The finding they will dispute

Somewhere on your page is a line that a specific person is going to read as a judgement about the last three years of their work. They will have an answer ready. They will deliver it in a room where they have context you do not, relationships you do not, and a strong incentive for your line to be wrong.

The instinct at that point is to soften the line in advance so the argument never happens. Resist it, and be clear about what softening costs: a finding written so nobody can disagree with it is a finding that could not have come out any other way, which is the definition of one that changes nothing. You already applied that test to the whole document in the lesson on who reads it. It applies hardest here.

Why the disagreement is structural

Start with the version that is not an accusation. A training publisher that sells finance courses rather than diligence engagements — and therefore has nothing riding on how the process is described — puts the asymmetry like this:

“The seller knows everything about its own business and the buyer knows far less. Making matters worse, the seller is incentivized to hide or downplay negative aspects of the business and exaggerate the positives.” (Wall Street Prep, updated 2023)

That describes a position, not a character. It holds for people acting in complete good faith, and the front-end version of it is entirely mundane: the engineering lead who tells you the design system is adopted believes it, and has never had a reason to count. Their evidence is the screens they work on. Yours is the screens you could reach. Both samples are real and neither is representative.

A diligence firm conceding the limits of its own product — worth weighing given which direction that incentive points — makes the same point about the material you are handed:

“That leaves you at the whims of the company to cherry-pick what you review, which in turn means you aren’t getting a good representative sample.” (Marty Abbott, AKF Partners, 2018 — a firm that sells technical diligence)

Put the two together and the disagreement stops being surprising. Two parties, both sincere, working from differently biased samples, with opposite exposure to the answer. Of course you land in different places. The question is what your document does about it.

What a disputed finding actually changes

Disagreement is only worth provoking if it goes somewhere. A firm selling compressed diligence reads to private equity — a vendor, and one that publishes no date on the page, so treat it as positioning rather than as a dated record — describes a severity ladder that is the clearest published account of where a finding lands. Critical findings reprice the deal or trigger an escrow. High findings need a plan agreed before close. Below that, findings become 100-day items, and below that they go on the company’s own timeline.

That ladder is why the disputed finding is worth the discomfort. The rungs are consequential in descending order, and a finding management successfully talks down does not merely lose the argument — it moves down a rung, from something that changes the price to something that goes on a list. If your evidence cannot hold the finding at its rung, the finding effectively disappears, whatever colour you printed.

Separate the observation from the inference

Most disputes that go badly go badly because two different claims got welded into one sentence, and the sentence can then be knocked over by attacking whichever half is weaker.

Take the welded version:

The design system is not really adopted and the team has been overstating its progress.

There is an observation in there and an inference about people, and the inference is doing all the emotional work while the observation carries all the weight. Management will dispute the inference, because it is about them, and the observation will go down with it.

Now unwelded:

Of the eleven screens reachable without an account, seven use button and form controls that do not match the published component library. That is the observation. My inference is that adoption is concentrated in newer surfaces, which implies consolidation work on the older ones.

Now the argument has somewhere useful to go. Management can accept the count and dispute the inference — those eleven screens are the legacy admin area, scheduled for retirement, and the inference is wrong for a reason you could not have known. That is a productive exchange that improves your page. What they cannot do is make the count go away by objecting to a characterisation of their team, because you did not make one.

Write their rebuttal before they do

The rubric’s disputed-finding block has four lines, and the third is the one people leave blank: what management will say.

Fill it in properly, in their voice, making the best version of their case rather than a straw one. Two things happen. The weak findings collapse at this desk instead of in the room, which is enormously cheaper. And for the ones that survive, you arrive having already thought about the strongest objection, which is the entire difference between answering it and being surprised by it.

If you cannot write their rebuttal at all, that is diagnostic. It usually means you do not understand the thing you are criticising well enough to have found the reason it is that way — and there is almost always a reason, often a constraint you would have accepted yourself.

Name what would change your mind

The fourth line is the strongest one on the page and it costs a sentence: what evidence would change my mind.

Committing to that in advance does three things at once. It proves the finding is falsifiable, which distinguishes it from taste. It hands management a concrete way to win, which converts an argument into a request for evidence they can actually satisfy. And it moves the dispute off you entirely: nobody is arguing about whether you are right, they are arguing about whether a specific artifact exists.

Written well it sounds like this: an adoption report from the design system team, dated within the last quarter, showing coverage across the surfaces I could not reach. If they produce it, you change the colour and say so plainly, and your credibility goes up rather than down. If they cannot produce it, the absence has become the finding, and it is now a finding about how the company knows things — which is usually worth more than the one you started with.

What this does not cover

Everything up to here has been built without running the instrument against anything real. The reader is named, the format is bounded, the thresholds are written, the estimate has assumptions under it, and the disputed finding has a rebuttal beside it. All of that is theory until a product has been through it.

The last lesson is where you pick a real target you can only see from the outside and score it end to end in one sitting. It will break questions. That is the intended outcome — the list of questions your rubric cannot answer without access is the most useful thing this course produces, because it is the part you can only learn by trying it rather than by designing it.