Search Console, logs and performance evidence

0:00 Search Console, logs and performance evidence

A crawler simulates. It shows you what could happen. Search Console shows what Google did with your URLs. Logs show what Googlebot actually requested. And field performance data shows what real users experienced. A strong audit joins all four. In this lecture, you'll gather that evidence for Kiran Home, segment it by market and template, add the newer twenty twenty-six sources, and build an evidence log that makes every claim in your report traceable.

0:32 Search Console questions

Start with Search Console, question by question. When did the decline start, and where? Use Performance, compare periods, and filter by page segment and country. For Kiran Home, the illustrative drop starts the week of relaunch, concentrated in UAE and Pakistan traffic to product pages. Which URLs aren't indexed, and why? Page indexing filtered by sitemap shows many product URLs as alternate page with proper canonical tag. What canonical did Google pick? URL Inspection on samples shows the UK product for UAE pages. And are rich results intact? Product items fell, with offers missing on some templates.

1:14 Faster, fairer comparisons

Use Search Console's newer features to make this faster. The branded queries filter, where it's available for the property, splits brand from non-brand, or you can use your own regex. Custom annotations let you mark the relaunch date directly on the chart. And always compare twenty-eight days before and after the relaunch, and the same windows a year earlier, to control for seasonality, for example if Ramadan fell in the comparison window one year and not the other.

1:48 Logs: what Googlebot did

Now logs. Cloudflare logs for four weeks, filtered to verified Googlebot and classified by segment, tell a clear story for Kiran Home. The large majority of requests go to facet parameters. A noticeable share goes to internal search URLs, linked from a popular searches widget. Only a small share goes to products, and many products weren't requested at all in four weeks. And Googlebot still requests pre-relaunch URLs heavily. Two conclusions: crawl attention is being spent on URLs with no value, and legacy URLs still matter to Google, so redirect gaps matter.

2:28 The unarguable join

Here's a join that makes the point unarguable. Take products with zero verified Googlebot hits in four weeks, and cross-reference them with revenue from the pre-relaunch period. If many former best-sellers haven't been crawled at all, that's a headline finding for the stakeholder report. Founders don't care about crawl budget. They care that their best-selling brass lanterns haven't been visited by Google in a month.

2:56 Performance + rendering evidence

Performance evidence. Run PageSpeed Insights for each template and read the field data first. For Kiran Home, illustratively, category pages have poor LCP because the hero carousel image is lazy-loaded, and product pages have poor INP and CLS because a reviews widget shifts layout and blocks the main thread. Where CrUX lacks data for a URL, use origin data and Search Console's Core Web Vitals groups, then use a lab trace to find the cause. Rendering evidence too: URL Inspection shows customers also bought links only load after interaction, so they add no discovery value.

3:37 New evidence sources

And the newer sources for twenty twenty-six. Search Console's generative AI report shows whether Kiran Home pages appear in AI Overviews or AI Mode, by page and country, so you can see if AI exposure fell too. Bing Webmaster Tools shows whether Bing indexing and Copilot citations dropped, and a second engine's view helps separate site problems from Google-specific changes. And CDN firewall events show whether real crawlers, including Bingbot and AI search bots, are being challenged since the relaunch. That's evidence for hypothesis five.

4:14 Hands-on: before/after by segment

Hands-on. The lesson text has a short script using the Search Console API that pulls clicks by page and country for twenty-eight days before and after the relaunch, assigns a segment from the URL, and pivots before versus after with a percentage change. It reuses the authenticated client from the Technical SEO Mastery course. For bigger sites, paginate with start row. This single table usually tells you where the drop is concentrated within minutes.

4:46 Example 2: seasonality, not SEO (illustrative)

Worked example two, a different client, illustrative, where the evidence exonerates the relaunch. A UK garden furniture retailer's segmentation shows the decline spread evenly across all templates and markets, starting two weeks before the relaunch, and matching the end of the spring buying season in last year's data too. Logs show healthy crawling, and indexing is stable. The evidence points to seasonality, not a technical regression. The audit says so, and saves the client from an unnecessary emergency rebuild. Good auditing sometimes means saying, it isn't technical.

5:24 Watch me do it: where did it drop?

Watch me do it. I'll build the evidence for where Kiran Home's drop happened. Step one: in Search Console Performance, I add a custom annotation on the relaunch date so every chart shows it. Step two: I apply the branded queries filter and choose non-branded, then compare twenty-eight days after the relaunch with twenty-eight days before. Step three: I add a page filter with the regex for UAE product URLs. Non-brand clicks are down sharply. I switch the filter to UK product URLs. Roughly stable. That contrast is the story. Step four: I repeat the comparison with the same dates last year to check seasonality. Last year's pattern was flat across those weeks, so seasonality doesn't explain it. Step five: I run the before-and-after API script by segment and country, and paste the pivot into the evidence log as E two. Step six: Page indexing, filtered by the UAE product sitemap: most URLs are alternate page with proper canonical tag. That's E eight. Step seven: I open the generative AI report and see UAE product pages also lost AI impressions, and in Bing Webmaster Tools the indexed product count dropped. Both go into the log. Each row has a source, dates, filter and file.

6:53 Check the calendar

Here's a quick mistake check before you trust your comparison. Look at the date ranges. If your before period includes the Ramadan and Eid weeks for UAE shoppers and your after period doesn't, or the other way round, the comparison is contaminated. The same goes for Black Friday in the UK. So for Kiran Home, I write both date ranges on the evidence log row, and next to them the holidays that fall inside each. If they don't match, I shift the windows or add a year-on-year comparison. It takes two minutes, and it stops the most common false conclusion in post-launch audits: blaming the relaunch for what was really a calendar effect.

7:42 The evidence log

Build the evidence log. Every claim in your final report must trace back to evidence: an ID, the source, the date range, the filter, the finding, and the file or screenshot. For Kiran Home: E seven, Cloudflare logs for March, verified Googlebot, facets dominate requests. E eight, Search Console Page indexing for the UAE product sitemap, most URLs alternate page with proper canonical tag. This protects you when a stakeholder challenges a recommendation, and it makes verification later straightforward.

8:16 Mistakes, recap, try this now

Common mistakes. Reading Performance data without segmentation, so a market-specific drop gets diluted. Comparing only with the previous twenty-eight days and ignoring seasonality. Using unverified Googlebot log lines. And treating Lighthouse lab scores as the verdict on Core Web Vitals. Recap: join Search Console, logs, field data and rendering evidence; segment everything; use the new AI and Bing sources; log every piece of evidence. Try this now: run a before-and-after comparison by segment and country for your own site's last major change.