Glossary
Published · 4 min read
In 1960, the psychologist Jacob Cohen published a short paper in Educational and Psychological Measurement arguing that when two independent reviewers assess the same material, raw agreement is misleading. Some of it is method. Some of it is coincidence. Cohen's kappa exists to subtract the coincidence and show what's left.
Conversion audit methodology refers to the process and scoring model behind a conversion diagnosis: how a tool or consultant decides what to examine on a page, how each finding is scored or ranked, and what evidence backs each conclusion. One answer to that, and the one this entry works from, is a dual-plane methodology: score the page twice, once for each of two very different readers. The human plane asks whether a person can understand this page, trust it, and act on it. The agent plane asks whether an AI agent can extract the facts on this page, verify them, and act on them. The full writeup lives at /methodology.
Cohen's problem transfers directly. Hand the same signup page to two consultants and you get two different finding lists. Without a fixed scoring model, nobody can say how much of the overlap was method and how much was two experienced people happening to notice the same headline.
A score without a visible methodology is an opinion with a number attached. What makes a score checkable is the machinery underneath it: a fixed set of dimensions applied to every page in the same order, evidence tied to each individual finding, and annotated screenshots showing the specific element the finding refers to. That combination is what lets a skeptical Head of Growth disagree with a finding on its merits rather than deciding whether to trust the person who wrote it.
Find out why your page loses people
Drop a URL. Get the first audit free when Nudgent opens.
A traditional CRO consulting deliverable is a human analyst or agency team reviewing a page against best-practice patterns and their own experience, usually over days to weeks, producing findings in a report or deck. It is genuinely deep and contextual. The cost is time, budget, and the reliability problem Cohen described: two reviewers, two lists, no way to measure the gap.
A dual-plane audit applies the same scored framework to every page, with annotated evidence attached to each finding, in minutes. It trades some contextual depth for consistency you can audit.
Every page is scored across 14 dimensions, 7 human-plane and 7 agent-plane, each rated 0 to 10 on a health scale where 10 is excellent and 0 is broken. A plane's overall score is the sum of its seven dimension scores multiplied by 100/70, producing a 0-100 composite per plane. That arithmetic is fixed and public, written up at /methodology, so you can recompute any score we give you. The 14 dimensions themselves are published for the same reason.
Here is the experiment worth running before you trust anyone's audit, including ours. Take one page. Hand it to two people on your team separately. Ask each for their top five conversion problems. Compare the lists. How much overlaps, and has anyone on your team ever actually checked?
A conversion audit methodology is the documented process and scoring model behind a conversion diagnosis: which dimensions get examined, how each is scored, and what evidence supports each finding. A methodology that isn't published can't be checked.
A traditional deliverable is a human review producing narrative findings, deep in context but inconsistent between reviewers. A dual-plane audit applies a fixed 14-dimension framework identically to every page, so two audits of the same page produce the same score.
Yes. Each finding is tied to annotated evidence showing the specific page element it refers to, so you can evaluate the finding rather than the source.
Find out why your page loses people
Drop a URL. Get the first audit free when Nudgent opens.
Get early access