Starting out
Length is not the product
Buyers notice specificity and skim volume, and the two are frequently in tension.
Guides on Starting out: A rubric is a production tool before it is a quality claim, The brief is rarely the request, Getting the first booking, in order
Write specifically by giving every sentence a referent, a comparison or a consequence, and delete any sentence that could move unchanged into someone else's assessment. The most common weakness in paid assessment writing is a piece long enough to look like value and general enough to predate the submission, and the review says "a bit generic".
Specificity is the product. Length is a proxy for it that stops working the moment the buyer reads two paragraphs, and much of it goes unread anyway: Nielsen Norman Group (2008), analysing 45,237 logged page views, estimated that users read about 20% of the text on an average web page.
The substitution test
Take any sentence in your draft and ask whether it would survive being moved, unchanged, into someone else's assessment.
If it would, it is padding. Not bad writing necessarily - often quite polished writing - but it carries no information that required you to have looked.
"There is a lot to work with here and the overall impression is positive" passes into any deliverable ever written. "The second image is the only one with even lighting, and it is the one the assessment mostly rests on" cannot be moved anywhere, because it refers to something.
Run the test on your opening paragraph first. Openings are where general sentences cluster, because they are written while you are still deciding what you think.
What makes a sentence specific
Three things, and any one of them is usually enough.
A referent. It points at a particular thing in the material rather than at the category. A comparison. It says relative to what, and the reader can check the comparison. A consequence. It says what follows, so the observation is not left for the buyer to interpret.
The strongest sentences carry two of the three. "The framing is tight enough that proportion is hard to read, which is why the score on that axis is lower than the rest" has a referent, a consequence, and an implicit comparison.
Numbers count as referents, with a caveat. A number you actually derived is specific; a number you produced to sound precise is worse than a qualitative statement, because it invites a challenge you cannot meet. Where you have no basis for a figure, state direction rather than magnitude - the same discipline that makes reading the market band without guessing work in pricing applies to the writing.
The four padding patterns
Almost all padding falls into one of four shapes, and each has a fix.
| Pattern | Example | Fix |
|---|---|---|
| The preamble | "Before getting into the detail, it is worth saying..." | Delete, start at the detail |
| The restatement | Repeating the brief back in your own words | One clause, not a paragraph |
| The hedge stack | "generally speaking, in most cases, arguably" | One hedge maximum, or none |
| The wrap-up | A closing section summarising what you just said | Cut it; the summary is already at the top |
The wrap-up is the expensive one, because it is often the longest and it is written when you are tired and want the job finished. A deliverable that summarises itself twice reads as a deliverable with one idea in it.
Being specific without being longer
The fear behind padding is that a specific piece will be short and the buyer will feel underserved.
In practice specific writing is denser, not shorter. A generic 600-word assessment usually contains four real observations; a specific one at the same length contains twelve, because the connective tissue between them was where the words were going.
If you genuinely run out of things to say at 300 words, the problem is upstream. Either the material did not support the assessment asked for - in which case say so and offer a remedy rather than inflating - or you looked at it once instead of twice. A second pass over the same material almost always produces the missing half, and it is faster than writing filler, because filler is slow to write and horrible to read back.
The pass that catches it
Do the padding pass separately from the writing pass, and do it against a fixed question rather than by reading for quality.
Read only the first and last sentence of each paragraph. Those two positions hold the majority of general statements, and reading them in isolation makes the vagueness obvious in a way that reading the whole piece does not.
Then read the piece with the substitution test in mind and delete rather than rewrite. Deleting a general sentence almost never leaves a gap. Rewriting one usually produces a longer general sentence.
This pass takes three or four minutes and belongs in the same slot as the final check before you send, which is where the rest of the pre-delivery routine lives.
Why buyers can tell
The reason generic writing fails so reliably is that buyers have already read the generic version, for free, before they came to you.
Automated assessment produces fluent, plausible, non-specific output at zero cost and instantly, and a buyer paying for a person has already seen what an automated score reliably says. Anything in your deliverable that a model could have produced without seeing the submission is priced, in their head, at what the model costs.
That is also the encouraging read. The observations only you can make are the ones with referents in them, and they are the cheapest part of the work to produce once you stop writing around them. Where a claim needs a figure rather than a judgement, cite a convention instead of inventing one - the site that documents what the underlying data supports is the right owner for that, and pointing at it is more credible than a number you improvised.
Buyers themselves describe the difference in almost these terms; their account of what makes a commissioned review worth keeping is largely a description of specificity by another name.
If you want to see what a fully specified deliverable looks like as the platform frames it, the format guidance on Rate Cock's judges page sets the expected shape, and the difference between meeting it and filling it is the subject of this post.