SEO and AEO audit for B2B SaaS: a revenue-page protocol
A practical B2B SaaS SEO and AEO audit method for finding technical, content, evidence, and measurement gaps across the pages that influence pipeline.
A B2B SaaS SEO and AEO audit should answer a commercial question: can a qualified buyer find, understand, verify, and act on the product’s claims? That requires more than crawling every URL and exporting warnings. It requires a close review of the page system around category, use case, comparison, integration, security, pricing, documentation, and customer proof.
The SEO portion establishes whether search engines can access, index, and interpret those pages. The AEO portion examines whether the same material is clear, supported, and usable when answer systems assemble responses. Neither audit can guarantee a ranking or citation. Both can expose avoidable defects, unsupported claims, unclear page ownership, and broken paths to conversion.
Start with the buying journey, not the crawl total
A SaaS site is rarely one coherent content library. It is a collection of page types owned by product marketing, demand generation, product, customer success, security, and documentation teams. An audit that treats all URLs equally will often prioritize large low-value patterns while missing a broken pricing or integration route.
Begin by mapping page families to buyer jobs:
| Page family | Buyer question | Evidence the page should provide | Useful conversion |
|---|---|---|---|
| Category or solution | “What kind of product solves this?” | Clear definition, fit, limits, differentiators | Explore use cases |
| Use case or industry | “Will it work in my situation?” | Workflow, constraints, relevant proof | View a demo or case |
| Comparison | “Why this instead of the alternative?” | Factual criteria, dated claims, honest trade-offs | Build a shortlist |
| Integration | “Does it work with our stack?” | Supported connection, setup path, limitations | Read docs or start setup |
| Security | “Can we approve this?” | Current controls, scope, contact route | Start security review |
| Pricing | “Can we afford and buy it?” | Pricing model, qualification rules, next step | Trial, purchase, or contact |
| Documentation | “Can our team implement it?” | Accurate procedures and prerequisites | Complete a task |
| Customer proof | “Has it worked in a comparable context?” | Approved facts, method, context, limitations | Read related solution |
This is a buyer heuristic, not a platform requirement. Its value is operational: it tells the audit team which templates deserve full review and which can be sampled. For a broader audit framework, see Leaf’s combined SEO and AEO audit guide.
Test access, index controls, and rendering on revenue pages
The technical review should follow the dependency chain. First confirm that important URLs return the intended status and content. Then inspect robots rules, meta robots directives, canonical tags, sitemap inclusion, internal links, and rendered output. Do not label a URL “indexable” because one crawler received a 200 response; indexability depends on several controls, and indexing remains the platform’s decision.
Google’s documentation is explicit that robots.txt controls crawler access, not whether a URL may appear in search. That distinction matters when a team blocks parameter pages, staging remnants, or documentation paths. A blocked URL can still create confusing signals if other pages link to it or refer to it as canonical.
For JavaScript-heavy product sites, compare initial HTML with rendered content on each important template. Check whether titles, headings, body copy, canonical tags, links, and forms survive rendering. Record the test method and date. A screenshot can show a visible symptom; the affected URL set and reproduction steps make the finding usable.
Performance deserves the same discipline. Use field data when available and label lab diagnostics as lab data. The official Core Web Vitals guidance defines the current user-experience metrics, but a failed lab run does not by itself establish lost pipeline. Prioritize issues that affect major templates, real users, or task completion.
Find intent collisions and missing decision content
SaaS sites often create overlapping pages during campaigns, positioning changes, and product launches. A category page, feature page, industry page, and three old landing pages may all target nearly the same problem without giving any one URL a distinct job. The result is not merely “duplicate content.” It is unclear ownership of the buyer question.
For every commercial page, write one sentence describing the intended audience, question, and next action. If two pages receive the same sentence, decide whether to consolidate them, narrow one, or reposition it. If a page has no credible next action, decide whether it belongs in the acquisition path at all.
Then inspect the answer itself. A useful page should state what the product does, who it is for, what it requires, and where it does not fit. It should connect claims to product documentation, current policies, or approved customer evidence. Google’s people-first content guidance is not an AEO formula, but it supports a sensible audit standard: content should be useful to its intended audience and demonstrate real knowledge rather than exist primarily to attract search visits.
Avoid turning this into a word-count exercise. A concise integration page with exact prerequisites can be more useful than a long generic guide. Conversely, a high-stakes security page may need deeper definitions, scope statements, dates, and owner information.
Audit product claims as a controlled system
Answer systems can surface claims from your site, third-party profiles, reviews, documentation, and old announcements. The first job is not to produce more claims. It is to make the important ones consistent and supportable.
Create a claim register for items such as supported integrations, deployment model, data handling, certifications, customer counts, pricing terms, and product capabilities. Each claim should have:
- the exact wording used on the site;
- the URLs where it appears;
- an internal owner;
- a supporting source;
- a last-reviewed date;
- any necessary scope or qualification.
This register is especially useful when marketing pages and product documentation drift apart. A stale comparison page can be more damaging than a missing FAQ because it gives buyers a confident but wrong answer. The audit should flag inconsistency; the relevant product, legal, security, or customer owner must approve the correction.
Structured data can help represent information already visible on a page. It cannot rescue inaccurate content or force inclusion in a result. Google states that structured data must describe the visible page and is not a guarantee of a rich result. Treat markup findings as precise technical work, not as a shortcut to AI citations.
Build a reproducible answer-engine sample
AEO testing is useful when it reflects actual buying questions and preserves the evidence. It becomes theater when a provider asks a few favorable prompts, screenshots the best answer, and reports an “AI visibility score” with no denominator.
Build a fixed panel across four stages:
- Problem discovery: questions about the category or workflow without naming the brand.
- Shortlisting: questions about tools for a defined company type or constraint.
- Evaluation: comparisons, integrations, security, implementation, and switching questions.
- Branded accuracy: direct questions about the product, pricing model, features, and limitations.
For every response, save the exact prompt, product or engine, date, account or location context where known, answer text, cited URLs, unlinked brand mentions, competitors, and factual errors. Keep mention, citation, and recommendation separate. A mention is not necessarily sourced; a citation is not necessarily an endorsement; and neither proves commercial influence.
Google says that its existing SEO fundamentals apply to AI features. That is a verified platform statement. It is not evidence that a specific rewrite will earn an AI Overview citation. Any expected effect from a content change should be labeled as an experiment and retested against the frozen panel.
Verify measurement before forecasting value
An audit should not claim revenue impact when form events, trial starts, demo requests, or qualified lead stages are not measured reliably. Review the path from landing page to the actual business event. Confirm that primary conversions fire once, preserve useful source context, and exclude internal or test activity where appropriate.
Separate three classes of measurement:
- Deterministic checks: a canonical points to the correct URL, a form submits, or a schema block parses.
- Platform observations: a URL appears in Search Console, a query earns impressions, or an answer cites a page on a recorded date.
- Business outcomes: a qualified opportunity or customer can be connected to an acquisition path under the company’s attribution rules.
The first category can usually pass or fail immediately. The second and third require observation over time and have external variables. That separation prevents teams from calling an implementation successful merely because it shipped.
Use this SaaS audit scorecard to set priorities
Score each material page family from 0 to 2 on the criteria below. This is a buyer prioritization heuristic, not a search-engine scoring system.
| Criterion | 0 | 1 | 2 |
|---|---|---|---|
| Access and rendering | Blocked, broken, or materially incomplete | Accessible with unresolved conflicts | Intended canonical content reliably accessible |
| Intent ownership | Duplicated or unclear | Partly distinct | One clear buyer job and next step |
| Claim support | Important claims unsupported or inconsistent | Partial evidence or stale review | Owned, sourced, qualified, and current |
| Answer clarity | Buyer must infer the answer | Answer exists but is scattered | Direct answer with useful detail and limits |
| Internal path | Orphaned or dead-end | Linked but weakly placed | Contextual path from discovery to decision |
| Measurement | Key action untracked | Event exists but quality is uncertain | Primary action tested and tied to reporting |
Do not average the score and hide critical failures. A zero for access on pricing or security can outrank several twos elsewhere. Use the table to expose where evidence is missing, then prioritize by revenue exposure, severity, confidence, effort, and dependency.
Turn the audit into a staged backlog
The final deliverable should read like a set of work orders. Each finding needs evidence, affected URL or template, business relevance, proposed change, owner, dependency, and a verification method. Group work into “now,” “next,” and “later” rather than presenting a flat list of warnings.
A sensible sequence is usually:
- restore access, correct canonicalization, and fix broken conversion paths;
- resolve inaccurate product claims and page ownership conflicts;
- strengthen decision content and internal links on exposed revenue pages;
- correct structured data that misrepresents visible content;
- run bounded answer-engine experiments and retest the same prompt panel.
Leaf’s $875 SEO and AEO audit follows this evidence-and-backlog model for established B2B sites; implementation is not included in that fixed audit. Teams that only need an initial homepage triage can instead use the free SEO and AEO assessment, which is deliberately narrower and cannot diagnose a whole SaaS site.
The useful outcome is not a larger list. It is agreement on which buyer path is failing, what evidence supports that conclusion, who owns the correction, and how the team will know the specific defect has been fixed.