How Senior Leaders Read Research

The disagreement has to arrive with its evidence

The preference is not for drama. Respondents volunteered exactly what the split must carry before they will move on it — and several said a split without that evidence is worse than useless: "If it stays at opinions, neither version gets my name on it."

And the counterweight that shapes the whole deliverable: 71 of the 100 decide off the split, then build the clean consensus one-pager themselves for the people above them. As one CFO-level respondent put it, "I can make A out of B, but not the other way around." The consensus line is not wrong — it is the output the reader manufactures, never the input they trust.

Finance and the C-suite agree on the answer, differ on the reflex

Both panels landed 46/50 on the disagreement and 48/50-and-46/50 on the two-page brief. Where they part is what they do with a report once it lands.

The real disagreements are about trust, not format

The format questions were near-unanimous. The genuine fault lines run through how a senior reader decides to believe a number in the first place.

The tells that read as low-effort or machine-made

Asked what makes them stop reading, the two panels named the same fingerprints. The reframe that matters commercially: "I do not care who wrote it — human or machine — if it is thin, vague, or unverifiable, it is not getting into my IC deck." The objection is verifiability, not provenance.

Missing methodology, chart crimes (unlabeled axes, 3D pies, truncated scales) and buzzword prose were each named by all 100 as reasons to stop reading. Sanitized verbatims are the tell that cuts closest for synthetic research: buyers authenticate by quote texture, so every quote on this page is untouched.

Trust is mechanical, and it is quantified

Before a senior reader puts their name behind a finding, the report has to clear a specific bar — the same bar in both panels.

Nobody forwards the report

All 100 described the same pipeline. The original document has a working life of about 48 hours; after that it lives as fragments inside the reader's own artifacts. What travels sideways and down is the full file; what travels up is a one-pager the reader wrote themselves.

Asked to design it, they converged on a bundle

The magic-wand question — design the perfect research deliverable for your own desk — produced one shape across both panels: a short decision brief, an exportable data pack, and version-frozen everything.

Roughly 45 of the 100 went further and built the disagreement structure into the opening of their own dream deliverable, unprompted — 16 making the split the literal lead. Charts carry the argument (~70% of the surface); narrative is connective tissue; verbatims run 2–8, each tagged with role, segment and date, and used to show dispersion, never to prove a point.

The one page

The version a reader would take upstairs — printable, liftable, and honest that it is the summary layer over a contested picture.

Who answered

Recruited and curated from a census-grounded synthetic population of 340,000 U.S. residents. Each panel was over-recruited to 120 and scored on the reality described in each persona's biography — job scope, budget authority, reports — not on the job title alone.

On the recruitment. The seniority bar was set honestly and it did not fall evenly. A genuine C-suite panel of 50 exists in the population — 120 executive recruits yielded 36 who clearly run an organization (CEO, COO, CFO with described board, budget and staff authority). A VP-and-up finance panel at that density does not: 120 finance recruits surfaced only three who unambiguously own the finance function. This finance panel is therefore senior finance decision-makers — function heads, senior FP&A, established advisors with real books — curated from 120 and disclosed as such, not a panel of CFOs. The pool tops out where real US org charts thin out.

How this study was made — and where to be careful