What to Do When Your Practice Score Stops Improving

Every serious exam candidate meets the plateau: weeks of honest study, and the practice score sits exactly where it was. The standard responses, study harder or despair, are both wrong, because a plateau is not a motivation problem. It is information. Scores flatten for four findable reasons, and each has a different fix, which is why "more of the same" is the one response guaranteed not to work.

First, define a real plateau: three or more weeks of completed sessions with no movement in the same measured skill. One flat mock is noise; see cause four.

Cause 1: practicing without feedback

The most common plateau by far, and it lives in Writing and Speaking. Reading and Listening come with answer keys, so errors self-correct. Rated skills do not: write forty essays with no one scoring them against the criteria and you are not practicing improvement, you are rehearsing your current band, mistakes included.

The fix: stop producing and start closing the loop. Every essay and speaking attempt gets criterion-level review, then one targeted second attempt fixing the single named weakness. A teacher does this well; AI evaluation does it instantly and cheaply between sessions (Verbola's writing feedback and speaking evaluation score the official criteria; treat any AI score, ours included, as a practice estimate, never an official result). The self-study versus tutoring guide covers how to budget this.

Cause 2: the wrong practice type for your stage

Practice has stages, and staying in a comfortable one flattens the curve. Untimed accuracy work keeps helping until it does not; then only timed pressure moves you. Full mocks build stamina but are terrible for fixing a specific question type; drills fix question types but never train pacing.

The fix: change the practice type, not the volume. Stuck on untimed accuracy? Everything goes on the clock. Stuck on mocks? Two weeks of single-type drills aimed at your log's worst category. Stuck on drills? Reassemble into full timed sections. The weakest-skill diagnostic tells you which stage you actually need.

Cause 3: fossilized errors

Some errors stop being mistakes and become habits: the article you always drop, the tense you drift into, the pronunciation blur you no longer hear. They survive ordinary practice precisely because practice reinforces them.

The fix: targeted extraction. Pull the two or three highest-frequency errors from your feedback history, and attack them outside full tasks: micro-drills, sentences written specifically around the structure, recordings listened to for one sound only. Then re-enter full tasks with a final-check routine that hunts exactly those errors. Fossils crack under specific pressure, never under general volume.

Cause 4: measurement noise

Sometimes the skill is moving and the measurement is not. Different practice sources vary in difficulty; a mock taken exhausted at midnight measures fatigue; rated-skill self-scoring wobbles by half a band on mood. Panicking over noise leads to plan-thrashing, which causes a real plateau.

The fix: standardize measurement before reacting to it. Same source series, same conditions, same time of day, and judge trends on three data points, never one. A progress view that tracks per-skill trend lines makes the signal visible; so does a simple spreadsheet, if it is honest.

If it is genuinely none of these

A long, real plateau across all skills sometimes means the ceiling of your current general English, especially at higher bands where exam technique runs out of room. That is not failure; it is a phase change. The move is input volume (reading, listening, vocabulary growth) for some weeks, and possibly a look at whether your target and timeline still fit. Plateaus end when something changes on purpose. Pick the cause, change the one matching input, give it two weeks, and measure again.