How to Do Google SEO: The Search Console First Checklist
You do Google SEO by getting a page into the index, making its purpose obvious in the HTML, proving the page deserves trust, and keeping the experience fast and stable. The generic steps still apply, but Google hands you specific instruments: Search Console for diagnostics, field data for page experience, and Discover for reach beyond the results page. Work in that order and most ranking drops explain themselves.
For B2B sites, export manufacturers and stores that sell outside their home market.
Fig. 1 — index first, then quality, then experience
What is different about doing SEO for Google?
Google SEO means working with Google's own systems in Google's order: its index for inclusion, its quality assessment for trust, its Core Web Vitals for page experience and Discover for reach beyond the results page. The mechanics are general, but the diagnostics, the evidence and the check sequence are specific to Google.
Start where Google starts: the index
Before keywords, before links, before copy. If the URL is not indexed, every other task is theoretical. This is the sequence we run in Search Console on day one.
Two reports carry most of the diagnosis. The indexing report tells you whether Google accepted the page, and the performance report tells you whether searchers did. A page that fails the first never appears in the second, which is why teams that start with keyword research spend weeks optimising pages Google has quietly excluded.
The Google-only checklist
Eight items that matter specifically for Google. The coloured dots group them: index, quality, experience, reach.
Confirm the URL is indexed, not merely discovered
"Discovered, currently not indexed" is a crawl-demand problem, not a content problem. Fix internal links before rewriting anything.
Declare one canonical per topic
Self-referencing canonicals on every page, plus redirects for retired URLs. Conflicting signals waste the crawl and split the ranking history.
Let the title and snippet say what the page does
Google rewrites many titles and descriptions. Put the query and the offer in the H1 and the first sentence, not only in the meta tags.
Name the person behind the page
A real author with a real profile, a role and a way to be contacted. Anonymous pages ask the reader, and the system, to take the claim on faith.
Show first-hand evidence
Test data, factory photos, process shots, comparison tables, measurement method. Experience is the part of E-E-A-T a competitor cannot paraphrase quickly.
Pass Core Web Vitals on field data
Good means LCP at or under 2.5 seconds, INP at or under 200 milliseconds, and CLS at or under 0.1, measured on real visits, not a lab score.
Opt in to large image previews
Without the large preview setting, Discover and image results show a thumbnail. Wide, high-quality images with honest captions are the entry ticket.
Keep dates and updates honest
Changing a date without changing the content is the fastest way to lose Discover eligibility. Update the substance, then the date.
Google-specific notes, next to the generic advice
The same task often needs a different standard once Google is the target. This is where the two diverge.
| Area | Generic advice | What Google actually checks |
|---|---|---|
| Indexing | Get the page crawled | Indexing report reasons, declared canonicals, and crawl demand created by internal links |
| Titles | Write a unique title tag | Pixel width truncation on mobile, and snippets that are rewritten from page copy |
| Quality | Publish long, useful articles | Experience, expertise, authoritativeness and trust, judged on evidence rather than word count |
| Speed | Improve PageSpeed score | Field Core Web Vitals per origin, segmented by mobile and desktop |
| Markup | Add structured data | Eligibility for a specific rich result type; invalid markup is ignored, not punished |
| Reach | Rank in blue links | Discover, image packs and AI summaries, each fed by different signals |
E-E-A-T without the mystique
It is not a score you can look up. It is the set of things a careful reader would check before believing you, and each one has a page-level fix.
Did you actually do the thing?
Photographs of the line, the sample, the install, the failed batch. A method section that only someone who ran the process could write.
Does the page show command of the subject?
Correct terminology, tolerances, standards and trade-offs, explained without padding. Specificity is the signal, not length.
Do other people treat you as a source?
Citations from trade bodies, suppliers, industry publications and directories that a buyer would recognise. Relevance before volume.
Can a stranger verify you?
Company address, contact route, warranty terms, clear authorship, no fake countdowns. Trust is the item the other three hang from.
One practical test: open your highest-traffic page and ask whether a sceptical engineer would forward it to a colleague. If the answer is no, the page has a trust problem that no amount of link building will fix.
Core Web Vitals: the three numbers that count
Field data at the 75th percentile, mobile and desktop assessed separately. Lab tools explain the cause; field data decides the verdict.
Most failures are unglamorous: an oversized hero image, a font that swaps late, a chat widget loaded before the copy, or a banner that pushes the page down after it renders. Reserve space for every element that arrives late, serve images at the size they are displayed, and defer anything that is not needed for the first screen. A page that scores well in a lab test and fails in the field is usually a page that behaves differently on a slow phone, which is the device your buyer is holding.
Discover is a different surface with different rules
Discover shows pages to people who were not searching. That means the page has to earn attention without a query to match, and it has to look like it belongs to a real publication.
Large image previews, a clear topic, a specific headline and a visible date all matter more here than in the results page. Content that reads like a template, repeats a press release, or promises something the page does not deliver gets suppressed quickly, and recovery is slow.
- Index status for the URL in question
- Canonical and rendered HTML
- Queries with impressions and no clicks
- Field vitals on the mobile template
- Trust block: author, address, contact
- Discover settings and image size
Work that list top to bottom and you will rarely guess wrong about a drop. Start in the middle, at "quality" or "algorithm", and you will spend weeks on changes that cannot be verified. If you need the underlying model first, our explainer on Whats Is SEO? covers the four stages in plain language before you touch a single setting.
Want the Google checklist run on your domain?
Send the domain. We will return the index reasons, the failing templates and the first five fixes in order.
When the index is clean and the pages are still losing, the gap is usually technical at the template level: rendering, canonical logic, hreflang or pagination. That is the work our Technical SEO Services are built around, and it is the part most content-only plans never reach.
Google SEO questions we hear every week
Is Google SEO different from SEO in general?
The mechanism is shared, but the diagnostics and evidence are not. Google gives you Search Console, field data and Discover settings; other engines give you different tools and different surface rules.
How often does Google crawl a site?
As often as the site's perceived value and change rate justify. A large, frequently updated, well-linked site is crawled more than a small static one, and crawl demand is a signal you can influence with internal links and sitemaps.
Do meta keywords still matter to Google?
No. Google has ignored the meta keywords tag for years. The words that count are in the visible copy, the headings and the internal link text pointing at the page.
Why did my page disappear from Google?
Check in order: a noindex tag, a canonical pointing elsewhere, a robots block, a server error on the template, or a manual action in Search Console. The cause is almost always one of those five.
Does Google use social signals for ranking?
Not directly as a ranking factor. Social activity can create the links and brand searches that do matter, which is why it helps indirectly and inconsistently.
How long does Google take to index a new page?
Hours for a strong site with a clear internal link, days to weeks for a new domain, and potentially never for a page that nothing links to and no sitemap lists.
Is structured data required to rank?
No, but it is required to be eligible for most rich results. Invalid markup is ignored rather than penalised, so the cost of trying is low and the upside is presentation.