The 2026 SEO Checklist (Technical + Content + AI)

SEO work fails most often not from a lack of knowledge but from inconsistent execution - the same ten items get checked on the flagship pages and quietly skipped on the other 200. A checklist run against every template type, not just hand-picked example pages, is what catches the systemic issues that actually move aggregate traffic, since a single broken component repeated across a template affects far more URLs than any one-off content problem.

Pass thresholds matter more than binary yes/no checks. 'Has a title tag' is nearly always true and tells you nothing useful; 'title tag is under 60 characters and contains the primary keyword within the first half' is a real pass/fail line that correlates with actual SERP display behavior. This checklist gives concrete thresholds wherever one exists rather than vague directional advice.

Sequencing the checklist matters as much as the items themselves. Fixing on-page content before technical crawlability is resolved wastes the content work, since Google cannot properly index or evaluate a page it cannot reliably crawl in the first place. Work top to bottom through the sections below in order the first time through a new site; subsequent passes can be run in any order as targeted maintenance.

Re-running the full checklist quarterly, with the technical section reviewed monthly, catches regressions before they compound. Sites redesign templates, marketing teams add third-party scripts, and developers ship changes that silently break canonical tags or sitemaps - none of which show up in rankings for weeks, by which point the fix is more expensive than a monthly check would have been.

HTTPS should be enforced site-wide with HTTP requests 301-redirecting to HTTPS equivalents, not just available as an option - mixed HTTP/HTTPS availability with no forced redirect is a fail, since it splits link equity and creates duplicate content between the two versions of every URL.

robots.txt must not disallow CSS, JavaScript, or image directories needed for rendering; the pass criterion is fetching robots.txt directly and confirming no Disallow rule matches your asset paths, then confirming in Search Console's URL Inspection rendered-view that the page visually renders correctly, since a render failure caused by blocked assets silently degrades how Google evaluates page experience.

An XML sitemap must exist, validate without errors, be submitted in both Google Search Console and Bing Webmaster Tools, and contain only canonical, indexable, 200-status URLs - a sitemap padded with redirects, 404s, or noindexed URLs is a fail even if it technically validates as well-formed XML, since it wastes crawl budget and signals poor housekeeping.

Zero uncontrolled crawl errors is the target in the Coverage report's error category specifically - a handful of expected 404s from genuinely removed content is normal and not a fail, but 404s or 5xx errors appearing on URLs that should resolve, or on URLs still linked internally, is a fail requiring either a fix or a redirect.

Canonical tags must be self-referencing on every indexable page by default, and cross-checked against Search Console's 'Google chose different canonical than user' report - any meaningful volume of URLs in that report is a fail indicating your canonical signals are being overridden by conflicting internal linking or sitemap entries.

Largest Contentful Paint passes at under 2.5 seconds measured at the 75th percentile of real-user field data from the Chrome UX Report, viewable in Search Console's Core Web Vitals report or PageSpeed Insights' field data section - lab data from a single Lighthouse run is not the pass criterion and should not be treated as one.

Interaction to Next Paint passes at under 200 milliseconds at the same 75th percentile field-data threshold. Sites failing INP should audit third-party script count first, since chat widgets, ad tags, and marketing pixels stacked without deferral or lazy-loading are the most common cause, followed by unoptimized event handlers on interactive components.

Cumulative Layout Shift passes at under 0.1. The fastest fail-to-pass fix is auditing every image and embedded iframe for explicit width and height attributes or CSS aspect-ratio, and checking whether web fonts are configured with font-display: optional or swap paired with a matched fallback font to prevent visible reflow on load.

Mobile usability must show zero issues in the dedicated Mobile Usability report - tap targets should have adequate spacing, viewport meta tag must be present and correctly configured, and text should be legible without a pinch-zoom gesture on a standard mobile viewport width.

Run this section's checks against your five highest-traffic template types separately, not just the homepage, since Core Web Vitals scores are commonly homepage-optimized while product, category, or article templates carrying the bulk of organic traffic quietly fail.

Every indexable page needs a unique title tag under roughly 60 characters (measured in pixel width in practice, since character count is an approximation) containing the primary target keyword within the first half of the string - duplicate titles across multiple URLs is a fail that indicates either genuine content duplication or a templating bug worth investigating immediately.

Meta descriptions should be unique, under roughly 160 characters, and written as a genuine value-proposition sentence rather than a keyword list - since search engines frequently rewrite descriptions dynamically anyway, the pass criterion here is less about guaranteed display and more about giving the algorithm a strong, accurate candidate to work from when it does use yours.

Exactly one H1 per page, matching the page's actual topic and ideally close to but not identical to the title tag, is the pass criterion - zero H1s or multiple competing H1s both count as fails, and this is checkable at scale with a simple site crawl rather than manual review.

H2 and H3 usage should follow a logical outline that mirrors how a knowledgeable person would explain the topic verbally, with no skipped levels (an H3 appearing before any H2 on the page) - skipped heading levels are a common CMS-plugin artifact and a fail worth fixing since it degrades passage-level parsing.

Every content image needs descriptive alt text that would make sense to someone who cannot see the image, not a filename or a keyword stuffed phrase - purely decorative images should have empty alt attributes rather than omitted ones, since an omitted alt attribute is treated differently by assistive technology than an intentionally empty one.

URLs should be short, lowercase, hyphen-separated, and descriptive of the actual content rather than containing session IDs, unnecessary parameters, or auto-generated numeric identifiers as the only distinguishing element - a URL like /blog/p?id=48213 is a fail even if the page content itself is excellent, since it communicates nothing to either users or crawlers scanning a SERP.

Every priority page should receive internal links from at least three other indexed pages using descriptive, varied anchor text rather than repeated exact-match phrases or generic 'click here' text - zero or one inbound internal link is a fail indicating an orphaned or under-supported page regardless of how strong its own content is.

Structured data should be present wherever a matching schema.org type exists for the content - Article, Product, FAQPage, BreadcrumbList, LocalBusiness - and validated with zero errors in the Rich Results Test, run per template type rather than per page, since a single broken shared component invalidates markup across every page using that template simultaneously.

Breadcrumb navigation, both visible in the UI and marked up with BreadcrumbList schema, should exist on every page more than one level deep in the site hierarchy - its absence is a common fail that both hurts crawlability of deeper pages and removes an easy SERP display enhancement.

Before publishing or auditing any page, check the actual top-ten results for its target query and confirm your page's format matches the dominant format shown - a fail here is a product page targeting a query where nine of the top ten results are educational guides, or vice versa, regardless of how well the page is written internally.

Content should cover the topic's reasonably exhaustive subtopic set rather than a single narrow angle - the pass check is whether a knowledgeable reader could think of an obvious follow-up question your page does not answer; if three or more such gaps exist, the page fails the depth criterion and needs expansion before further optimization is worthwhile.

Every page should be reviewed and meaningfully updated at least once every 6 to 12 months for evergreen topics, and far more frequently for time-sensitive topics like pricing, statistics, or software version references - a stale statistic or an outdated screenshot is a common, easily fixed cause of declining trust signals on otherwise strong pages.

Pages with zero organic clicks over a trailing 12-month window and no meaningful internal linking purpose should be candidates for consolidation into a stronger page or outright removal with a redirect, since a large volume of thin, unvisited pages dilutes the perceived average quality of the site as a whole in aggregate quality assessments.

Google Business Profile, where applicable, should be fully completed - category, hours, all photo slots reasonably filled, services listed, and posts published at least monthly - a profile that is merely claimed but sparse is a common fail that caps local pack visibility regardless of website-side SEO quality.

Business name, address, and phone number should match exactly, character for character, across all citation sources - directories, social profiles, and the website footer - since inconsistent NAP data is one of the more persistent, quietly damaging local SEO failures and is entirely within the business's control to fix.

A healthy quarterly target for most small-to-midsize sites is 3 to 5 newly earned backlinks from topically relevant, genuinely independent domains, rather than a larger volume of low-relevance links purchased in bulk - the pass criterion is relevance and independence of the linking domain, not raw count.

Run a backlink profile review at least twice a year specifically hunting for toxic patterns - link farms, clearly paid link networks, or foreign-language spam domains linking with exact-match commercial anchors - and use the disavow tool for any pattern that looks deliberately manipulative rather than naturally acquired, since ignoring an actively toxic pattern is a fail even if current rankings have not yet visibly suffered.

Robots.txt should explicitly allow the AI crawlers you want citing you - GPTBot, Google-Extended, PerplexityBot, and ClaudeBot among the most relevant as of 2026 - since a default-deny or accidentally blanket Disallow rule silently removes a site from AI training and retrieval consumption without any visible error in traditional SEO tools.

Priority content should include a genuine Q&A or FAQ-structured section addressing the specific questions users actually ask about the topic, marked up with FAQPage schema where eligible, since generative answer engines consistently extract more reliably from clearly bounded question-and-answer pairs than from prose paragraphs covering the same information.

Track citation presence across at least ChatGPT, Gemini, and Perplexity for a defined list of 15 to 25 target prompts monthly, since traditional rank tracking tools do not capture this and the only reliable method is running the actual prompts and logging which sources get cited - a fail here is simply never having checked, which describes most sites even in late 2026.

Third-party reviews, comparison articles, and forum mentions on sites AI engines are known to trust and cite heavily (community forums, established review sites, industry-specific publications) should be actively pursued as their own line item, since generative engines frequently cite these intermediary sources discussing your product rather than your own site directly, meaning your own on-site AI optimization is necessary but not sufficient.

Google Search Console and Bing Webmaster Tools should both be verified and actively monitored, not merely set up once years ago and forgotten - the pass check is whether someone has reviewed the Coverage and Performance reports (Google) and SEO Reports and Search Performance (Bing) within the last 30 days.

Conversion tracking should be connected to organic traffic specifically, not just aggregate site conversions, so that SEO impact can be measured in outcomes rather than only in traffic or ranking proxies - a fail here is a site that can report organic sessions but cannot answer how many of last month's leads or sales originated from organic search.

A defined core query set of 20 to 50 target keywords should have position tracked consistently over time, whether through Search Console's own position data or a dedicated rank tracker, so that ranking movement can be attributed to specific work rather than inferred anecdotally from overall traffic trends.

Analytics should distinguish branded from non-branded organic traffic, since branded query growth often masks stagnant or declining non-branded visibility, which is the segment that actually reflects new-customer-acquisition SEO performance rather than existing brand awareness driving direct-navigation-style searches.

Keyword cannibalization exists when two or more URLs on the same site rank for the same core query within the same rolling 90-day window, visible in Search Console by filtering the Performance report to a specific query and checking the Pages tab for multiple entries with meaningful impressions - the fix is consolidating into the stronger page or clearly differentiating intent between the two, never leaving both to compete indefinitely.

Orphan pages - pages with zero internal links pointing to them despite being indexed or intended to be - should be identified quarterly by cross-referencing the full sitemap URL list against a site crawl's internal link graph; any indexable URL absent from the internal link graph is a fail requiring either a link to be added or the page to be reconsidered for removal.

Thin content is defined here as any indexable page falling well short of what competing top-ten results demonstrate as necessary depth for its target query, not by an arbitrary universal word count - the pass check is comparative to the current SERP, since a 300-word page can be entirely sufficient for a narrow factual query while a 300-word page competing against 2,000-word comprehensive guides is a fail regardless of writing quality.

Run this three-part audit (cannibalization, orphans, thin content) every quarter as a fixed calendar item rather than reactively after a traffic drop, since all three problems develop slowly and silently, and a single quarter of neglect can erode months of otherwise solid execution across every other section of this checklist.

Monthly: technical crawlability spot-check, Core Web Vitals field-data review, Search Console and Bing Webmaster Tools report review, and AI-crawler citation tracking against your defined prompt list - these are the fast-moving categories where a month of drift is enough to cause a measurable, avoidable regression.

Quarterly: full on-page pass criteria review across every template type, content depth and freshness audit against current SERPs, cannibalization and orphan page audit, backlink profile toxicity check, and a fresh look at whether your current content clusters still map to how your actual customers are searching, since query patterns shift gradually and a quarterly cadence catches the drift before it becomes a full content-strategy gap.

Annually: revisit the entire checklist structure itself, since search engine behavior genuinely changes year over year - new Core Web Vitals metrics get introduced, new AI crawlers emerge, and new schema types become eligible for rich results - meaning a checklist frozen from a prior year will itself become a source of missed opportunity rather than a safeguard against one.

The single highest-leverage habit across this entire operating rhythm is treating every pass criterion as genuinely binary and checkable, not aspirational - a checklist item that cannot be verified with a specific number, report, or tool output is not yet specific enough to be useful, and should be rewritten into a concrete threshold before it earns a permanent place in the recurring review.

Before any new site or major redesign goes live, engineering and SEO should jointly sign off on a staging-environment crawl using the production robots.txt and sitemap configuration, confirming zero unintended noindex tags remain from the staging environment and that every canonical tag resolves to itself or to the correct intended target rather than to a staging URL. Acceptance criteria: a full crawl report with zero staging-domain references and zero unintended noindex directives, reviewed and approved in writing by both engineering and SEO leads before the launch window opens.

URL structure and redirect mapping for any migrated or renamed pages must be finalized and tested before launch, with a complete old-URL-to-new-URL mapping spreadsheet covering one hundred percent of previously indexed pages, each redirect tested to confirm a single 301 hop rather than a redirect chain. Acceptance criteria: redirect mapping spreadsheet complete with test results logged for every row, zero chains longer than one hop, and zero mapped URLs returning a 404 or 5xx status.

Core Web Vitals must be measured on the actual production build under realistic network conditions, not just on a fast development environment, with Largest Contentful Paint, Interaction to Next Paint, and Cumulative Layout Shift all falling within passing thresholds on the top templates by expected traffic share. Acceptance criteria: passing Core Web Vitals scores on a field or lab test simulating a mid-tier mobile device and throttled connection for every major template type, documented with timestamped test results.

Structured data implementation must be validated against Google's Rich Results Test and Schema.org validators for every template type using it, with zero errors and any warnings explicitly reviewed and accepted rather than ignored by default. Acceptance criteria: a validation report per template type showing zero errors, attached to the launch sign-off document with named reviewer approval.

Analytics and conversion tracking must be verified end to end in a staging or pre-launch environment before go-live, confirming pageview, event, and conversion tracking fire correctly across desktop and mobile for the primary user journeys. Acceptance criteria: a completed tracking QA checklist with screenshots or recordings of each tracked event firing correctly, signed off by whoever owns analytics reporting post-launch.

Within the first 24 hours post-launch, submit the updated XML sitemap in both Google Search Console and Bing Webmaster Tools, and run URL Inspection on a representative sample across every template type to confirm the live version matches what was tested pre-launch. Acceptance criteria: sitemap submission confirmed accepted in both tools, and a sample of at least ten URLs per template type showing correct indexable status with no unexpected noindex or canonical mismatch.

Daily through day seven, monitor the Coverage or Page Indexing report for any spike in errors, excluded pages, or a sudden drop in total indexed count relative to the pre-launch baseline captured during the launch phase. Acceptance criteria: a daily logged snapshot of indexed page count and error count, with any single-day change exceeding five percent of total pages triggering an immediate engineering investigation before day's end.

Server response time and error rate monitoring should run continuously through this window, since launch-day traffic patterns and any lingering caching or CDN misconfiguration tend to surface within the first week under real load rather than in staging tests. Acceptance criteria: average server response time within the same range measured during pre-launch testing, and 5xx error rate below a defined threshold agreed with engineering, checked at least twice daily.

Redirect integrity should be spot-checked again on day three and day seven using a fresh crawl rather than relying solely on the pre-launch test, since DNS propagation delays, CDN cache staleness, or configuration drift occasionally introduce redirect issues that were not present during the original test. Acceptance criteria: a follow-up crawl on day three and day seven showing the same zero-chain, zero-404 result as the pre-launch redirect audit.

Any Search Console or Bing Webmaster Tools manual action, security issue, or spam alert triggered during this window must be treated as a same-day priority regardless of other workstreams, since early false-positive flags from a botched migration are far easier to resolve immediately than after they compound with weeks of continued indexing under the flagged state. Acceptance criteria: zero unresolved manual action or security alerts remaining open past 24 hours from first detection.

By the end of week two, every page identified in the pre-launch audit as having a duplicate, missing, or truncated title tag or meta description should be rewritten and republished, with each rewrite following the site's established title tag length and keyword-inclusion guidelines. Acceptance criteria: zero remaining duplicate or missing title tags across the full site as confirmed by a fresh Search Console or third-party crawl, logged against the original issue list.

Internal linking gaps identified during the technical audit, particularly orphaned pages with zero internal inbound links, should be resolved by week three through targeted linking from relevant high-authority pages, with priority given to pages supporting revenue-relevant query intent. Acceptance criteria: zero orphaned pages remaining among priority URLs, verified through a crawl tool's internal link count report.

Thin or near-duplicate content pages flagged during the audit should be consolidated, expanded, or removed with appropriate redirects by week three, following a documented decision for each affected page rather than a blanket action, since some thin pages serve legitimate functional purposes and should be intentionally excluded from indexing rather than deleted. Acceptance criteria: a decision log covering one hundred percent of flagged pages, each marked consolidated, expanded, or intentionally excluded with a stated reason.

Structured data coverage should expand by week four to cover any template type that launched without it due to time constraints, validated the same way as the pre-launch templates were. Acceptance criteria: validation report showing zero errors for all newly covered template types, matching the pre-launch validation standard.

By week four, a full Core Web Vitals recheck across all templates should confirm real-world field data from the Chrome User Experience Report, where available, aligns with the lab data gathered pre-launch, since lab and field data can diverge once real user device and network diversity enters the picture. Acceptance criteria: field data passing thresholds for at least seventy-five percent of tracked page views per template, with a documented remediation plan for any template falling short.

During month two, complete a query-intent mapping exercise comparing your target keyword universe against your existing page inventory, identifying valuable intent clusters with no matching page. Acceptance criteria: a documented gap list with each identified cluster assigned an owner and a target publish date within the quarter, reviewed by the content lead.

Existing pages showing high impressions but low click-through rate in Search Console's Performance report should be prioritized for title tag and meta description testing, since this is one of the highest-leverage, lowest-cost improvements available at this stage. Acceptance criteria: at least the top twenty affected pages rewritten and republished, with a follow-up click-through rate check scheduled four weeks after each change to measure impact.

A backlink outreach plan targeting a small number of topically relevant, genuinely earnable placements should be scoped and initiated, explicitly avoiding low-quality directory or paid-link schemes that risk a manual action. Acceptance criteria: an outreach target list of relevant, vetted domains with outreach status tracked per domain, reviewed monthly against actual placements earned.

Competitor content and backlink gap analysis should be completed to identify topics or link sources competitors rank for or hold that your site currently lacks, feeding directly into the content gap list and outreach target list rather than existing as a standalone report nobody acts on. Acceptance criteria: gap analysis findings explicitly cross-referenced into the month two content and outreach plans with owners assigned.

Local SEO elements, where relevant to the business, including Google Business Profile and Bing Places listing accuracy, category selection, and review response cadence, should be audited and brought current by the end of month two. Acceptance criteria: both listings verified accurate against current business information, with a documented review-response process in place and a named owner for ongoing maintenance.

By month three, a full comparison against the original pre-launch or pre-project baseline should be completed across organic traffic, indexed page count, average position for priority queries, and Core Web Vitals pass rate, presented in a single consolidated report rather than scattered across separate tool dashboards. Acceptance criteria: a completed comparison report reviewed with leadership, with each metric showing either measurable improvement or a documented explanation and remediation plan for any metric that has not improved.

AI-visibility checks should be added to the standard review at this stage, including manual testing of priority queries against Google AI Overviews and, where relevant, Copilot, to assess whether the site is being cited, and if not, whether content structure, passage clarity, or crawlability is the likely blocker. Acceptance criteria: a documented AI-visibility test log covering the top twenty priority queries, with a remediation note for any query showing zero citation despite a strong organic ranking.

Structured internal reporting cadence should be formalized by month three so that the weekly, monthly, and quarterly review rhythms described elsewhere in this guide continue automatically rather than depending on the initial launch team's continued manual attention. Acceptance criteria: a written recurring review calendar with named owners for each cadence, stored somewhere the whole team can reference without depending on any single person's memory.

Any remaining items from the month one and month two gap and remediation lists that are still open should be explicitly triaged at the month three review: either committed to a specific near-term date, formally deprioritized with a documented reason, or escalated if blocked by a dependency outside the SEO team's control. Acceptance criteria: zero items remaining in an ambiguous, undated state after the month three review meeting.

For teams using paid tooling to support this ongoing checklist workflow, plan selection should match team size and site count rather than defaulting to the largest tier: a Starter plan at $29 per month suits a single site with one or two owners running this checklist, a Pro plan at $69 per month fits a growing in-house team covering several properties, a Guru plan at $125 per month supports larger content and SEO teams needing deeper historical tracking and more seats, and a Business plan at $249 per month is built for agencies running this exact ninety-day checklist across many clients simultaneously; every tier includes a 15-day free trial with no credit card required up front, and there is no free-forever plan, reflecting the genuine ongoing cost of the crawling, monitoring, and reporting infrastructure behind the checklist.