20+ years
Test-prep specialization
Bay Area + national
In-person and remote programs
Make It Stick methodology
Anchored in cognitive-science research
First-party outcomes
Our students, our published numbers

📊 The headline numbers

Average score gain, by where students start.

Multi-year analysis of students who completed the program, current through 2026. Each card shows the average improvement for students starting in that range.

ACT · starting 27 or below
+5.7
ACT points · average improvement
From mean 22 (69th pctile) ending mean 28 (90th pctile)
ACT · starting 28+
+3.4
ACT points · average improvement
From mean 30 (94th pctile) ending mean 34 (99th pctile)
SAT · starting 1350 or below
+190
SAT points · average improvement
From mean 1210 (77th pctile) ending mean 1400 (93rd pctile)
SAT · starting 1350+
+120
SAT points · average improvement
From mean 1400 (93rd pctile) ending mean 1520 (98th pctile)

Eligibility: students who completed 12+ instruction hours and 80%+ of assigned homework. Starting scores from official PSAT, decommissioned SAT, or decommissioned ACT — taken before instruction began.

⭐ The story in the SAT data

Below-elite students end the program where elite students start.

Students entering the program scoring below the elite tier (≤1350) gain +190 SAT points on average — landing at a mean of 1400, the same percentile where our elite-tier students begin (93rd percentile).

And many don't stop there. 63% of below-elite SAT students gain 200+ points; 20% gain 300+ — meaning they cross the elite line and keep climbing into 1510+ territory (top 2%).

Start 1210
End 1400
▼ Elite tier (1350)
20% reach 1510+

Yellow line = the 1350 SAT cut-off between "below elite" (≤89th percentile) and "elite" (90th+). The cohort's mean ending score sits well past it.

🔬 vs. published research baselines

How does this compare to typical test prep?

ACT's own published review of test-prep research (College Entrance Exams: How Does Test Preparation Affect Retest Scores?) summarizes the average gain a student can expect from a typical retest with preparation. The benchmarks: about +1.4 points on the ACT (Schiel & Valiga, 2014) and about +30 points on the SAT (NACAC review, Briggs 2009). Meta-analyses on SAT-like tests put it at 0.25 standard deviations — roughly 1.3-1.7 ACT points (Bangert-Drowns et al., 1983).

Test Prep Gurus students gain several multiples of those benchmarks.

ACT, starting 27 or below
4.1× the typical retest gain
Typical retest
+1.4 ACT pts
TPG students
+5.7 ACT pts
ACT, starting 28+
2.4× the typical retest gain
Typical retest
+1.4 ACT pts
TPG students
+3.4 ACT pts
SAT, starting 1350 or below
6.3× the typical retest gain
Typical retest
+30 SAT pts
TPG students
+190 SAT pts
SAT, starting 1350+
4.0× the typical retest gain
Typical retest
+30 SAT pts
TPG students
+120 SAT pts

Source studies (all cited in the ACT.org review paper)

  • Schiel & Valiga (2014a) — students who took and prepared for a 2nd ACT gained an average of 1.4 ACT Composite points.
  • Schiel & Valiga (2014b) — students preparing 20+ hours scored only 0.7 ACT points higher than those preparing 3-6 hours.
  • Briggs (2001) — commercial prep classes and tutoring produced gains that did not exceed a full ACT subject-test point.
  • Briggs (2009) / NACAC — review of SAT prep efficacy studies found average gain of approximately 30 SAT points.
  • Bangert-Drowns et al. (1983) — meta-analyses on SAT-like prep show a modest 0.25 SD gain — about 1.3 to 1.7 ACT points.
  • Powers (1993) — short-term prep activities like commercial workbooks produced 1.2 to 1.5 ACT Composite point gains.

Caveat: the cited studies span different test versions and prep formats, and the SAT figure includes a wide variety of programs in its average. We use these as honest reference points, not as a like-for-like control group.

🎯 ACT bucket

ACT: starting below the elite tier

ACT students starting 27 or below.

Starting score
22
69th percentile
Mean gain
+5.7 pts
≈ +220 on the SAT
Ending score
28
90th percentile

Percentile journey (0 → 100th)

Start
End
69th
90th
median

Achievement breakdown

97%
improved by 3+ points (≈ +120 SAT)
68%
improved by 5+ points (≈ +200 SAT)
21%
improved by 8+ points (≈ +320 SAT)

🎯 ACT bucket

ACT: starting in the elite tier

ACT students starting 28 or higher (90th percentile+).

Starting score
30
94th percentile
Mean gain
+3.4 pts
≈ +140 on the SAT
Ending score
34
99th percentile
Gap closed to perfect
58%
of the way from start to 36

Percentile journey (0 → 100th)

Start
End
94th
99th
median

Achievement breakdown

98%
scored 31+ (top 5%)
77%
scored 33+ (top 2%)
41%
scored 34+ (top 1%)

🎯 SAT bucket

SAT: starting below the elite tier

SAT students starting 1350 or below.

Starting score
1210
77th percentile
Mean gain
+190 pts
SAT composite
Ending score
1400
93rd percentile

Percentile journey (0 → 100th)

Start
End
77th
93rd
median

Achievement breakdown

93%
improved by 100+ points
63%
improved by 200+ points
20%
improved by 300+ points

🎯 SAT bucket

SAT: starting in the elite tier

SAT students starting 1350 or higher (90th percentile+).

Starting score
1400
93rd percentile
Mean gain
+120 pts
SAT composite
Ending score
1520
98th percentile
Gap closed to perfect
60%
of the way from start to 1600

Percentile journey (0 → 100th)

Start
End
93rd
98th
median

Achievement breakdown

96%
scored 1430+ (top 5%)
67%
scored 1500+ (top 2%)
38%
scored 1530+ (top 1%)

🎬 Real students, real interviews

Behind the averages: students you can hear from.

Averages tell one story; individual journeys tell another. Watch (or read) the students and parents who lived through the kind of score jump the data above describes.

ACT · +12 points
2234
Ella Ryan
69th99th+ percentile

From a 22 (69th percentile) to a near-perfect 34 — a 12-point jump that put Ella in the top 1% of all ACT takers.

Watch Ella's interview →
ACT · +7 points
2532
Luke Stark
Parent interview with Diana Stark
83rd97th percentile

Luke went from a 25 (83rd percentile) to a 32 (top 3%) — and his mom Diana sat down with us to talk about the journey.

Watch Diana's parent interview →
⭐ Perfect 36 — top 1 in 3,500
2836
Alyssa Kerscher
Santa Margarita Catholic High School, CA
91stperfect percentile

Alyssa came in at a 28 (already in the 91st percentile) and finished with a perfect 36 — a feat earned by only ~300-500 students per year out of ~1.7 million ACT takers.

📖 Read Alyssa's story →
📚 The long view
Tommy Poletti
Now at USC Law — Sept 2024 interview

The strategies students learn at TPG don't end with the SAT or ACT. Tommy is now at USC Law, and the self-testing, color-coded review, and growth-mindset framing he picked up here are how he prepares for law-school exams.

"The hard way is usually the right way."
Watch Tommy's interview →

Perfect-score context: media outlets place perfect ACT scores at approximately 300-500 per year out of roughly 1.7-2 million annual ACT takers — about 1 in every 3,500 students. (We're awaiting an official figure from College Board / ACT research.)

🔧 The method behind the numbers

How we move scores: four evidence-based techniques.

The gains above aren't from doing more of what schools already do. They come from a small set of high-leverage learning practices grounded in cognitive-science research — most prominently Make It Stick (Brown, Roediger & McDaniel, Harvard University Press, 2014). Four practices we lean on hardest:

📝

Self-testing (retrieval practice)

Pulling information out of memory beats re-reading it. Every session is built around recall under pressure, not exposure — because the test itself is recall under pressure.

🔗

Elaboration

Connecting new material to what a student already understands. We don't just teach problem types — we build the mental models that survive once the problem looks different on test day.

⛰️

Desirable difficulties

Easy practice feels productive but doesn't last. The right kind of hard — interleaving topics, varying conditions, spaced repetition — builds durable skill that holds under exam stress.

🌱

Growth mindset, earned objectively

Most growth-mindset coaching fails. What works is letting a student watch their own score move in an area they thought was fixed — then helping them connect the dots between effort, strategy, and result.

🧠 Why the score gain matters beyond the score

A score change is also a self-fulfilling prophecy.

Most students who arrive at TPG below their target are certain they can't move. They've tried before. Teachers and parents tell them otherwise — "you can do this, you have a growth mindset" — but the words bounce off, because the student has internal evidence that says they've already failed. The technical term is learned helplessness, and pep talks rarely fix it.

The one thing we've found that does fix it is an objective experience of improvement in the exact area a student was sure they couldn't change. A score gain isn't just a number for the application — it's the first piece of hard evidence that contradicts the story they've been telling themselves. From there, a new self-fulfilling prophecy installs itself: they can succeed at hard things, and they start acting like it. Some carry it into college. Tommy, who's now at USC Law, carried it past that.

"The hard way is usually the right way." — Tommy, TPG student, now at USC Law

Tommy still uses the same techniques he picked up here — self-testing, color-coded review, elaboration — to prepare for law-school exams. The score change was where it started.

Why we measure elite-tier students by "gap closed to a perfect score"

Raw point gains are capped by the test ceiling: a 1530 SAT student can only gain 70 points before hitting 1600. By raw-gain math, that student looks like an underperformer — even if they scored a perfect 1600. So for students who start near the top, we ask a simpler, fairer question: how much of the distance from their starting score to a perfect score did they actually close?

Gap closed = actual gain ÷ (perfect score − starting score)

A 1530 → 1600 student closes 100% of the gap. A 1530 → 1565 student closes 50%. Our elite ACT students close 58% of the gap on average; our elite SAT students close 60%. Both numbers are well above what one extra cycle of self-study typically produces — at 60%, the average elite SAT student moves from 1400 to 1520, the last 80 points of which are the hardest to earn.

📅 Digital SAT update (March 2024 onward)

ACT vs. SAT: what changed when the SAT went digital

During most of this analysis (2017–2024), about 65% of our students chose the ACT — usually on our recommendation. The reason was simple: the old paper SAT covered a broader range of material, which made big short-term gains harder to engineer than on the ACT.

The digital adaptive SAT (March 2024+) overhauled the question structure and narrowed the content range. Our data on students since then shows the score-gain pattern on the digital SAT has caught up to the ACT — when students apply our learning strategies, we see similar dramatic improvements on either exam. The choice between SAT and ACT is now closer to a fit question than a points question.

How we measured this — methodology

Window
Multi-year analysis of student outcomes — original cohorts from 2017–2024 finalized December 2024, refreshed through 2026 as new cohorts complete the program.
Who counts
Students who completed at least 12 instruction hours and turned in at least 80% of assigned homework. This filter excludes students who started and stopped — those would inflate the starting baseline without producing the corresponding ending exam. We measure who did the work.
Starting scores
Drawn from official PSAT exams, decommissioned official SAT exams, or decommissioned official ACT exams — completed before instruction began. PSAT is treated as a valid starting SAT baseline because the PSAT is a validated SAT predictor; the Pre-ACT is not used as a starting ACT baseline because it has not been shown to be a calibrated predictor of the live ACT.
Ending scores
Each student's official final SAT or ACT score — the score they actually submit to colleges.
Buckets
Students are split into below elite (starting in the 89th percentile or lower) vs. elite (starting in the 90th percentile or higher). The line falls at ACT 28 / SAT 1350. Different buckets call for different success metrics — raw-point gain for below-elite, I/P ratio for elite — for the reasons explained in the I/P box above.
Percentiles
A student's percentile rank represents the percentage of students with scores equal to or lower than their score — the standard College Board / ACT convention.

📚 The deeper story

Want the cognitive science behind it?

Our methodology is anchored in Make It Stick: The Science of Successful Learning by Peter Brown, Henry Roediger III, and Mark McDaniel (Harvard University Press, 2014) — the book that translates 30+ years of laboratory research on how the brain actually learns into specific, testable techniques. We invited the team onto our channel to talk through the practical implications for high-school test prep.

▶ Watch the Make It Stick interview

Your score is the most coachable part of your application.

Most things in college admissions can't be changed in six months. Your test score can — and these numbers are what that change looks like with structured prep.