How Google Ranks Websites: Crawling, Indexing, and Ranking Factors

Google ranks pages for particular searches. A page first needs to be discovered and processed, then compete with other eligible pages for that search. A technically tidy homepage does not automatically become the best answer to every service question.

That distinction matters when you receive an SEO audit. A missing element can deserve attention without being the reason a competitor appears above you.

We examined 18 plumbing and HVAC business homepages across New England on September 20, 2026 to see what basic technical checks actually reveal. We measured downloaded HTML and robots files, not rankings. The findings give us concrete examples of where a checklist stops and further investigation begins.

Crawling, indexing and ranking are different stages

Google's explanation of how Search works separates crawling, indexing and serving results.

Crawling is retrieving a page. Internal links and other discovery paths help a crawler find URLs. Access rules, server responses and rendering can affect what it retrieves.

Indexing is processing the content and deciding what to keep in the search index. Retrieving a URL does not guarantee indexing. Duplicate pages, canonical choices and content assessment can affect what is retained.

Ranking happens in response to a search. Google evaluates eligible pages against the request and its context. That is why a business can perform well for its own name and poorly for an unfamiliar customer's service query.

When diagnosing a page, establish which stage needs attention before changing its wording. A crawl restriction and an unhelpful answer require different fixes.

What we found on 18 business homepages

Our sample included three businesses per New England state. All 18 homepage fetches returned HTTP 200. Their markup still varied:

See the full observations and methodology. This was a convenience sample, not a ranking study or a representative estimate of business websites.

The biggest lesson is what the results cannot establish. All eighteen returned a page and none exposed a checked noindex directive. We still cannot conclude that Google indexed any particular URL, chose its declared canonical or ranks it for a valuable query. Those questions require different evidence.

Interpret each audit flag before fixing it

Multiple H1 elements need inspection, not a penalty claim

Five sampled homepages had multiple H1 elements. A source parser counts markup even when a heading belongs to a hidden slide or an alternate layout. We did not render every site to determine what a visitor sees.

On your own page, check whether the visible heading explains the service and whether the reading order makes sense. Simplify confusing duplication where needed. A count alone is not a diagnosis of a ranking penalty.

A canonical tag needs the right destination

Sixteen sampled homepages declared a canonical URL. Our count measures the presence of an element, not whether the destination was appropriate or Google accepted it.

If a service page points to the homepage by mistake, adding more copy will not resolve that configuration issue. Inspect the selected canonical in Search Console alongside the site's declaration and internal links. Google's canonicalization guidance explains how these signals work together.

A description is a presentation detail worth checking

Seventeen pages had a nonempty meta description. That does not tell us whether the description accurately summarizes the page or appears in search results. Google can generate a different snippet from page content; its snippet documentation explains the distinction.

Write descriptions to help a person understand the page they would visit. Avoid treating a character-count pass as proof that the search result is persuasive.

What to improve after access is confirmed

Google describes multiple ranking systems, not a public checklist with fixed weights you can add up. A useful review therefore starts with the search and the page's ability to satisfy it.

For a repair query, check whether the page identifies the repair, service area, relevant limitations and next step. For a comparison query, explain the tradeoffs and the evidence supporting them. Do not pad a page to meet an arbitrary word count.

Review usability separately. Can a visitor read the page on a phone, understand the offer and reach the business? Performance problems deserve investigation with appropriate measurements. Our HTML audit did not measure load speed or Core Web Vitals, so we make no performance claims about the sampled businesses.

How do Google ranking and GEO relate?

Generative engine optimization (GEO) concerns visibility in AI-generated answers. A conventional search position and a citation in an AI answer describe different outcomes. Our homepage audit establishes neither: it measures specific elements in the HTML the sites returned.

For the AI-search side of that work, read our generative engine optimization (GEO) guide. It explains crawler controls, structured business information and how to measure mentions separately from visits.

A practical diagnosis for one underperforming page

1. Choose a specific service query and the page intended to answer it.
2. Inspect the URL in Search Console for retrieval, indexing and canonical information.
3. Read the live page on a phone. Check whether its first screen establishes what the business does and where.
4. Compare the information a customer needs with what the page actually provides.
5. Make the relevant fix, record its date and watch the same page and query group over time.

Treat a subsequent change in impressions as an observation, not automatic proof that one edit caused it. Demand, competitors and other site changes can move at the same time.

For an owner-friendly starting point, read what SEO involves in practice. If you have a page with an unresolved access or content problem, send Lookelo the URL.