Before You Open a Single Tool
The technical SEO audits I see most often have the same two problems. They're done once a year, and
they're built around the same checklist from 2022. Both of those things are now actively costing sites
rankings.
The pace of change accelerated significantly after March 2024, when INP replaced FID as a Core Web Vital,
and then again after Google's March 2026 Core Update, which tightened how heavily performance signals
weigh against content quality. There are also entirely new categories to check that didn't exist two
years ago — AI crawler governance, llms.txt, bot-specific rendering decisions.
This checklist covers 22 items across six categories. Some you'll clear in five minutes. A few will
require developer time. I've flagged each one so you know what you're walking into before you start.
⚠
The audit order matters
Work through these categories in sequence. Crawlability and indexation first — there's no point
fixing INP on a page that isn't indexed. Core Web Vitals second. AI crawler governance third. Schema
and structure last. Jumping to performance fixes on a poorly indexed site is the most common audit
mistake I see.
40%
of sites that passed FID don't pass INP — the new interactivity metric
47%
of sites globally pass all Core Web Vitals thresholds in 2026
20%
of server traffic on many sites now comes from AI crawlers
Everything downstream — rankings, AI citations, Core Web Vitals signals — depends on Google being able to
find and process your pages. These four checks take about fifteen minutes and catch the issues that
quietly tank sites for months before anyone notices.
✓
#01
Verify robots.txt isn't accidentally blocking important pages
Fetch yourdomain.com/robots.txt in a browser.
Cross-reference with Google Search Console → Coverage → Excluded → Blocked by robots.txt.
I've audited multiple e-commerce sites where staging-environment robot rules were
accidentally pushed to production and sat there for months. It's more common than it should
be.
Critical
✓
#02
Run a crawl and check for orphaned pages
A page with zero internal links pointing to it is invisible to Googlebot
and AI crawlers alike. Use Screaming Frog or Sitebulb on your top 50 URLs. Any page without
internal links pointing in is either intentionally excluded or a mistake — and most of the
time it's the latter.
Critical
✓
#03
Check canonical tags on paginated and duplicate content
Self-referencing canonicals are fine and recommended. The problem is
when canonicals point to the wrong URL — typically after a site migration or CMS update.
Check your ten most important pages manually with view-source: and
confirm the canonical tag matches the URL you want indexed.
Quick fix
✓
#04
Confirm your sitemap is accurate and submitted
A sitemap listing 404 pages or pages with noindex directives sends conflicting signals. After Google's December
2025 Rendering Update, pages returning non-200 status codes may be excluded from the
rendering queue entirely. Pull your sitemap into a spreadsheet and spot-check 20 URLs for
live status and correct indexation.
Critical
INP is the metric that most teams are still handling incorrectly. It's not a faster version of FID. It
measures something fundamentally different — and fixing it requires a different kind of investigation.
"FID was a sprint. INP is a marathon. You could game FID by optimising the first click. INP catches
every slow interaction across the entire session."
| Metric |
What it measures |
Good |
Needs work |
Poor |
| LCP |
Time for largest visible element to load |
≤ 2.5s |
2.5–4s |
> 4s |
| INP |
Responsiveness of all interactions in session |
≤ 200ms |
200–500ms |
> 500ms |
| CLS |
Unexpected layout shifts during load |
≤ 0.1 |
0.1–0.25 |
> 0.25 |
| VSI NEW 2026 |
Layout stability across full user session |
Low score |
Mid |
High |
| TTFB |
Server response time |
≤ 800ms |
800ms–1.8s |
> 1.8s |
⚡
Why Lighthouse scores will mislead you on INP
INP only exists in real user data from Chrome. Lighthouse simulates a single automated page load — it
cannot measure what happens when a user opens a mobile menu 30 seconds into their session. A healthy
Total Blocking Time (TBT) in Lighthouse does not mean your INP is fine. Use Chrome User Experience
Report (CrUX) data via PageSpeed Insights or Search Console as your primary INP source.
✓
#05
Pull your INP score from CrUX data, not Lighthouse
Go to PageSpeed Insights → "Discover what your real users are
experiencing" section. If INP shows red or yellow on mobile, you have a JavaScript
main-thread problem. Do this for your ten highest-traffic pages. Treat any page with INP
over 200ms as a priority fix — it's a confirmed negative ranking signal.
Critical
✓
#06
Identify heavy JavaScript using DevTools → Performance → Coverage
INP is almost never fixed by compressing images or tweaking server
settings — that's LCP. INP is fixed by auditing what JavaScript runs during user
interactions. Open Chrome DevTools, go to Performance → Coverage, reload the page, and look
for scripts with high unused bytes. Removing 5–10 unnecessary scripts often improves INP
more than any other single optimisation.
Critical
✓
#07
Add fetchpriority="high" to your LCP element
Your LCP element — almost always a hero image or H1 — needs the browser
to know it's the most important resource on the page. Add fetchpriority="high" to that element's HTML tag. It's a one-line
change that consistently moves LCP by 200–400ms on sites I work on. Takes ten minutes to
implement if you know where the element lives.
Quick fix
✓
#08
Audit Google Tag Manager for tag bloat
GTM containers accumulate tags from every campaign, experiment, and
analytics integration that ever ran on the site. Each one adds JavaScript execution time.
I've seen containers with 40+ active tags where half were from campaigns that finished 18
months ago. Export your GTM container, audit every tag, and kill the ones that serve no
current purpose. This is often the fastest INP win available.
Quick fix
✓
#09
Reserve space for images and embeds to fix CLS
Images and iframes without explicit width and height attributes cause
layout shifts as they load. The browser doesn't know how much space to reserve, so content
jumps around. Set explicit dimensions on every image above the fold, and add aspect-ratio CSS for responsive images. CLS is one of the easiest Core
Web Vitals wins — and one of the most neglected.
Quick fix
✓
#10
Check for the new Visual Stability Index (VSI) — 2026 addition
Google's Core Web Vitals 2.0 (introduced early 2026) adds VSI — layout
stability measured across the entire user session, not just initial load. Unlike CLS, VSI
accounts for whether layout shifts were predictable (a dropdown opening on click) vs.
unexpected (an ad pushing content). Check your Search Console Core Web Vitals report for any
new VSI-related flags appearing.
New in 2026
Mobile-first indexing has been Google's default for years, but the rendering question has a new dimension
in 2026. Most AI crawlers don't execute JavaScript. If your critical content loads client-side, it's
invisible to Perplexity, ChatGPT's crawler, and others. That's a GEO problem sitting inside a technical
SEO problem.
✓
#11
Confirm critical content renders server-side, not client-side
Open your most important pages and disable JavaScript in DevTools
(Settings → Debugger → Disable JavaScript). Reload. If your main content — headings, body
text, product descriptions — disappears, it's JavaScript-dependent and invisible to most AI
crawlers. Server-side rendering (SSR) is not optional for content you want cited in AI
Overviews and chatbot answers.
CriticalAI visibility
✓
#12
Check mobile navigation is crawlable in HTML — not JS-only
Hamburger menus that only render links after a JavaScript click are
invisible to crawlers that don't execute JS. View the page source and check whether your
navigation links appear in the raw HTML. If they don't, those link equity signals aren't
being passed, and AI crawlers can't map your internal structure.
Critical
✓
#13
Remove or make compliant any intrusive interstitials
Large pop-ups covering the main content above the fold are a confirmed
Google ranking demotion signal. Cookie banners that block content on mobile are the most
common offender across the legal, healthcare, and financial sites I audit. Test your pages
on a real mobile device and check what appears before any user interaction.
Quick fix
This entire category didn't exist in a 2022 audit. By 2026, AI crawlers — GPTBot, ClaudeBot,
Google-Extended, PerplexityBot — collectively represent a meaningful share of server traffic for most
sites. How you manage them determines whether you appear in AI-generated answers, how much your server
costs, and whether your data is being used to train models without your consent.
🤖
The strategic decision you need to make first
Blocking AI crawlers entirely stops your content from appearing in AI Overviews and chatbot
responses. Allowing them all opens you to training scrapers that consume server resources without
providing attribution. The right answer for most sites: explicitly allow retrieval-focused crawlers
(GPTBot, PerplexityBot, ClaudeBot) and block aggressive training scrapers (CCBot). These are
different things and require different robots.txt directives.
✓
#14
Audit your robots.txt for AI crawler rules
Many sites still have a blanket User-agent: *
Disallow: that was written before AI crawlers existed and now accidentally blocks
them. Separately, check whether you have explicit rules for GPTBot,
ClaudeBot, PerplexityBot, Google-Extended, and CCBot. Each requires a
deliberate decision. Most sites have none.
New in 2026AI visibility
✓
#15
Implement llms.txt to guide AI content consumption
A new protocol in 2026, llms.txt lives at your
root domain and tells AI systems which pages are your canonical sources of truth —
documentation, pillar content, glossaries. Unlike robots.txt, it doesn't restrict access —
it guides it. List your most authoritative pages here. It's becoming a forward-looking GEO
signal, especially for sites targeting Perplexity and AI Mode visibility.
New in 2026AI visibility
✓
#16
Check server logs for AI crawler volume and impact
Some AI training bots crawl aggressively enough to affect server
response times. Pull 30 days of server logs and filter by user agent. If CCBot or other
undifferentiated scrapers represent more than 5% of requests, it's worth blocking them
explicitly. This is infrastructure cost management, not just SEO — and it's a conversation
to have with your dev team, not just your SEO team.
New in 2026
Schema is no longer an optional enhancement. It's the language AI search systems read to understand what
your content is, who wrote it, and whether to cite it. Sites without proper structured data are asking
AI Overviews to guess — and they often guess wrong or skip it.
✓
#17
Validate schema markup with Google's Rich Results Test
Run your ten most important pages through search.google.com/test/rich-results. Check for errors, not just
warnings. Schema drift — where your JSON-LD says a product is "InStock" but the page
visually shows "Sold Out" — is a trust signal problem. After any CMS update or AJAX-powered
page change, your schema can desynch from the visible DOM without anyone noticing.
Critical
✓
#18
Add FAQ schema to every cluster page targeting question-based queries
FAQ schema is the most direct signal you can give AI Overviews that your
page contains a clean, extractable answer to a specific question. If you're writing content
targeting "how to" or "what is" queries and you don't have FAQ schema, you're relying on the
AI to interpret your answer structure. That's a competitive disadvantage on every page it
affects.
AI visibility
✓
#19
Implement Article schema with author, datePublished, and dateModified
EEAT signals flow partly through structured data. An Article schema with
a named author, publication date, and accurate modification date tells Google and AI systems
that this content has an identifiable human expert behind it and is being maintained. Sites
without this on content pages are leaving authoritativeness signals on the table.
Quick fix
These three don't fit neatly into a category, but they show up repeatedly in audits I run — and they're
almost always flagged as new findings, not things the team was already aware of.
✓
#20
Confirm hreflang is implemented correctly for international pages
Hreflang errors are silent. A missing reciprocal annotation — where the
UK page references the US page, but the US page doesn't reference the UK page back — breaks
the whole signal without throwing any visible error. For any site serving multiple languages
or regions, run hreflang tags through a validator after every deployment that touches the
<head>.
Critical
✓
#21
Check TTFB and server response time from multiple locations
Time to First Byte over 800ms is the hidden bottleneck behind slow LCP
scores. A CDN fixes this for static assets, but if your origin server is slow — due to
unoptimised database queries, no object caching, or a shared hosting environment under load
— every other performance fix you make will hit a ceiling. Test TTFB from India, the US, and
the EU separately if you have international traffic.
Quick fix
✓
#22
Set a quarterly audit cadence — not an annual one
This is the meta-item. A technical SEO audit done once a year catches
problems after they've already cost rankings. Core Web Vitals scores can degrade after a
single third-party script update. AI crawler rules need reviewing as new bots emerge. Schema
can drift after any CMS deployment. A quarterly 90-minute sweep of the top 10 pages catches
regressions before they compound into traffic drops.
Critical
Using Claude to Run This Audit Faster
You can work through this checklist manually in about an hour. You can also use Claude to accelerate the
diagnostic layer — particularly for identifying INP-causing JavaScript, spotting schema drift, and
prioritising which issues to fix first based on your site's traffic distribution.
The prompt below is built for technical SEO analysis. Give Claude your PageSpeed Insights data, your
Search Console Core Web Vitals report, and a list of your highest-traffic pages. It will identify
patterns you might miss and sequence the fixes by expected impact.
# TECHNICAL SEO AUDIT ANALYSIS PROMPT
# Paste your data below the DATA START marker.
# Include: PageSpeed Insights results, Search Console Core Web
# Vitals report export, top pages by traffic (URL + session count),
# and any known issues you're already tracking.
You are a senior technical SEO engineer specialising in Core Web
Vitals, site performance, and AI-era crawlability. You understand
INP, CrUX data, schema implementation, and AI crawler governance.
I'm sharing my site's technical performance data. Analyse it and
give me a prioritised fix list. Focus on what will move rankings
and AI citation rate — not what will just improve lab scores.
SITE CONTEXT:
- CMS / tech stack: [e.g. WordPress / Shopify / Next.js / custom]
- Monthly organic traffic: [approx. sessions]
- Industries served: [e.g. e-commerce / healthcare / legal / SaaS]
- Known performance issues: [e.g. slow INP / LCP failing on mobile]
- Dev team capacity: [e.g. we have a developer 2 days/week]
- AI visibility goal: [e.g. appear in AI Overviews / Perplexity]
ANALYSIS TASKS — address all of these:
1. INP DIAGNOSIS: Based on the data, which pages have INP issues?
What are the likely causes given the tech stack? Suggest the
3 most impactful JavaScript fixes to investigate first.
2. LCP SEQUENCE: LCP failures are rarely just image problems.
Walk through the dependency chain: TTFB → render-blocking
resources → LCP element discovery → LCP element load time.
Where is the actual bottleneck in my data?
3. CRAWLABILITY FLAGS: From the pages listed, identify any that
look like they may have indexation issues (low impressions
relative to traffic, unusual crawl patterns, redirect chains).
4. SCHEMA GAPS: Based on my site type and content, which schema
types am I likely missing? List the 5 most impactful additions
for AI Overview citation visibility.
5. AI CRAWLER STATUS: Given my robots.txt (if I've shared it),
am I blocking or allowing the right crawlers? What should
my llms.txt list based on the pages I've shared?
6. PRIORITY FIX LIST: Rank every issue you identify by:
Impact (High/Med/Low) × Effort (Dev hours) × Urgency.
Output as a table with: Issue | Page(s) | Fix | Priority score.
OUTPUT FORMAT:
- Specific to my data — no generic advice
- Developer-actionable language for the fixes
- End with a "this week / this month / next quarter" timeline
━━━━━━━━━━━━━━━ DATA START ━━━━━━━━━━━━━━━
[PASTE YOUR DATA HERE: PageSpeed results, Search Console
Core Web Vitals export, page inventory, robots.txt content]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
The Tools I Actually Use for This Audit
The audit sequence that saves the most time: Start with Search Console → Core Web
Vitals report to identify which pages are failing. Then use PageSpeed Insights on those specific pages
to see field data. Then use Chrome DevTools Performance tab on the ones with INP issues. This top-down
approach takes about 45 minutes for a 50-page site and surfaces the issues that actually matter — not a
list of 200 low-priority warnings.
What EEAT Looks Like at the Technical Layer
EEAT is usually discussed as a content question. But there's a technical dimension to it that most guides
skip.
E
Experience
Author schema with datePublished and dateModified — proof the content has a
human author and is maintained
E
Expertise
Organisation schema, Person schema — connecting your content to a verifiable
entity with domain credentials
A
Authoritativeness
FAQ and HowTo schema — giving AI systems clean extractable answers from pages
that have already demonstrated authority
T
Trust
HTTPS throughout, no schema drift, accurate canonical tags — consistency
between what you say and what the crawler sees
The Bottom Line
Technical SEO in 2026 is not harder than it was — but it's wider. The same foundations matter:
crawlability, indexation, speed, mobile rendering. What's new is the AI layer on top. AI crawlers need
governance. Schema needs to be accurate for machine reading, not just search result features. INP
requires a fundamentally different kind of investigation than the old FID metric did.
The sites that are pulling ahead in organic search right now are the ones treating technical SEO as a
quarterly practice, not an annual project. The gap between a site audited monthly and one audited
annually widens faster than it used to — because the environment changes faster than it used to.
Use the interactive checklist above. Run the Claude prompt against your real data. Fix the highest-impact
items first, and set a calendar reminder to repeat this in 90 days.
.webp)
A
Akif Qureshi
Senior SEO Specialist & Marketing Analyst | Content Strategist
5+ yrs experience
Google Certified
6 guides
About
Driven by advanced SEO expertise, deep marketing analytics,
high-impact content strategy
With 5+ years of hands-on experience, I specialize in holistic search strategies that don’t just rank—they drive real, measurable business growth. I’ve worked across industries including healthcare, hospitality, legal, e-commerce, and professional services, helping brands dominate their target markets. My approach bridges the gap between raw data and creative execution. Every strategy I build is rooted in rigorous market analysis, structured SEO frameworks, and tailored content ecosystems—no templates, no shortcuts. Whether you’re a single-location brand or scaling across multiple cities, I create data-driven marketing systems designed to compound results and grow with you.
No sponsored content
No affiliate links
Reader supported
Ready to run your technical SEO audit?
Use the interactive checklist above, copy the Claude prompt, and paste in your PageSpeed and Search
Console data. Share this guide with anyone who manages SEO on your team.
Jump back to the Toolkit →
