Band descriptors should become one revision step.

Common mistake

Memorising criterion names does not change essay evidence by itself. The value is knowing which criterion holds the target band back.

• Whether four criteria are separated
• Whether one main blocker is chosen
• Whether revision can verify evidence change

What the 6/7 boundary actually says

The public descriptors in plain words: TR at 6 is "relevant ideas, insufficiently developed"; at 7, "ideas extended and supported" — the line is the mechanism sentence. CC at 6 is "cohesion effective but mechanical"; at 7, "logical flow, clear progression" — the line is logic doing the linking rather than linkers. LR at 6 is "adequate range, attempts at sophistication with errors"; at 7, "flexibility and precise less-common items" — the line is accuracy, not word difficulty. GRA at 6 is "accurate simple sentences, error-prone complex ones"; at 7, "frequent error-free sentences" — the line is error rate.

The same idea written as a Band 6 and a Band 7 paragraph

Band 6: "Online learning is convenient. Students can study anytime and anywhere. For example, many people use online courses. Therefore, online learning is popular." Every sentence is correct, but the second restates the first, the example has no specificity, and the close circles back. Band 7: "Online learning removes the fixed timetable that excludes working adults. A nurse on rotating shifts, for instance, can complete a degree module at midnight after work — something no campus schedule allows. This flexibility explains why enrolment growth now comes mostly from people over 25." Same four sentences; each moves one step forward.

The three most common misreadings of the descriptors

Misreading one: "more advanced words, higher score" — from Band 7, LR punishes inaccurate difficult words; a wrong "utilize" costs more than a right "use". Misreading two: "long complex sentences show grammar level" — GRA wants range plus accuracy; a triple-nested clause with a broken tense is a net loss. Misreading three: "more words, better TR" — the descriptors care whether each idea is extended; two fully developed ideas reliably beat four mentioned ones.

Turning descriptors into post-writing checks

After each essay, run one 30-second check per criterion — more effective than rereading the whole text. These four actions map directly onto the four 6-to-7 boundaries; whichever fails becomes the revision target for this essay.

• TR: under each claim, count the because/therefore chain
• CC: delete every linker and reread — does the paragraph still flow
• LR: circle your three hardest words; verify meaning and collocation
• GRA: check only your known high-frequency error structures