We Checked the SEO Report Against Google's Own Data
When a report says your website has an SEO problem, the next question should be simple: what did the tool actually measure?
Did it find something in your website's public code? Estimate traffic from a keyword database? Read your own Google Search Console account? Those are different kinds of evidence. They should lead to different kinds of conclusions.
We recently compared a public-facing audit of Scarborough House, prepared by our sister site Sitebuilder.solutions, with our authorized Google Search Console and Google Analytics data. Some findings were confirmed. Some needed important qualifications. The exercise was useful because it helped us separate a real implementation issue from an assumed search-performance problem.
Sitebuilder.solutions contributed the public-facing research: what could be observed in website responses, page content, metadata, and sitemaps. Scarborough House supplied the private Google measurements for comparison. The value of that approach is practical: website research can identify worthwhile questions even when the researcher does not have access to the owner's Google accounts.
Ownership disclosure: We own both Scarborough House and Sitebuilder.solutions. This was our separate verification of its findings, not an independent third-party review of a product we do not own. It was a limited case study, not a full-site accuracy benchmark or a controlled ranking experiment.
Google's measurements and an SEO tool's estimates answer different questions
Our operating rule is, “If Google measures it, then we maximize it.” That means improving the useful outcome, not increasing every number. We want fewer errors, stronger eligible search coverage, relevant clicks, and visitors who take meaningful action.
Google's own guide to using Search Console and Analytics together makes the distinction clear. Search Console is the source of truth for Google Search performance. Analytics is the source of truth for measured behavior inside the website.
That is not a claim that either tool is a complete, error-free record of a business. Search Console has reporting limits, privacy protections, and delays. Analytics depends on tracking, consent, configuration, and attribution. Contracts and deposits still need their own business records.
Nor does Search Console explain every ranking cause or report performance on every other search engine. It tells us what Google reports for the property and report we are examining.
Ahrefs is useful, but an estimate is still an estimate
Third-party tools can uncover backlinks, competitor pages, keyword opportunities, and technical problems. Those are useful research functions.
But a traffic estimate is not your website's measured traffic. Ahrefs explains this directly: its organic traffic figures depend on the keywords it tracks, search-volume estimates, rankings, and a click-through-rate model.
That does not make the tool dishonest. It means the number has a specific purpose and limitations. A vendor's traffic estimate should not overrule your first-party Google measurements. Equally, Search Console cannot replace the competitor research you cannot obtain from your own property.
The useful question is not which dashboard looks most confident. It is which source can actually answer the question being asked.
A real canonical conflict did not mean Google chose the wrong homepage
The public audit found that our homepage's HTTP header pointed to a capacity page as a canonical, while the HTML canonical correctly identified the homepage.
We reproduced that inconsistency. It is a legitimate cleanup item.
Then we checked Google's URL Inspection results. Google selected the correct separate canonical for both pages. The observed inconsistency had not become the specific indexing failure someone might infer from it.
Google's canonicalization guidance also matters: declaration details affect how annotations are interpreted. A typed canonical header is not simply interchangeable with every other canonical mechanism.
The proportionate recommendation is to make the declarations consistent. It is not to delete the capacity page, redirect the homepage, or announce that Google has replaced it.
A correct observation does not automatically prove the predicted consequence.
“My crawler failed” is not the same as “Google is blocked”
A request using a Python-default client received a 403 response for a sitemap. Other requests retrieved valid XML. Google's own sitemap reports showed successful downloads without errors or warnings.
We retained those different request conditions. We did not establish exactly which request characteristic caused the difference.
This is where a report needs careful wording. “This request received 403” is evidence. “Google cannot crawl your sitemap” is a separate claim that needs support.
The sitemap counts were also useful: 2,723 distinct page entries, with overlapping image and video inventories. Those counts describe published inventory. They do not tell us how many pages Google indexed or how many pages deserve to rank.
Live pages, valid markup, and Google recognition are separate milestones
Thinking about touring? Most couples know within the first ten minutes on the grounds. See tour times or check pricing first.
Our individual November 14 and November 15 event pages were live, contained initial-HTML event markup, and appeared in the sitemap. At the inspection date, Google still reported both URLs as unknown.
Three existing video pages also remained unindexed. That gives us something specific to investigate. It does not establish that the website needs a new video library or that missing markup caused the exclusion.
For each feature, we need to distinguish implementation, technical validity, policy eligibility, discovery, indexing, search appearance, and results.
The same principle applies to optional image-credit warnings. Missing attribution fields are not permission to invent a photographer or copyright owner. Correctness comes before clearing a warning.
More impressions did not bring more clicks
Our finalized Search Console web-search comparison produced this result:
| Metric | Aug 27 to Sep 23 | Sep 24 to Oct 21 |
|---|---|---|
| Clicks | 427 | 387 |
| Impressions | 10,462 | 12,569 |
| Click-through rate | 4.08% | 3.08% |
| Average position | 13.18 | 11.20 |
Impressions increased by about 20%. Clicks declined by about 9%. Average position improved numerically, but changes in query and page mix can affect that average.
These totals do not identify a cause. They do identify a better next question than “How do we add more pages?” We need to see which queries, pages, devices, and search appearances account for the movement.
If a useful page is appearing for the right searches but earning fewer clicks, its search presentation deserves investigation. If new impressions come from less relevant searches, the solution may be different. We should examine the evidence before rewriting titles across the website.
Analytics needs the right comparison, too
Google Search clicks and Analytics sessions are different measurements. They should not be expected to match exactly.
For an SEO comparison, Google's guide recommends isolating sessions whose source is Google and medium is organic. All-site traffic can include advertising, direct visits, social referrals, and other channels.
Our source review found an opportunity in our own historical SEO collector: it requests all-traffic Analytics data rather than a separate Google-organic series. The all-traffic view remains useful. The improvement is to distinguish the views clearly before comparing them with Search Console.
We also need to verify what our recorded key events actually mean. A key event is not automatically a confirmed tour, and zero Analytics revenue does not establish zero real-world bookings.
Useful tools for checking your own website
These resources answer different questions. Start with the question you need answered, then choose the appropriate tool rather than treating every dashboard as interchangeable.
- Sitebuilder.solutions: Our sister site, which prepared the public-facing research used in this case study. Its observations helped us identify issues to verify; they did not replace our private Google reports.
- Google Search Console: Your first-party reference for Google Search clicks, impressions, queries, indexing reports, and URL Inspection. Access to a property's private reports requires authorization.
- Google Analytics: Understand measured visits, traffic sources, engagement, and configured events on your own website. Check tracking and event definitions before interpreting those actions as business outcomes.
- Ahrefs: Research backlinks, keywords, competitors, and technical issues. Use its estimated traffic for research, not as a substitute for your own measured Search Console results.
- PageSpeed Insights: Investigate speed and user-experience issues. Keep lab diagnostics separate from real-user field data, which is not available for every page.
- Google Rich Results Test: Check supported structured data on a public page. A passing test does not guarantee indexing, policy eligibility, or an enhanced search appearance.
- Bing Webmaster Tools: Consult Bing's own search-performance reports and diagnostics. Google Search Console does not measure every search engine.
- Google Trends: Explore relative search interest and seasonal patterns. It is not a report of your website's traffic or an exact keyword-volume count.
For the reasoning behind a recommendation, link to the relevant official guidance as well as the tool. A reader investigating conflicting canonicals needs Google's canonicalization documentation, not just a generic SEO score. Useful references should help readers check the claim and take the next step, not simply fill the page with outbound links.
The useful audit ends with a testable next step
For our website, the next opportunities are to correct the confirmed header inconsistency, investigate the click decline at query and page level, separate organic Analytics reporting, and investigate the specific event and video discovery gaps.
We still need valid performance evidence before making a speed-related claim. We also need reliable links between website activity and verified business outcomes before claiming SEO produced a booking.
None of those improvements requires treating every external report as worthless. They require asking it to show its work.
A useful recommendation explains what was observed, which Google measurement it relates to, what remains unknown, and how we will check the result. That is the standard we want for outside tools and for our own reporting.
Estimates help us investigate. First-party measurements help us verify. Business records tell us whether the work produced something worth having.
Start with your own evidence: Open Google Search Console, choose a commercially important page, and compare its searches and clicks across two equal periods. For more owner-led examples, read our Tech Talk articles and how we fact-check AI claims about our venue.