The band I actually got

I sat the real IELTS Academic test on September 19. The Test Report Form arrived on the 21st. Overall band 6.0, CEFR B2. Component by component: Listening 6.0, Reading 6.0, Writing 5.5, Speaking 6.0.

I am publishing that, with my real name attached, on the site I built to help people prepare for exactly this test. It felt like the only honest way to close the loop on eleven weeks of work.

The number I forgot to save

Back in July I shipped a page asking strangers to do something very specific: write down the band ParrotTalk gave them before their exam, then come back after the real result and tell us both numbers. That page has been live since September 13, and until today it had collected exactly zero pairs. I built the form. I did not fill it in for myself.

I do not have the precise band ParrotTalk gave me before September 19. I remember it sitting somewhere around 6 overall, which is close to what I actually got, but "I remember it being close" is not a data point, it is a feeling, and the whole reason that proof page exists is that feelings are exactly what it was built to replace with two real numbers side by side. I was the first person in a position to give it a clean pair, and I did not, because I was too busy shipping the next thing to screenshot my own dashboard.

So this is not a calibration result. It is one person's honest account of what happened, with a gap in it I should not have left. If you are about to sit your own exam, the fix is one click: open the proof page, write down what ParrotTalk tells you today, and come back when your Test Report Form arrives. One real pair is an anecdote. A hundred is the answer I actually want.

What this one result does and does not prove

Writing was my lowest component, 5.5 against 6.0 everywhere else. That happens to line up with something the feedback on this site keeps saying: more candidates report their Writing band as lower than expected than they do for Speaking. I would love to tell you that is confirmation. It is not. One result cannot confirm a pattern measured across hundreds of people, and treating it as if it could would be exactly the kind of sloppy reasoning I would not accept from anyone else writing about this site. What I can say is that nothing about my own score contradicted what the feedback has been showing, and that is worth exactly as much as it sounds: not proof, not nothing either.

The honest version is smaller and less satisfying than a confirmed theory: Writing was hard for me, same as it is for a lot of the people who use this site, and I do not yet know whether that is the exam being strict on Writing in general or ParrotTalk's own grading running a little severe on that one component. The real calibration data, once enough of it exists, will say which.

Using my own site under exam pressure

I practiced on ParrotTalk the way anyone else would, in the weeks before the test, switching between sections the way the real exam forces you to, with a timer I could not pause and a correction I could not argue with. Building a tool and actually depending on it for your own result are two different kinds of trust, and the second one is the only kind that counts. Band 6 is not a triumphant number. It is a real one, earned on a Friday morning in a test centre, not inflated by me owning the grader.

If you have already sat the real exam and tried ParrotTalk beforehand, the proof page is where that comparison actually becomes useful to the next person.