An audit is thirty questions, not a scan.

The data & method

The best SEO audit is not a tool that scores a page red or green. It is a fixed set of questions, asked in the same order every time, of the one dataset that records what a search engine actually did: Search Console. Get the questions right and the answers write the strategy for you.

The answers here come from sixteen months of Search Console performance exports across five live accounts — the same five in the field study — spanning e-commerce, local trades and B2B software. Every figure is measured from the raw query, page, device and search-appearance tables. Identities are NDA-protected, so each account is labelled by sector only, exactly as in the study.

One honest constraint shapes what can be claimed. Google caps the query and page exports at 1,000 rows each. So counts of clicking queries — all comfortably under a thousand — are true and uncapped; anything measured over the impression set is measured over each site’s top ~1,000 queries, and is described that way. The full limitations are listed at the end, not buried.

The query network

How wide, how commercial and how distributed is the demand the site actually captures. Breadth of the query network is the first and hardest-to-fake signal of authority.

1. 01

   How wide is the query network — how many queries earn clicks, not just impressions?

   The clearest breadth signal, and unfakeable: Auto locksmith earns clicks on 798 distinct queries, Signage e-commerce 536, Glazing repair 196, Compliance SaaS 80, Prestige automotive 67. A wide clicking network is coverage doing its job; a narrow one is a few pages carrying the site.

2. 02

   What share of clicks is non-brand demand versus people typing the brand name?

   Signage e-commerce is 93% non-brand — almost pure demand capture. Auto locksmith 76%, Prestige automotive 40%, Glazing repair 28%, Compliance SaaS just 5%. Non-brand is the traffic authority earns and the traffic that grows a business; brand clicks were coming anyway.

3. 03

   Is authority distributed across the network, or concentrated in a handful of head terms?

   The top ten queries carry 44% of clicks at Auto locksmith and 53% at Signage e-commerce (distributed = real coverage), rising to 96% at Compliance SaaS — where “the network” is essentially one query, the brand. Distribution is the difference between a covered subject and a lucky page.

4. 04

   Does the demand the site captures overlap what the business actually sells?

   At the demand-captured accounts the non-brand clicks are category and buyer-intent terms — the things that convert. At Compliance SaaS the non-brand demand exists only as impressions: shown for the category, chosen for the name. Coverage that does not capture commercial demand is a library, not a funnel.

Where the coverage sits

Coverage is worth what its position is worth. Where in the results does the mapped subject actually rank, and what is one push from breaking through?

1. 05

   What share of the query set ranks top-3, top-10, or beyond page two?

   Across each site’s top ~1,000 queries: Prestige automotive is strongest (20% top-3, 50% top-10); Auto locksmith 18% / 44%; Signage 9% / 55%. Glazing repair and Compliance SaaS still sit mostly beyond page two (84–85% past position ten) — coverage in place, position still climbing.

2. 06

   How many “striking-distance” queries sit one push from page one?

   Queries at positions ~8–20 with real impression volume — the near-authority pipeline: Signage e-commerce has 78, Auto locksmith 31, Compliance SaaS 29, Glazing repair 8, Prestige automotive 2. That is the fastest ROI on the whole audit: not new coverage, a nudge to coverage that already half-ranks.

3. 07

   Is average position improving over the sixteen months, or parked?

   Every account measured is climbing: Glazing repair 46→24, Auto locksmith 38→19, Compliance SaaS 36→24, Prestige automotive from page two to the top of page one. Position is the leading indicator — the engine ranks you better before anyone clicks — and here it leads in the right direction across the board.

Earning the click

Ranking is being shown; the click is being chosen. At the same position, does the site earn the click — and where is it shown for things it never wins?

1. 08

   At the same rank, does the site earn its clicks — is CTR at or above what the position should yield?

   Top-3 click-through ranges from 26.8% at Glazing repair (snippet and intent pulling hard) to just 2.61% at Signage e-commerce, whose head terms are the most competitive and the most compressed by AI answers. Same rank, wildly different pull: visibility is not satisfaction.

2. 09

   Where are impressions high but click-through near zero — shown, but never chosen?

   Compliance SaaS at positions 11–20 converts 0.04% of impressions to clicks — effectively shown and ignored. That is coverage the engine is testing but users are not selecting: a signal to sharpen intent and snippet, not to publish more.

3. 10

   Do the SERP features the site owns actually earn the click?

   Signage e-commerce wins both Product snippets and Merchant listings — but Merchant listings convert at 26.6% CTR against the rich snippet’s 1.1%. The high-intent shopping feature earns the click; the decorative one mostly does not. Owning a feature is not the same as being chosen through it.

The page architecture

Authority spread across a mapped network is durable; authority balanced on one page is a single point of failure. How is the click-earning load distributed?

1. 11

   Is authority carried by one page, or spread across many?

   The single top page accounts for just 19% of clicks at Prestige automotive — healthy distribution across the map — versus 62% at Glazing repair and 66% at Compliance SaaS, where one URL is doing most of the work. Concentration is fragility: one algorithm shift on one page and the site moves.

2. 12

   How efficient is the coverage — how many pages earn clicks versus merely get impressions?

   Compliance SaaS has 140 pages earning clicks and 860 shown-but-ignored; Auto locksmith is lean at 68 earning against 39 ignored. A large ignored set is not always waste — it can be the outer section still ripening — but it is retrieval spent without return, and worth watching.

3. 13

   Are there pages with impressions and zero clicks that signal thin or mismatched coverage?

   Thousands, portfolio-wide — most heavily at Compliance SaaS and Signage e-commerce. Read against the topical map, each is either outer-section coverage on its way up or a page answering the wrong intent. The audit’s job is to tell those two apart, page by page.

Intent & source context

The same query means different things in different hands. Device, geography and query shape tell you the search context — and whether the coverage matches it.

1. 14

   What does the device split reveal about the search context?

   Auto locksmith is 78% mobile (locked out, on the move) and Prestige automotive 77% mobile — urgent, in-the-field intent. Compliance SaaS is 72% desktop: a considered B2B search done at a desk. The channel mix confirms the intent the content has to serve.

2. 15

   Is the demand local, national or broad?

   Signage e-commerce is 93% UK — a national store — while the trades concentrate in their service areas. Geography tells you whether the topical map needs place-nodes (town-level service pages) or category depth. Get that wrong and you build the right coverage for the wrong map.

3. 16

   What is the dominant query intent, and does the coverage match it?

   Transactional at Signage e-commerce and Auto locksmith, local-service at Prestige automotive and Glazing repair, informational-then-commercial at Compliance SaaS. Authority is only useful where the coverage answers the intent the queries actually carry — the map has to be drawn to the demand, not to the keyword list.

Momentum & historical data

Authority is coverage plus track record. Is the network growing, is visibility converting to clicks, and did the coverage survive the updates that reset thin tactics?

1. 17

   Is the query network expanding over time, or static and decaying?

   The single question this export cannot answer directly — GSC does not give query-by-month in a bulk pull — and the honest next data request. What the sixteen-month chart does show is total clicks and impressions climbing at four of five accounts, which only happens when new queries keep entering the network.

2. 18

   Is visibility converting to clicks, or decoupling from them?

   Only one account decouples cleanly: Compliance SaaS, impressions up ~230% while clicks rose ~18% — the AI-era “great decoupling”. Glazing repair does the opposite, converting rising visibility into a rising click-through. Same era, opposite trajectories, and the difference is authority.

3. 19

   Is there a seasonal cycle the content calendar should anticipate?

   The monthly series shows demand cycles — clearest in the seasonal trades — that a reactive calendar always publishes too late for. The audit turns that from a surprise into a schedule: cover the seasonal outer-section before the season, not during it.

4. 20

   Did the coverage hold through the 2025–2026 core updates, or reset?

   It held and climbed. None of the five reset; the accounts built on comprehensive coverage rode the updates that were designed to reward exactly that. That is the whole thesis of authority in one observation — thin rankings evaporate on update day; covered ones gain.

Cost of retrieval

Authority the engine cannot cheaply crawl and read does not count. How much indexed coverage is paying its way, and how much is retrieval spent for nothing?

1. 21

   How much indexed coverage never pays back the cost of crawling it?

   Compliance SaaS carries 860 pages that earn impressions but no clicks — the sharpest retrieval overhang in the set. Every one is crawled, rendered and indexed; a large fraction returns nothing. Lowering that ratio — prune, consolidate or fix intent — is cost-of-retrieval work that frees budget for pages that rank.

2. 22

   Is the click-earning page set concentrated enough to crawl efficiently?

   Auto locksmith is efficient — a tight click-earning set with little dead weight — while Signage e-commerce (338 earning, 662 ignored) spreads crawl across a much larger surface. A big surface is fine if it is a real map ripening; it is a problem if it is duplication the engine keeps re-reading for nothing.

The strategic reads

The point of thirty questions is one paragraph of verdict per account — what the data says to do next, and what it says about the method itself.

1. 23

   Which account shows the strongest topical-authority signal?

   Auto locksmith: 798 clicking queries, 76% non-brand, distributed across the map (top ten only 44%), position climbing 38→19. Wide, commercial, distributed and improving — every fingerprint of authority at once. Its lead volume tells the same story from the business side.

2. 24

   Which is most exposed to the AI decoupling?

   Compliance SaaS: 5% non-brand clicks, 860 zero-click pages, click-through collapsing to 0.04% deep in the results. It is shown for its category and chosen for its name. The fix is not more pages — it is position and intent on the coverage already indexed.

3. 25

   Where is the fastest return available?

   Signage e-commerce’s 78 striking-distance queries — coverage already at positions 8–20 with real demand behind it. Nudging half-ranking pages onto page one returns faster than any new content, because the authority is already built; only the placement is missing.

4. 26

   Whose growth is brand-dependent versus demand-captured?

   Glazing repair (28% non-brand) and Prestige automotive (40%) lean on their names; Signage e-commerce (93%) and Auto locksmith (76%) are captured demand. Brand-led growth is real but capped by brand awareness; demand-captured growth scales with coverage. The audit says which lever each account has left to pull.

5. 27

   Where is the problem “shown but not chosen” rather than “not shown”?

   Signage e-commerce’s compressed head terms and Compliance SaaS’s deep-ranking category pages. Both are visibility without selection — a snippet, title and intent problem, not a coverage problem. Publishing more there would be answering a question nobody asked.

6. 28

   For each account, is position the bottleneck or is CTR?

   Position is the bottleneck at Glazing repair and Compliance SaaS (the coverage is there, the rank is not). CTR is the bottleneck at Signage e-commerce’s head terms (the rank is there, the click is not). Naming which one, per account, is what turns an audit into a plan.

7. 29

   Which account should shift from building coverage to converting it?

   Prestige automotive: position won (top of page one, 50% of queries in the top ten), so the next unit of work is the click and the conversion, not another outer-section page. Knowing when to stop building is as valuable as knowing what to build.

8. 30

   Does the method predict the outcome?

   Yes — and that is the whole point. Broad, distributed, non-brand, position-climbing coverage lands on exactly the accounts that produced the revenue and leads in the field study. The signal the audit reads and the business result it predicts are the same accounts. The questions work because the method does.

Run this on your own Search Console

None of these numbers need trusting on faith. Every one comes from a plain CSV export and a few lines of pandas — so you can run the same audit on your own site this afternoon. First, the exact definitions behind each figure, so nothing rides on a fuzzy term:

Clicking query

A query with at least one click in the window. Breadth is counted this way because clicking-query totals sit under Google’s 1,000-row export cap, so the count is true rather than truncated.

Non-brand

Any query whose text does not contain a brand token, matched case-insensitively. A close approximation — a few branded long-tail queries fall either side of the line.

Striking distance

Average position between 8 and 20 with at least 500 impressions: real demand, one page from page one. The fastest-return segment in the audit.

CTR by position

Clicks ÷ impressions within a position band (top-3 = positions 1–3). Reads click satisfaction independent of how many queries sit in the band.

Top-page concentration

Share of all clicks earned by the single highest-clicking URL. Lower is healthier — authority spread across the map, not balanced on one page.

Then the recipe. Export Performance → Search results → Export → CSV from Search Console (you get Queries.csv and Pages.csv), drop in your brand tokens, and run:

# pip install pandas
import pandas as pd

q = pd.read_csv("Queries.csv")           # Top queries, Clicks, Impressions, CTR, Position
BRAND = ["your brand", "yourbrand"]      # lowercase brand tokens
brand = q["Top queries"].str.lower().str.contains("|".join(BRAND))

# 1 · Query-network breadth — queries that earn clicks (uncapped; all under 1,000)
clicking = q["Clicks"].gt(0).sum()

# 2 · Non-brand share of clicks — the demand authority actually captures
nonbrand = q.loc[~brand, "Clicks"].sum() / q["Clicks"].sum()

# 3 · Striking distance — real demand, one push from page one
striking = (q["Position"].between(8, 20) & q["Impressions"].ge(500)).sum()

# 4 · Click satisfaction — CTR in the top three positions
top3 = q[q["Position"].between(1, 3)]
ctr_top3 = top3["Clicks"].sum() / top3["Impressions"].sum()

# 5 · Authority concentration — clicks on the single strongest page
p = pd.read_csv("Pages.csv")
concentration = p["Clicks"].max() / p["Clicks"].sum()

print("clicking queries :", clicking)
print("non-brand share  :", round(nonbrand * 100, 1), "%")
print("striking-distance:", striking)
print("CTR top-3        :", round(ctr_top3 * 100, 2), "%")
print("top-page share   :", round(concentration * 100), "%")

Limitations

The honest boundaries on what this data can and cannot claim. An audit that hides its limitations is a sales deck.

1. Five accounts, one operator, UK-centric. This is field evidence from a real portfolio, not a controlled study or a representative sample. It shows what the questions surface, not a population average.
2. Google caps the Search Console query and page exports at 1,000 rows each. Clicking-query counts (all well under 1,000) are therefore true and uncapped; anything measured over the impression set is measured over each site’s top ~1,000 queries and is described that way throughout.
3. Brand versus non-brand is classified by matching the brand name in the query. It is a good approximation, not a perfect one — a handful of branded long-tail queries can fall on either side of the line.
4. Position and CTR are averages over the window, which hide within-period movement. The sixteen-month trend lines in the field study are the honest view of change over time; the tables here are the snapshot the audit reads.
5. Correlation, not proof of causation. These questions diagnose the shape of a site’s coverage and the relationships in its data; they do not isolate a single cause. The field study is where the causal argument, with before/after windows and significance tests, is made.

Where to go next

This is the diagnostic discipline behind the method. The field study proves the thesis with sixteen months of before/after data; the complete guide teaches how to build the authority these questions measure. And the receipts are public:

• The compounding curve — the field study 16 months, five businesses, real Search Console + GA4 data Read →
• Topical authority — the complete guide What it is, why it beats keywords, and how to build it Read →
• Signage & personalisation e-commerce organic revenue · converts at 2× its traffic weight £11,244
• Prestige-marque automotive specialist retained · 68% of traffic organic · #1 local for the marque 5+ yrs
• Domestic glazing & window repair organic clicks in the recent quarter, YoY · page 5 → page 2 ~5×
• Auto locksmith & vehicle key replacement qualified leads from organic · 77% of new customers 505
• Compliance & quality-management software category head-term visibility built over 16 months Near-zero → p2