What Evaluators Really Want To Read
Evaluators score against a rubric under time pressure. Writing for how they actually work is the cheapest score improvement available.
An evaluator is not your ideal reader. They are usually an operational manager who has been handed a scoring pack on top of a full workload, reading eight submissions against a published rubric, and they will have to justify every mark they give to a moderation panel. Write for that person and your scores move.
They score, they do not read
Evaluation is a search task, not a reading experience. The evaluator is looking for evidence that maps to a descriptor such as "demonstrates a comprehensive understanding, supported by relevant evidence". Your job is to put that evidence where it can be found in seconds.
Mirror the question's wording in your first sentence. Use sub-headings that match the sub-criteria exactly. If the question has four parts, give it four labelled sections in the same order. Moderators who cannot locate evidence tend to score the middle band.
Give evidence in a consistent shape
The most reliable pattern is context, action, result, relevance. Set out the situation briefly, say what you did, quantify what changed, then connect it explicitly to this buyer's circumstances. That last step is the one most bidders skip, and it is the one that turns a case study into a score.
Keep the numbers honest and specific. Percentages without a baseline invite scepticism. "Complaints fell 22 percent, from 148 to 115 per year, across a comparable 30-site contract" survives moderation. "Significant improvement in complaints" does not.
Make the marking easy
Bold the metric, not the marketing language.
Use tables for anything comparative, such as resourcing, SLAs or mobilisation milestones.
Keep paragraphs under five lines so the page does not look like a wall.
Spell out acronyms once and only use those the specification itself uses.
Never split one requirement's evidence across two distant sections.
What loses marks quietly
Long preambles before the point. Claims about your organisation that have nothing to do with the question. Graphics that carry information not repeated in the text, which fails if the pack is printed in greyscale. Repetition across answers, which reads as padding when the same paragraph appears three times in one submission.
The greyscale, eleventh-submission test
Before you submit, print a section in black and white, read it standing up, and time yourself. If you cannot find the evidence for each sub-criterion within about 30 seconds, the evaluator will not either. That single test tends to be worth more than another round of polishing the prose.