Semrush Site Audit: A Practical Walkthrough (2026)

A Semrush Site Audit crawls your site the way a search engine would, runs 140+ technical and on-page checks, and — unlike a raw crawler — tells you what to fix first. Technical SEO is where ranking potential quietly leaks away, and this is the tool I reach for whenever a site underperforms for no obvious reason. This walkthrough covers the whole job: configuring the crawl properly, the limits that bite on each plan, how to actually read the report, the thematic reports most people never open, and the new AI-search checks added in 2026.

What the Site Audit actually does

Point it at your domain and Semrush crawls your pages, scores overall site health from 0–100, and sorts every issue into three buckets:

• Errors — the serious problems (broken pages, broken internal links, redirect chains, non-secure pages). Fix these first.
• Warnings — mid-priority issues (missing or duplicate meta descriptions, thin content, slow pages, missing alt text).
• Notices — minor, good-hygiene items worth a pass once errors and warnings are handled.

The genuinely useful part is that each issue comes with a “why it matters” and a “how to fix it” — so it’s a to-do list, not a wall of jargon.

Setting up the crawl (the step most people rush)

A bad configuration produces a scary, useless report. Before you hit go:

1. Create a Project for the domain — Site Audit lives inside Projects, not the standalone search tools.
2. Pick your crawl source. Website (follows links from your homepage), Sitemaps (crawls what you’ve declared), or URLs from file (a list you upload). If you want to know what Google can actually discover, crawl the website. If you want to check your declared inventory, crawl the sitemap. They surface different problems — orphan pages only show up in the gap between them.
3. Set the page limit for this crawl, within your plan’s cap (below).
4. Choose the crawler: mobile or desktop. Google indexes mobile-first, so unless you have a specific reason, crawl as mobile — it’s what actually gets indexed.
5. Handle the awkward cases — allow/disallow specific URL paths, remove URL parameters that create infinite crawl traps (session IDs, filters), set a crawl delay if your host is fragile, or bypass restrictions for a staging site.
6. Schedule it. Daily or weekly automatic re-crawls, with an email when it finishes. Set this once and the trend line builds itself.

The limitation that catches people out: JavaScript rendering is only available from the middle tier up — that is Pro+ and Advanced today, and it was Guru and Business before Semrush retired those names in May 2026. If your site is React, Next.js, Vue or any JS-heavy framework and you’re on the entry plan, the crawler may see far less content than Google does — and your audit will be misleading rather than merely incomplete. That single constraint is often the honest reason to move up a tier; I break the rest down in Semrush Pro vs Guru.

Crawl limits by plan

The limits matter more than the feature list, because they decide whether the tool can even see your whole site.

Semrush Site Audit crawl limits by plan, 2026

Plan	Pages per audit	JS rendering	Concurrent audits
Free account	100	No	1
SEO $139 and Starter $199 (was Pro)	20,000	No	2
Top pickPro+ $299 (was Guru)	20,000	Yes	2	See Semrush plans→
Advanced $549 (was Business)	100,000	Yes	5

The free limit, plainly: a free Semrush account crawls 100 pages per audit, with one audit running at a time and no JavaScript rendering. That is enough to audit a small brochure site or to sanity-check a section of a big one, and not enough for anything else — 100 pages is the point at which the tool stops being free rather than a soft warning.

Monthly crawl budgets scale on top of the per-audit cap — roughly 100,000 pages a month on the entry tier, 300,000 in the middle and a million at the top — so a big site re-crawled often can hit the monthly ceiling before the per-audit one.

A wrinkle worth knowing: Semrush retired the Pro / Guru / Business names in May 2026, but its own knowledge-base articles on crawl limits are still written against the old names. The crawl numbers map across unchanged, which is why the table above lists both. If you see a support page quoting “Guru,” it means what is now sold as Pro+. Note that SEO ($139) and Starter ($199) are separate tiers that share the same crawl allowance; Starter’s extra $60/mo buys AI-visibility volume, not more crawling. The crawl figures here were last confirmed against Semrush’s documentation, not re-read during the September 2026 pricing check.

Site Audit is not sold separately, so there is no Site Audit coupon as such — the only discount that moves the price is annual billing on the base plan. I explain why the “17% off” codes everyone advertises are just that in the Semrush coupon breakdown.

Run a Site Audit in Semrush→

How the Site Health score really works

This is the most misunderstood number in the tool, and getting it right changes how you use it.

Site Health is not a count of your problems — it’s a measure of problem density. Semrush weighs the issues it found against the number of checks it performed, with errors weighted more heavily than warnings. That has one very practical consequence:

A 50,000-page site with 2,000 issues can score higher than a 20-page site with 15 issues.

So don’t compare your score to anyone else’s, and don’t panic at a low first number on a large site. What matters is your own score’s trend over time. Semrush doesn’t publish the exact weighting coefficients, and you should be suspicious of any blog that claims to know the formula — I don’t, and I’m not going to invent one.

Practical rule: never chase 100/100. Some notices carry almost no ranking impact, and the last few points cost more effort than they return.

Re-running, pausing and deleting a Site Audit

This is the question the documentation answers worst, and it comes up constantly: how do I delete a Site Audit? The reason it’s hard to find is that an audit is not a standalone object you can delete. Site Audit is a campaign that lives inside a Project, so there is no “delete this audit” button next to the report — you are looking for a control that does not exist at that level.

What you actually want is one of three different things:

• Stop it re-crawling. The scheduled re-crawl is a setting on the audit’s own configuration, not a property of the Project. Turn the schedule off and the existing report stays where it is; it simply stops refreshing. This is what most people mean, and it is the reversible option.
• Free up a crawl slot. Your plan caps how many audits run concurrently (see the table above). If you have hit that cap, you don’t need to delete anything — reconfigure an existing audit to point at the new domain and the slot is reused.
• Remove it entirely. Because the audit belongs to the Project, removing it for good means removing the Site Audit campaign from that Project — or deleting the Project itself, which takes every other tool configured on it (Position Tracking, On-Page SEO Checker) with it. Check what else is attached before you do that; deleting a Project to tidy up one audit is a mistake people only make once.

Re-running on demand is separate from all three: an audit can always be re-crawled immediately without touching the schedule, and doing so after a fix is the fastest way to confirm the fix actually landed.

How to read the report without panicking

A first audit almost always looks alarming. Work in priority order:

1. Clear the errors. Broken links, redirect chains and crawlability problems block rankings directly. Fastest wins live here.
2. Triage warnings by impact. Duplicate titles and thin content affect rankings; a missing alt tag on a footer icon does not. Fix the ones touching pages that actually rank.
3. Batch the notices. One housekeeping pass; don’t let them derail real work.
4. Re-crawl and watch the trend. Steady improvement beats a perfect snapshot.

The thematic reports (where the real value hides)

Most people read the issue list and stop. The thematic reports are where a site audit stops being a checklist and starts being a diagnosis:

• Crawlability — how much of your site is actually reachable, plus crawl-depth problems. If important pages sit five clicks deep, this is where you find out.
• Internal Linking — orphan pages, pages with too few incoming links, broken internal links, and link-depth distribution. Genuinely the most underrated report in the tool.
• Site Performance — load times and page-size problems, with Core Web Vitals broken out.
• HTTPS — mixed content, certificate issues, insecure resources.
• International SEO — hreflang conflicts and missing return tags, if you run language variants. (The specific error is a common one — I wrote a plain-English fix for hreflang ‘no return tags’.) If this report reads “not implemented,” that is not an error: it means the crawler found no hreflang annotations on the site at all. On a single-language site that is the correct state and you should ignore it. It is only a problem if you do publish language or country variants — in which case “not implemented” means Google is being told nothing about how they relate, and the fix is to add the annotations, not to change a setting in Semrush.
• Markup — structured data problems, including the unparsable structured data errors that quietly kill your rich results.

What’s new in 2026: the AI-search checks

This is the part that makes the tool worth re-visiting even if you’ve used it for years. Semrush added AI-oriented checks to Site Audit:

• AI Search Health — scores how well your pages are set up for AI crawlers and answer engines.
• Blocked from AI Search — flags pages your robots.txt blocks to AI user-agents specifically. This catches a genuinely common accident: a blanket crawler rule that also shuts out the AI engines you actually want citing you.

That second one is worth running today even if you do nothing else. If you’re unsure which user-agents you’re blocking, cross-reference the AI crawler user-agent reference and the guide to blocking or allowing AI crawlers. For the wider picture, see technical SEO in the AI era.

Worth noting for context: Adobe completed its acquisition of Semrush in April 2026, and the platform has been reorganising around AI-visibility features since. Nothing about the Site Audit workflow changed as a result, but expect the AI-oriented reporting to keep expanding.

Turn the audit into rankings

An audit only pays off if you act on it. My routine: run it monthly (scheduled, so it happens whether I remember or not), fix every new error, keep the health score trending up, and pair it with Organic Research so you know which pages are worth the technical effort — fixing a page that ranks and earns beats polishing one nobody visits.

How it compares: Screaming Frog gives you more raw crawl control and a one-off licence fee, and it’s what I’d still use for a forensic deep-dive on a huge site. Site Audit’s advantage is prioritisation, scheduling, the trend line and plain-English fixes — it tells you what to do, not just what exists.

New to Semrush? Start with how to use Semrush for free — the free account includes a 100-page audit, which is enough to try this on a small site before paying.

See Semrush plans→

Frequently asked questions

What does the Semrush Site Audit check?

It runs 140+ technical and on-page checks — broken links, redirect chains, crawlability, HTTPS problems, duplicate or missing metadata, thin content, page speed, Core Web Vitals, hreflang and structured data — and scores overall site health from 0 to 100. In 2026 it also includes AI-search checks that flag pages blocked from AI crawlers.

How is the Semrush Site Health score calculated?

It measures the density of problems, not the raw count — the issues found weighed against the number of checks performed, with errors weighted more heavily than warnings. That is why a large site with many issues can score higher than a small site with few. Semrush does not publish the exact weighting, so track your own trend rather than comparing scores between sites.

How many pages can Semrush crawl per audit?

A free account is capped at 100 pages per audit. The SEO, Starter and Pro+ tiers allow up to 20,000 pages per audit and Advanced up to 100,000, with monthly crawl budgets scaling on top of that (about 100,000, 300,000 and one million pages respectively). These are the same limits the old Pro, Guru and Business plans carried before the May 2026 rename.

Does Semrush Site Audit render JavaScript?

Only from the middle tier up — Pro+ and Advanced today, formerly Guru and Business. On the entry plan, a JavaScript-heavy site (React, Next.js, Vue) may be crawled with far less content visible than Google actually sees, which makes the audit misleading. For JS-driven sites, JS rendering is often the single strongest reason to move up a tier.

Is a perfect 100 site health score necessary?

No. Some flagged notices have little ranking impact, and the final few points usually cost more effort than they return. Aim for all errors cleared and a steadily improving score rather than a perfect number.

How often should I run a site audit?

Monthly suits most sites, plus a fresh crawl after any big change such as a migration, redesign or bulk content update. Schedule it to run weekly or daily so the trend line builds automatically and regressions surface before they cost you traffic.

Is Semrush Site Audit free?

There is a free tier: a Semrush free account includes one project with a Site Audit limited to 100 pages, which is enough to test a small site. Beyond that you need a paid plan for meaningful crawl limits, scheduling and JavaScript rendering.