What a Small TOEFL Prep Platform Learns From Its Own Data
Most TOEFL prep advice is written as if every student has months to spare. The data behind FluentPrep AI, a small bootstrapped TOEFL practice and mock test platform, suggests something different: many test takers arrive with a date already booked and only a few weeks to work with. We spoke with founder Noah van der Vleuten about what the platform's own numbers show, where its scoring has missed, and what students tend to get wrong right after their first scored mock.
When the mock score and the real score diverge
The first question was about accuracy. Has a practice score on FluentPrep ever failed to match what a student actually walked out of the real TOEFL with?
Van der Vleuten was candid about the limits of what he can currently claim. "Honestly, I only know of one case so far, because very few students come back to tell us their real score." In that case, a learner scored a 5.5 band on the mock and then earned a 6.0 on the real exam in August, meaning the mock ran half a band too strict for her.
He was careful not to generalize from a single data point. "One person is not enough to say how accurate the mock is, and it's also constantly evolving, and I don't want to claim more than I know." That caution has shaped a change in how the platform collects feedback: every student who finishes an exam is now asked to send in their real score. Over time, that stream of reported results should allow the team to check the mock properly and adjust it.
For test takers, the practical takeaway is to treat any single mock score as an estimate rather than a verdict. A mock that is half a band off in either direction can change how a student feels about their readiness, but it says little on its own about what will happen on test day.
The gap most prep guides leave out
The second question turned to what users say that the TOEFL prep pages they find online tend to miss. The answer was about timelines.
"Most students prepare in weeks, not months," van der Vleuten said. The platform asks departing subscribers why they are leaving, and more than half say it is because they have already taken their exam. For those students, the usual time between joining and cancelling is about four weeks, and several wrote variations of "my exam is in a month."
That sits awkwardly against the material many students encounter. "Many prep guides describe long study plans of three to six months, but most of our students already have a test date booked and only a few weeks left. They need to know what to practice first, not a long course."
This is a familiar problem in test preparation generally: advice written for an ideal timeline gets applied to a compressed one. A student with four weeks left cannot follow a six-month syllabus, and trying to may leave them spreading effort thinly across every section instead of addressing the weakest one. The takeaway is to match the plan to the calendar you actually have, and to prioritize ruthlessly when time is short.
The decision students make too quickly
The third question asked what students decide too fast after their first scored mock test. Here the platform's internal review produced the most concrete finding.
"Many decide the score is the final answer and stop there," van der Vleuten said. When the team looked at its data in September, more than half of the people who visited on only one day took the full two-hour mock test and then never answered a single practice question. By contrast, students who did even a few practice questions after their mock were almost three times as likely to come back.
His framing of what the score is for was direct: "The mock score is meant to be the starting point: it shows you which section to work on. The students who improve are the ones who start practicing that same day."
That is a useful reframe for anyone preparing for a standardized test. A TOEFL mock test is a diagnostic instrument, and a diagnostic that is not followed by targeted work produces no improvement. The students who gain from a mock are the ones who read the section breakdown and immediately begin on their weakest area, rather than filing the number away as a prediction.
What the data can and cannot say
Across all three answers, van der Vleuten drew a consistent line between what the platform observes and what it can prove. The sample of reported real scores is still tiny. The retention figures come from one September review. The platform is, by his own description, a very small bootstrapped startup that has been running for almost a year.
That honesty is itself part of the practical value here. Mock tests and prep platforms are tools for directing effort, not for forecasting outcomes with certainty. The students who benefit most appear to be those who use the score to decide where to start, and then start the same day.
For readers comparing options, our roundup of online TOEFL mock test platforms covers FluentPrep AI alongside others in the category.
