Here is the offer in one sentence: every Pro tier now carries a written, diagnostic-banded score-lift guarantee — and the remedy is in the contract, not the marketing copy. This is the first version of that document. The bands are below. The eligibility is below. What we will do when we miss is below.
We have been quietly making this promise to candidates across the alpha and the first Pro cohorts since the spring. Writing it down is the part that has taken until November.
The bands
The guarantee applies to every Pro tier — 1-month, 4-month, and the new 6-month — at the lift amount determined by the candidate's diagnostic score. Higher diagnostic, smaller promised lift; the reason is in the next section.
| Diagnostic score | Guaranteed lift | Endpoint |
|---|---|---|
| 205–505 | +110 points | 615 |
| 515–555 | +80 points | 635 |
| 565–605 | +60 points | 665 |
| 615–645 | +50 points | 695 |
| 655–675 | +40 points | 715 |
| 685–705 | +30 points | 735 |
| 715–745 | +20 points | 765 |
The same table applies to every Pro tier. The eligibility criteria below — three full Mock Exams, sixty hours of adaptive practice — are tight on a 30-day window. Pro 1-month candidates who cannot finish them inside the window simply do not trigger the guarantee; they are not penalised for it, and the tier still ships the engine, the Mock, and the analytics surface in full.
Why the lift shrinks as the diagnostic rises
The first question candidates ask is why a 420 candidate is promised +110 and a 720 candidate is promised +20. At first glance it looks like we are less confident in our product at the top end.
It is the opposite. The lift shrinks because the variance in score gain shrinks. Expected variance in score outcomes is empirically lower as candidates approach the upper band — this is consistent with the published standardized-test variance literature and visible in any honest analysis of GMAT score-improvement data. A candidate at 420 has more raw point-room above them and the marginal item they get right is worth more in the scoring curve. A candidate at 720 has less room and a steeper marginal cost per point.
Promising +110 from 720 would be either dishonest or a coin flip. Promising +20 is something we can stand behind and write into a contract.
Eligibility, in full
The guarantee is real, which means it has terms. The terms are the same ones we would ask of any candidate we were genuinely trying to coach to their target: do the work.
- Full diagnostic completed within the first two weeks of the subscription. This establishes the starting band.
- Three full-length Mock Exams completed inside the guarantee window, spaced out — not stacked in the final week. Mock Exam access is included in every Pro tier and is uncapped during the public beta.
- At least 60 hours of adaptive practice inside the guarantee window. The engine logs hours; we do not ask candidates to self-report.
Candidates who hit all three thresholds and miss their banded lift trigger the guarantee. Candidates who do not hit the thresholds do not — because the guarantee is a claim about what happens when the method is used as designed.
What happens if you miss the band
The remedy is renewal. If a candidate completes eligibility and misses the banded lift, the next six months on the engine are free — a new window the candidate can use to close the gap. The engine keeps running, the candidate keeps practicing, the next mock sits. There is no second invoice and no clawback of access.
Renewal is not a placeholder for "send us your complaint and we will see." Once eligibility is logged and the official score is recorded against the candidate's account, the renewal is automatic in the billing system. The conversation is whether the eligibility was met, not whether we feel like honoring the contract.
Why renewal, not a refund
Two reasons we wrote the remedy as a renewed window rather than a refund.
First, a candidate who completes eligibility and misses the band is almost always a candidate the engine's analytics surface can identify a gap for in the next window. Continued access is the path that materially closes that gap. A refund pays the candidate out but leaves the score where it is, which is usually not what the candidate actually wants. A candidate who wanted out would not have signed up to Pro in the first place.
Second, incentive alignment. The renewal mechanic keeps us on the hook for delivering the score the next window, not just for refunding the last one. The contract holds us to the outcome — getting the candidate to their banded endpoint — not to a money-back gesture that squares the books without solving the problem.
What this replaces
Until this morning, the score-lift promise was an unwritten goodwill policy. Candidates who finished a Pro window and missed their target got extended access anyway, on a case-by-case basis, decided in a co-founders' inbox. The case-by-case version worked because the cohort was small. It would not work at the cohort sizes Pro is now serving.
Writing this down does two things. The candidate now knows what the terms are before they subscribe — which is the only version that respects the candidate's time. And we now have a single contract to defend internally, which is the only version that survives growth.
Closing
Every number above is something we are willing to write into a contract and pay out on. The bands are not marketing. The eligibility is not marketing. The renewal remedy is not marketing. They are the terms.
We will publish another note like this one whenever the terms change. They do not change often, and they do not change quietly.
— Nicola & Joel