Loading tutorials…
Loading tutorials…
Site Audit returns 3,000 issues. You have time to fix 30. Picking the wrong 30 wastes a quarter; picking the right 30 lifts rankings within 60 days. This is the priority order specialists use.
Who this is forOwners or in-house SEOs who ran SEMrush Site Audit, saw the issue count, and froze. The tool gives you data; this tutorial gives you the priority order. Most teams can ship the top 30 fixes in 2-3 sprints if they know which 30 to ship.
What you'll need
Step 1
Site Audit → Issues tab → filter "Errors" only. Warnings and Notices are 90% noise on most sites.
Open Site Audit → Issues tab. By default SEMrush shows Errors + Warnings + Notices together (often 1,500-3,000 issues).
Filter to 'Errors' only. The count drops to 50-200 on most sites.
Errors are issues Google treats as crawl/index blockers or hard ranking penalties: 5xx server errors, broken canonicals, broken internal links to important pages, missing title tags on indexable pages.
Warnings are 'best practice' issues: missing alt text, no H1, low text-to-HTML ratio. Some matter; most don't. Defer until after Errors are clean.
Notices are minor: missing meta description, mixed-case URLs, long URLs. These almost never affect rankings.
Mental model: Errors first (1-2 sprints), Warnings second (1 sprint), Notices last (or never).
Step 2
Inside Errors tab, sort by URLs affected. A bug affecting 800 URLs is 80x the leverage of 10 individual 4xx pages.
In the filtered Errors view, sort by 'URLs affected' descending.
The top 5-10 errors by URL count almost always represent your highest-leverage fixes. A 'broken internal link template' affecting 800 pages is one fix that resolves 800 URLs.
Compare to the bottom of the list: '12 pages with 4xx response.' Those 12 pages might be old URLs that should 410 anyway.
Always prefer template-level fixes over URL-level fixes. One change to a layout, sitemap, or robots.txt often resolves hundreds of issues.
Export the top 30 errors as CSV. This is your work list.
Step 3
The 8 errors that actually move rankings: 5xx errors, broken canonicals, noindex on indexable pages, broken internal links to top pages, slow LCP, duplicate titles on money pages, redirect chains, broken hreflang.
These 8 categories are the ones that materially affect crawl, index, and rank. Tackle in this order:
1. 5xx server errors on indexable URLs — Google deprioritizes crawl of 5xx-prone domains. Fix server config, deploy stability, or rate-limiting.
2. Broken canonical tags (self-referencing canonicals pointing to 404, or canonicals pointing across hosts). This breaks index consolidation.
3. Noindex tags on pages that should be indexable (especially money pages, blog posts). Common deploy-config bug.
4. Broken internal links to your top 50 traffic pages. Internal link equity is real; broken links to top pages bleed it.
5. LCP > 4 seconds on top pages (Site Audit → Performance tab). Core Web Vitals are a ranking signal post-Mar 2024.
6. Duplicate title tags on money pages — confuses Google about which page to rank. Especially common on paginated lists and faceted nav.
7. Redirect chains 3+ hops (Site Audit → Issues → Redirect chains). Each hop loses ~10% link equity.
8. Broken hreflang tags (for international sites). Misconfigured hreflang breaks geo-targeting entirely.
If you have 3-7 of these in your top 10 by URL count, that's the entire first-sprint scope.
Step 4
GSC → Pages → Indexed + Not indexed. If SEMrush says 500 errors but GSC indexed count is fine, the errors may not be impacting real Google behavior.
Open Google Search Console → Pages. Note the 'Indexed' count and 'Not indexed' count.
Compare against SEMrush Site Audit crawled URLs. If SEMrush is crawling 8,000 URLs and GSC has 6,500 indexed, the 1,500 'extra' URLs SEMrush is crawling may be parameter URLs Google ignores anyway.
For each high-URL-count error in SEMrush, sample 5 URLs and check in GSC URL Inspection. If GSC says 'Indexed, submitted, served,' the error may not actually affect rankings.
Conversely, GSC errors that SEMrush doesn't surface are real — GSC reflects Google's actual experience, SEMrush reflects what its crawler sees.
When the two disagree, GSC wins. Always cross-validate before spending dev time on a SEMrush error.
Step 5
Site Audit → Thematic Reports → Crawlability. Surfaces robots.txt blocks, orphan pages, broken sitemap entries.
Open Site Audit → Thematic Reports → Crawlability.
This report surfaces structural issues that hide in the issue list: robots.txt blocking indexable pages, orphan pages (no internal links pointing to them), sitemap entries returning 4xx, redirect loops.
Orphan pages: pages SEMrush found in your sitemap or via GSC but couldn't find any internal links to. Often old blog posts or category pages. Either link to them properly or remove from sitemap.
Robots.txt blocks on indexable pages: classic deploy bug — staging robots.txt promoted to production blocks /blog/ or /products/. Fix immediately.
Sitemap entries returning 4xx: tell Google the page exists, then 404 the URL. Bad signal. Either fix the URL or remove from sitemap.
Step 6
Thematic Reports → HTTPS + Internal Linking. Mixed content and broken internal-link patterns are common ranking drags.
Thematic Reports → HTTPS. Surfaces mixed-content issues (HTTP assets on HTTPS pages), broken cert config, HTTPS pages with HTTP canonicals.
Mixed content is sometimes blocked by browsers entirely, causing real conversion drops. Fix all HTTP asset references to be protocol-relative or HTTPS.
Thematic Reports → Internal Linking. Surfaces: pages with 0 incoming internal links, pages with too many outgoing internal links (200+), and broken internal-link patterns.
The 'Internal Link Distribution' chart shows your link-equity flow. Money pages should have the highest internal-link counts. If your About page has 500 internal links and your /pricing page has 12, that's misallocated link equity.
Set up redirects for any URLs that should be consolidated, then audit the internal links pointing to the old URLs and update them. Internal redirects waste crawl budget.
Step 7
After each fix sprint, wait 7-14 days, then re-run Site Audit. Measure: Errors count delta, URLs affected delta, Site Health Score delta.
After your fix sprint ships, wait 7-14 days for CDN caches and Google crawl-discovery to catch up before re-crawling.
Re-run Site Audit. Compare to the prior crawl: Errors count delta, URLs affected delta, and Site Health Score delta.
Capture before/after in a shared doc: 'Sprint 1 — fixed 800 broken internal links across template. Errors dropped from 1,847 to 1,032. Site Health: 72 → 81.'
If a fix didn't move the count, dig in. Sometimes a fix shipped to staging but didn't deploy to prod. Sometimes the fix surfaced a new related error.
Track ranking movement alongside in Position Tracking. Critical fixes (5xx, noindex, broken canonicals) often lift rankings within 14-30 days. Cosmetic fixes (alt text, H1) often don't move anything.
Common mistakes
Chasing Site Health Score
What goes wrong: You spend 40 hours fixing low-impact warnings (alt text, H2 count) and lift Site Health 8 points. Rankings don't move. You concluded SEO doesn't work. Real cost: ~$3,200-4,800 of dev time on non-ranking-affecting fixes.
How to avoid: Use Site Health as a trend indicator only. The real metric is critical Errors affecting indexable URLs. Tackle the 8 critical error categories first; cosmetic warnings go to the bottom of the backlog or get deprioritized entirely.
Treating every error as equal priority
What goes wrong: Engineering treats the 'fix' list as a checklist. They fix 200 broken redirect chains on archived blog posts before fixing 5xx errors on /pricing. ~$2,000-4,000 of dev time spent on near-zero-impact fixes while real ranking issues bleed traffic.
How to avoid: Sort by URLs affected + page importance. A 5xx on /pricing is 100x the priority of a redirect chain on /blog/old-archived-post. Tag your fix list with priority levels before assigning.
Not cross-validating against GSC
What goes wrong: You spend a sprint fixing 800 'duplicate title' errors SEMrush flagged. GSC URL Inspection on those URLs shows 'Indexed, submitted, served.' Google never cared. ~$1,500-3,000 of dev time on a non-issue.
How to avoid: Sample 5 URLs from any high-URL-count error. Run GSC URL Inspection on each. If GSC says 'Indexed, submitted, served' for all 5, deprioritize that error category.
Fixing without measuring
What goes wrong: Sprint after sprint of fixes ship. Nobody tracks before/after. Six months later, nobody knows which fixes worked. Decisions get made on gut. Wrong lessons get extrapolated to the next site.
How to avoid: Always document before/after for every sprint: Errors count, URLs affected, Site Health, and rankings delta on a sample of affected URLs. Build a 'fix → impact' table over time.
Re-crawling immediately after fixes ship
What goes wrong: Fix ships Monday. You re-crawl Tuesday. SEMrush still shows the old errors because the crawl was already mid-flight or CDN caches haven't expired. You think the fix didn't work and revert it.
How to avoid: Wait 7-14 days minimum between fix-ship and re-crawl. CDN caches, sitemap re-fetch, and SEMrush crawl-discovery all take time. Patience.
Ignoring orphan pages and sitemap mismatches
What goes wrong: Your sitemap declares 2,400 pages. SEMrush crawls 2,400 URLs but finds 180 of them have zero internal links. Google de-prioritizes orphan pages for indexing. The pages exist but never rank.
How to avoid: Run the Crawlability thematic report monthly. For orphan pages: either link to them from a hub page, or remove from sitemap. Decide deliberately.
Recap
Done — what's next
How to set up a SEMrush project the right way
Read the next tutorial
Hand it off
Running Site Audit once is a project. Running it weekly, triaging issues, shipping fixes, and validating in GSC + Position Tracking is a job. A vetted technical SEO specialist owns the audit-fix cycle — typically $400-1,000/mo at $14-16/hr depending on site complexity.
See specialist rates
It's directional. Better signal: how many critical Errors (5xx, broken canonicals, noindex bugs) do you have affecting indexable URLs? A site with Health 88 and 0 critical Errors is in great shape. A site with Health 88 and 50 critical Errors has 50 problems to ship.
No. Fix Errors first, always. Warnings are 'nice to have' best practices. Most warnings don't affect rankings on a well-built site. Defer them to the second sprint or skip them entirely.
Filter to Errors only (drops to ~50-200). Sort by URLs affected descending. Top 10 errors = your first sprint scope. Almost always one or two of those 10 errors will affect 500-2,000 URLs as a template-level bug — that's where 80% of your impact will come from.
Screaming Frog is a one-off desktop crawler — you run it, get a snapshot, walk away. SEMrush Site Audit is a scheduled cloud crawler that runs weekly and tracks trends over time. Specialists usually use both: Screaming Frog for deep one-off analysis, SEMrush for ongoing monitoring.
Three possibilities: (1) you're fixing the wrong issues (warnings, not critical errors); (2) you fixed real issues but haven't waited long enough (Google takes 14-60 days to re-process at scale); (3) site issues weren't the bottleneck — your rankings are blocked by backlink profile or content quality. Diagnose all three before concluding the fixes didn't work.
Weekly for active sites with frequent deploys. Monthly for static sites. Daily only if you're shipping site changes daily and have credits to burn. Most teams over-crawl and burn credits without acting on the data.
SEMrush
A SEMrush Project is the container that pipes data into Site Audit, Position Tracking, On-Page SEO, Listing Management, and Social Tracker. Configure it wrong on day one and every downstream module produces noise for months. This is the right setup.
SEMrush
On-Page SEO Checker returns 300 'ideas' across six categories. Most users implement the first 5 in the list. The right workflow triages all 300 down to 30 high-impact actions in about 90 minutes. This is the playbook.
SEMrush
You're paying $139-$500/mo for SEMrush. The question isn't whether the tool is worth it — it's whether you have the time and skill to actually use what you're paying for. This is the honest framework.