How to build publishable evidence.
The first article on this site introduced the Search Opportunity Framework. The second answered Q1: how do you know what a keyword is really worth? The third answered Q2: how do you read a SERP like a market map, not a scoreboard?
This article answers Q3: can you publish something nobody else can?
Q3 is where most content projects fail. Demand checks out. The SERP looks beatable. The asset potential is real. Then the writing starts — and the evidence never shows up. The draft replaces what should be first-hand data with opinion. The end result reads like everything else on the SERP, and nobody can tell why it should rank above them.
The framework calls this the evidence gap. This article explains how to close it — before you write a single paragraph.
What counts as publishable evidence? Three criteria.
Not everything you know counts as evidence for a content page. Three criteria separate what you can build from what you should not.
Criterion 1: It is about the world, not about you.
“From our experience” is not evidence. It is a credibility claim without verification. “From ten years of experience” is the same claim with a number attached. Neither tells the reader anything falsifiable about the market you are analyzing.
Publishable evidence describes something the reader could check if they had access to the same source. It names the observation, not the observer.
Criterion 2: It is specific enough to be wrong.
Good evidence makes a claim sharp enough that somebody with better data could disprove it. “The market is competitive” is not evidence. “In July 2026, four of the top ten results for this query were forums — none were original research” is.
The specificity of the second version is what makes it useful to the reader. It is also what makes it survivable: if the SERP changes, the next researcher can update the number without discarding the method.
Criterion 3: It is proportionate to the claim it supports.
A single Semrush screenshot does not prove a market is winnable. Two years of ABA data does not require a single anecdote. The weight of the evidence should match the ambition of the claim. This rule keeps you from overclaiming on thin data — the single most common error in content-driven research.
If your evidence does not meet all three criteria, it is not evidence. It is something else — a belief, a preference, a hunch, or a guess. None of those belong in a page that is asking the reader to trust your judgment over the SERP’s.
The four evidence layers. Most content stops at layer one.
Not all publishable evidence is equally valuable. Four layers describe the depth of what you can offer. Most content on the web stops after the first two.
Layer 1: Observation. You looked at something and recorded what you saw. A SERP screenshot. A set of keyword metrics. A list of competitors. Observation is the cheapest layer and the easiest to produce. It is also the easiest for a competitor to reproduce — which makes it the weakest differentiator.
Layer 2: Aggregation. You collected observations across time, categories, or sources and organized them into a pattern. A table comparing four SERPs. A chart tracking keyword difficulty across a category over 36 weeks. Aggregation raises the cost of reproduction. A competitor can screenshot one SERP in thirty seconds. Rebuilding your comparison table takes hours.
Layer 3: Inference. You have enough aggregated data to draw a conclusion the raw numbers do not state outright. “The highest-volume keyword in this category converts fewer searchers into buyers than three lower-volume alternatives” — this is an inference. The data is aggregated. The conclusion is yours. Inference is where the evidence starts becoming an asset rather than a reference.
Layer 4: Framework. You have generalized the inference into a repeatable method that works across categories. The Search Opportunity Framework itself is layer-four evidence. It is not about one market. It is a method for evaluating markets — and the method is the evidence.
Most content on the web operates at layers one and two. The best content on the web — the kind that earns links, citations, and defensible rankings — operates at three and four.
The framework’s Q3 is asking you a simple question: which layer can you reach with the evidence you actually have?
The evidence map: a pre-writing exercise.
Before you write a word of the article, map your evidence against the claims it supports. This exercise takes fifteen minutes and saves you from discovering mid-draft that your best argument rests on a single tool screenshot.
The map has four columns:
| Claim | Evidence source | Layer | Publishable? |
|---|---|---|---|
| The SERP is fragmented | Semrush Top 10, 4 queries, July 2026 | Layer 2 | Yes |
| The market has real demand | ABA 36 weeks, category aggregate | Layer 3 | Yes (anonymized) |
| Nobody has published original research | SERP content type scan across 4 queries | Layer 2 | Yes |
| This market is winnable | All of the above | Layer 3 | Yes — if all three prior claims hold |
If a claim has no evidence source, it is a belief. Either find the evidence or remove the claim. If the evidence exists but does not meet the three criteria, it is not publishable. Either strengthen it or flag the claim as a judgment — and label it accordingly.
This is what the HTML comments at the bottom of every Search Asset Lab article are doing. Every claim that is not self-evident from the surrounding narrative is tagged: Evidence: for sourced data, Judgment: for editorial interpretation. The tagging discipline forces you to confront your own evidence gaps before the reader does.
The worked example: walking pad evidence map.
The second article on this site used a walking pad category as its worked example. Before the article was written, the evidence map looked like this:
| Claim | Evidence | Layer | Publishable? |
|---|---|---|---|
| Walking pad has purchase intent on Google | Semrush: CPC $1.62, 2 active ad creatives | Layer 2 | Yes |
| Amazon ABA can observe buyer proportion directly | ABA 36 weeks, walking pad category | Layer 3 | Yes (anonymized) |
| Google tools label it “informational” despite purchase signals | Semrush intent label vs CPC contradiction | Layer 2 | Yes |
| Keyword value = buyer count × order value × margin | Cross-platform inference + ABA data | Layer 4 | Yes — formula is the framework-level asset |
Every claim in the published article maps to a row on this table. Nothing was invented during writing. Nothing was generalized from a single observation. The evidence existed before the prose — and the prose only had to carry what was already proven.
This is the discipline Q3 demands.
What happens when the evidence is not there.
Not every market will survive the evidence map. Some will look promising at Q1 and Q2, then fail at Q3 because you simply do not have the data to make the claims the article needs.
This is not a failure of the framework. It is the framework working. The first article said: “Most search markets should never become websites.” Q3 is the gate where that judgment most often proves true.
When the evidence fails:
– Do not invent data to fill the gap.
– Do not lower your standards to make claims sound supported.
– Do not publish a page that reads like everything else on the SERP.
Record what you have. Flag the gap. Decide whether to invest in collecting the evidence — or pass on the market.
The Search Opportunity Framework is not a formula for always finding a market to build. It is a framework for stopping you from building in the wrong one. Q3 is where the framework earns that reputation.
From evidence to asset.
Q3 is the hinge of the framework. Q1 and Q2 tell you whether there is a market you can reach. Q3 tells you whether you can do it in a way that lasts.
Demand can shift. SERP competition can intensify. Asset models can change. But evidence you own — data you collected, patterns you documented, methods you built — is the one defensible layer that does not erode with algorithm updates.
The fourth question — Q4, the subject of the next article — asks whether the traffic can compound into something you keep. But Q4 only matters if Q3 passes. You cannot compound what you never had.
—
