when that list was fixed. |
| How was the result confirmed? | A stated verification process and any independent review scope. |
| What supports the comparison? | Direct institutional sources or a clearly defined comparison cohort. |
| What does the firm claim the number means? | A distinction between historical description, association, and causal improvement. |
Families do not need other students' unredacted applications, financial records, or decision letters to ask these questions. An anonymized aggregate table and a clear methods note can provide meaningful transparency. If an outside reviewer checked documents, ask what was checked: the offer count, inclusion rules, calculations, missing cases, or the claim of additional benefit. The word “verified” is incomplete without its object.
A document check can establish that selected offers occurred. It cannot, by itself, show that all eligible clients were included or that counseling caused those offers. Similarly, a testimonial can be a truthful account of one family's experience while providing no denominator for a success rate.
FTC guidance explains that exceptional-result testimonials should not misleadingly imply generally expected results, and material relationships can require disclosure. Its guidance is context dependent. [E33] This report does not make a legal finding about any particular provider; it applies the practical lesson that a striking story needs context.
What The Ivy Institute's selected internal records can support
A selected November 2026 planning entry recommends continuing other applications while awaiting an early result. It describes counselor optimism as qualitative rather than a statistical prediction. That record supports a description of planning under uncertainty. It does not supply an estimated probability of acceptance or evidence that consulting changed the outcome. [IN06]
A selected activity-review email asks a student to check numerical claims in proposed activity descriptions. It documents a request for verification, not independent confirmation of the final application. [IN10] The distinction is the same one families should apply to company outcome reporting: a plan to check, a completed check, and a verified result are different evidentiary stages.
Readers can examine the company's published case studies as individual narratives. They should request a separate cohort definition before treating a collection of stories as a rate. The team descriptions and service descriptions help identify the offering under consideration; they do not substitute for an outcome audit.
The selected internal material reviewed for this article contains no matched nonclient comparison and was not assembled as a complete outcome ledger. It should not be used to validate a company-wide acceptance multiplier. A future company report could improve accountability by publishing a dated aggregate table using the fields above, while protecting individual student information.
How a family can use the answer
After requesting clarification, summarize each claim in one sentence: “Among this defined group, during this period, this many students or applications achieved this outcome.” If the provider supplies an institutional comparison, add a second sentence explaining whether the populations are actually comparable.
Then classify what you have. A documented case shows that one result occurred. A defined cohort shows how often a result occurred within that cohort. A well-aligned descriptive comparison shows how two measured groups differed. A credible causal study investigates how much the service itself changed the outcome. A provider may offer valuable work without possessing the strongest form of evidence, but its marketing should not imply evidence it has not supplied.
Do not make admission percentages the only purchasing criterion. Ask what the counselor will do for the student's current needs, how the student responds to the working relationship, and whether the agreement is affordable. A rate can inform those questions; it cannot answer them.
For related evidence reviews, the research collection provides additional context. Keep the central distinction visible when reading them: accurate historical reporting is useful, and a personalized claim of improvement requires more than dividing one published percentage by another.
References and source notes
- Common Data Set Initiative. Definitions and standards. Accessed November 10, 2026.
- MIT Institutional Research. 2025–26 Common Data Set, section C. Fall 2025 first-year admissions data; accessed November 10, 2026.
- Barr, A., and Castleman, B. (2025). Increasing Degree Attainment among Low-Income Students: The Role of Intensive Advising and College Quality. American Economic Review, 115(11), 4075–4103. DOI:10.1257/aer.20240669.
- Phillips, M., and Reber, S. (2022). Does Virtual Advising Increase College Enrollment? Evidence from a Random-Assignment College Access Field Experiment. American Economic Journal: Economic Policy, 14(3), 198–234. DOI:10.1257/pol.20200515.
- Gurantz, O., Pender, M., Mabel, Z., Larson, C., and Bettinger, E. (2020). Virtual Advising for High-Achieving High School Students. Author manuscript; published in Economics of Education Review, 75, 101974. DOI:10.1016/j.econedurev.2020.101974.
- Federal Trade Commission. FTC's Endorsement Guides: What People Are Asking. Guidance addressing the 2023 revisions; accessed November 10, 2026.
- The Ivy Institute. Deidentified application-prioritization meeting-note entry. November 2026. Unpublished company operational record; selected, not representative.
- The Ivy Institute. Deidentified activity-description verification email. October 2026. Unpublished company operational record; selected, not representative.
Research questions or corrections? Contact The Ivy Institute. This report distinguishes public evidence, selected internal observations, and original planning frameworks. It does not promise an admissions result.
