The SEO Audit Checklist We Run Before We Quote A Rebuild
This is a real SEO audit checklist, not a lead magnet with forty headings. Forty-two checks, eight categories, one pass test each. Work through it yourself and you will find most of what a paid audit would find — often in a single afternoon.
How to use this without wasting a week
Score each item 0, 1 or 2. Do not fix anything mid-audit. The moment you start fixing, the list stops describing the site and starts describing your afternoon.
Sample, do not crawl everything by hand. Twenty URLs from each template type will reveal the same pattern as five hundred, at one fortieth of the cost.
Write the evidence next to each score: a URL, a screenshot, a log line. An audit you cannot prove is an opinion with a cover page.
The checklist
Each item has one pass test. If you cannot answer yes or no in under two minutes, the item is not measurable on your site and that itself is a finding.
Crawling and indexing
6 checksPass: no disallow rule matches a page you want ranked, and no CSS or JS folder is blocked.
Pass: spot-check ten sitemap URLs — none redirect, none return noindex, none are duplicates.
Pass: every "Excluded" bucket has a written reason and an owner. Unexplained exclusions are the norm, not the exception.
noindex on commercial pagesPass: crawl the top fifty revenue pages and confirm zero accidental directives, including in HTTP headers.
Pass: every internal link resolves with a single 301 to a 200. Chains waste crawl budget and dilute signals.
Pass: filters either canonicalise, are blocked, or add unique value. Thousands of parameter pages are a bug, not a feature.
Technical foundations
6 checksPass: LCP under 2.5s, INP under 200ms, CLS under 0.1 on mobile for each major template type.
Pass: disable JavaScript and load a product page. If the price, specs or body copy vanish, rendering is a risk.
Pass: same headings, same links, same structured data. Hidden mobile blocks are a common parity failure.
Pass: canonical points to the preferred URL, is absolute, and is not contradicted by an hreflang or sitemap entry.
Pass: http, www and non-www all resolve to a single HTTPS version with one redirect, no loops.
Pass: no errors in the rich results test, and no markup describing prices, ratings or FAQs that a user cannot see.
On-page elements
6 checksPass: no duplicate titles across indexable pages, and the target phrase appears near the start.
Pass: 120–155 characters, one clear reason to click. They do not rank you, but they decide whether you get clicked.
Pass: H1 then H2 then H3. Decorative headings styled as H2 for font size are a hierarchy bug with a design disguise.
Pass: descriptive, not keyword-stuffed, and empty only for decoration. Check the top templates, not all 8,000 images.
Pass: a human can guess the topic from the path. No session IDs, no dates that will age badly, no ?p=1841.
Pass: no /search?q= or ?print=1 URLs indexed. They are infinite thin-content generators.
Content quality and intent
5 checksPass: if the top five results are comparison tables and yours is a 2,000-word essay, intent is mismatched regardless of quality.
Pass: open five near-identical pages. If only the city or product name changes, they are thin duplicates with different URLs.
Pass: search your main phrase and confirm one page ranks, or that two rank for clearly different sub-intents.
Pass: a reader cannot find a stale year, discontinued product or expired offer in the top twenty pages.
Pass: named authors, real photos, specific examples, first-hand numbers. "Trusted by thousands" is not evidence.
Internal linking and architecture
5 checksPass: trace the shortest path to each top page. Four or more clicks means the page is buried by structure.
Pass: compare sitemap URLs against internal link sources. Pages with no inbound internal link will not rank on their own.
Pass: no "click here" or "read more" as primary anchors. The anchor should make sense read alone.
Pass: breadcrumb trail reflects the same hierarchy the URL implies, with breadcrumb structured data on top.
Pass: every supporting article links up to its hub and the hub links down to its main children.
Off-page and links
5 checksPass: reading the referring domains, a stranger could guess what your site sells. Ten relevant links beat a hundred generic ones.
Pass: no burst of sitewide footer links from unrelated domains, no paid-placement network pattern.
Pass: a written list of ten domains linking to two competitors but not to you, with a reason each would link.
Pass: mentions found and requested, not just counted. This is the cheapest link source most sites ignore.
Pass: something a journalist or trade editor would reference — original data, a benchmark, a calculator. Without it, outreach is begging.
Conversion and user experience
5 checksPass: a stranger can answer both in five seconds without scrolling. Clever headlines fail this test.
Pass: the main CTA appears above the fold and repeats at natural decision points, using the same wording each time.
Pass: every field has a named owner. Twelve-field forms exist because nobody deleted anything since 2016.
Pass: named case studies with numbers, real certifications, working contact details. Logos without context convince nobody.
Pass: buttons and links are at least 44px, not crowded, and phone numbers are tappable.
Measurement
4 checksPass: a form submit, call click and chat start each fire one consistent, named event.
Pass: owner access, sitemap submitted, no unread coverage warnings older than a month.
Pass: every enquiry carries an original-source field, so closed revenue can be traced back to a landing page.
Pass: sessions, non-brand clicks, enquiries and revenue screenshotted before you touch anything. Without it, you cannot prove improvement.
From scores to a plan, in four steps
A score is not a plan. This is the order we work in once the list is filled, and it is the order that keeps an audit from turning into a three-month backlog nobody finishes.
Fix whatever blocks indexing
Redirects, directives, rendering, sitemaps. Nothing else you do will be measured until these are clean, so they come first even when they feel unglamorous.
Consolidate before you create
Merge near-duplicates, pick one URL per intent, redirect the rest. Fewer stronger pages beat more weaker ones in every crawl budget model we have worked with.
Fix conversion on traffic you already have
Before chasing new keywords, make the current top twenty landing pages answer the query and offer one obvious next step. This is usually the cheapest win on the list.
Then add content, one cluster at a time
One hub, four to six support pages, internal links wired both ways, and a scheduled review in ninety days. Repeat only after the first cluster earns impressions.
What this list cannot see
Forty-two checks catch structural damage. They do not catch the four problems that cost more money than any broken canonical, because none of them show up in a crawler export.
What Googlebot actually fetches
Server logs show crawl frequency per template, which URLs consume budget and whether Google sees the same page you do. No browser tool reveals this. On large or fast-changing sites, logs are where the real answer lives.
Intent misalignment at the category level
A page can pass every on-page check and still be the wrong format for its query. Finding that requires comparing your page type against what actually ranks, cluster by cluster — judgement, not automation.
Template-level thin content scale
Five near-duplicate pages are a sample finding. Nine thousand of them is a strategy problem: it changes how you consolidate, redirect and rewrite, and it changes the timeline by quarters.
Whether demand exists at all
A technically clean site can rank first for terms nobody converts on. Validating that requires revenue data, competitor ad spend and sales-call language, not a checklist.
If your score comes out above 120 of 160 and organic still is not producing enquiries, the list is not the problem. The diagnosis is deeper — and that is usually the point at which a second pair of eyes is cheaper than another six months of guessing. The fundamentals behind all of this sit in SEO Services
Run it, then send us the scores
We read self-audits for free and tell you which three findings actually matter. If the honest answer is that the site needs consolidation rather than more content, we will say so before quoting anything.
What audits found on real sites
Machinery exporter
4,100 indexed pages, 620 real ones. Consolidation plus 301 mapping cut crawl waste; organic enquiries went from 6 to 19 per month in five months.
EU + US, 5 monthsB2B distributor
Every product page carried the manufacturer's boilerplate. Rewriting 180 pages with real specs and stock data lifted non-brand clicks 2.6x.
UK, 7 monthsSaaS, post-migration
Redirect chains four hops deep after a replatform. Flattening them recovered 31% of lost organic sessions in nine weeks.
US, 9 weeksQuestions people ask after running the list
How many of the 42 should pass?
Above 120 of 160 weighted is healthy. Below 90, fix structure before writing anything new — new content on a broken base will not rank.
How often should it be re-run?
Fully, twice a year. The crawl and index sections, quarterly. After any migration or template change, immediately.
Do I need paid tools?
No. Search Console plus a free crawler covers 34 of the 42. Paid tools speed up backlink and log analysis, but they do not change the verdict.
What is the single highest-value fix?
Usually consolidation: fewer, stronger pages. It is unglamorous, it takes a spreadsheet, and it outperforms most content calendars.
Can you audit without site access?
Yes, from the outside. Access to Search Console and analytics roughly halves the time and materially improves the accuracy of index findings.
Is this the same checklist you use internally?
Yes, with two internal additions: log-file sampling and a manual review of the top twenty competitors' page types. Both need a practitioner.
Who wrote this list
This checklist comes from nine years of hands-on organic search work, seven of them on B2B and ecommerce sites in the US, UK and EU, plus log-file audits on sites up to 400,000 URLs. It is deliberately unglamorous: the items that matter are the ones that stay broken for years because nobody put them on a list.
We use it as an intake document, not a sales asset. If you want the execution sequence that follows an audit — what to fix first, what to leave alone, and how long each stage really takes — the working order is in Google SEO Services.