How to read a search results page before you write anything

The results page for a query is the most useful research document available about that query, and it is free.

It tells you what Google currently believes the query means, which kinds of pages satisfy it, how strong the incumbents are, and whether there is a gap. Most people skip it and go straight to a difficulty score in a tool — a number produced by a model, from sampled data, by a company selling subscriptions. The page itself is primary evidence.

This is a method. Ten minutes, before you write, in a fixed order so nothing is skipped.

Set up so you see the real page

Two adjustments first, because the default view is misleading.

Use a private window and sign out. Signed in, you are seeing results shaped by your own history — including your own site, which you visit constantly and which will appear higher for you than for anyone else.

Know where you are searching from. Location changes results, dramatically for anything local. If you sell in a city you do not live in, the page you are looking at is not the page your customers see.

Neither takes long, and skipping them produces confident conclusions from a page nobody else is looking at.

1. Count the page types

Before reading anything, categorise the ten results by *kind*:

• editorial articles
• product or service pages
• category or listing pages
• forums and community threads
• videos
• tools or calculators
• directories and marketplaces

Write down the counts. This single number is the most decisive thing on the page.

If seven results are category pages on retailers and you intend to write an article, you are competing against a page type Google currently prefers for this query. That is not forbidden, but it should be a decision, not a discovery six weeks later. The reasoning is set out in search intent.

2. Look at what is *not* a result

The features occupying space tell you as much as the links.

| What you see | What it means for you | |---|---| | Map results | Local intent. A web page cannot win this; a Business Profile might | | Shopping listings | Transactional. Articles will struggle | | A featured snippet | A short answer exists and can be captured — see who holds it | | People also ask | Adjacent questions worth covering, and a hint at the subject's shape | | Video block | The audience prefers watching. A text page may be second-best by default | | Many ads | Commercially valuable. Also means the organic slots sit lower | | Almost no ads | Little commercial value. Often easier, often worth less |

A page where the first screen is entirely map results and ads has perhaps two organic positions above the fold. Ranking fourth there is worth much less than ranking fourth on a clean page, and the arithmetic should account for it.

3. Judge the incumbents honestly

Now read who is there. For each of the top five, ask three things:

Is this site about this subject, or is this a stray page on a huge site? A specialist site ranking is a stronger signal of difficulty than a general publisher ranking with one article. Large sites rank for many things loosely; that does not mean their page is good.

Is the page actually good? Open it. Genuinely read it. A remarkable proportion of first-page results are thin, outdated, or answer a slightly different question. If three of the top five are poor, that is your opening — and you will only ever find it by opening them.

When was it last meaningfully updated? Not the displayed date, which is often automated. Look at whether the content refers to things that have changed.

4. Check whether the page shape is consistent

Within the dominant type, look for a repeated structure. Are they all lists of ten? All step-by-step? All comparison tables? All short direct answers?

Consistency across results is Google reporting what this audience repeatedly chooses. It is information, not an instruction — you may deliberately differ — but differing without noticing is how good writing ends up in the wrong shape.

5. Ask the question that decides it

After the four passes above:

Can I produce something clearly better than the weakest page currently in the top five, for the same intent?

Not better than the best. Better than the *weakest*, because that is the position you would be taking. If the answer is no, the honest move is a different query — usually a more specific one where your actual knowledge is the differentiator.

If the answer is yes, you also know *how*, because you have just read what the incumbents failed to do.

A worked example

Illustrative. A firm fitting home EV chargers considers targeting "how much does it cost to install an ev charger".

• Types: four editorial articles, three energy-supplier pages, one government-adjacent information page, two installer pages.
• Features: a featured snippet held by a supplier, a People Also Ask block, moderate ads.
• Incumbents: two of the articles are national publishers with generic ranges. The installer pages are thin and mostly a quote form.
• Shape: most lead with a price range, then break down what changes it.
• Freshness: prices referenced are visibly out of date on two of them.

Conclusions: the intent is informational-with-commercial-edge, satisfied by a price breakdown. The weakest top-five pages are the generic national ranges. The firm has something none of them do: actual current prices from jobs it has completed. That is a genuine advantage, and the page should lead with a real range and what moves it.

Ten minutes produced a specific brief and a reason to believe it can win. That is the return on this method.

A second example, where the answer is no

The method is only worth anything if it sometimes says stop, so here is one that does.

The same firm considers "ev charger installation".

• Types: three national installer chains, two energy suppliers, a comparison site, a marketplace, the manufacturer of a popular charger, and two paid results above everything.
• Features: a large ad block, a shopping-style module, no featured snippet, no People Also Ask worth speaking of.
• Incumbents: all commercial, all national, all with proper service pages and long histories.
• Shape: completely consistent — every organic result is a service page selling installation.
• Freshness: irrelevant; nothing here is dated content.

Conclusions: the intent is transactional, the field is entirely national commercial pages, and nothing in the top ten is weak. A local installer's service page has no realistic route into this within a year, and an *article* would not belong here at all — the format is wrong before the competition is even considered.

That is a ten-minute save of what would otherwise have been a month. And it points somewhere useful: the firm's route into this market is the informational query in the previous example, plus the map results for its own area, not this term.

Reading the results page for a commercial query

The five steps hold, but three things read differently when the intent is transactional rather than informational.

The ad block is information, not noise. Heavy advertising on a query means someone is paying to be there, repeatedly, which is a strong signal that the traffic converts. It also means the organic results sit lower on the screen than they appear in a rank report.

Consistency matters more. On an informational query, a mixed results page is an opening. On a commercial one, ten results of the same type usually means the format is settled and there is no gap to enter through.

"Weak incumbent" means something different. A thin article is genuinely weak. A thin service page from a company with a decade of history and national brand recognition is not weak in the way that matters — the page is thin because it does not need to be thick.

The practical upshot: informational queries are where a small site finds openings, and commercial ones are usually won on standing rather than on the page. Which is why the route in is normally the former, feeding the latter.

What this replaces

It replaces trusting a difficulty score. Those scores are estimates built by third parties on sampled link data and modelling — useful as rough comparators across many keywords, unreliable as a verdict on one. They cannot tell you that three of the top five pages are bad, which is the fact that most often decides whether you should proceed.

Use scores to shortlist. Use the page to decide. More on that in why every competitor traffic estimate is wrong and judging whether you can realistically win a keyword.

When the page tells you to walk away

Sometimes the honest reading is that this query is not available to you:

• Every result is a large marketplace, directory or retailer
• The intent is transactional and you have no product page that fits
• The page is dominated by institutional sources on a subject where trust is the barrier
• Every incumbent is genuinely excellent and deeply resourced

That is a useful outcome, not a wasted ten minutes. It redirects effort to a query you can win, which is worth far more than a good article in an unwinnable position.

If the whole subject reads that way — the queries your customers use are structurally held by platforms — then the realistic routes are a narrower query, or being on one of those platforms. We run one, so weigh that accordingly: the trade-offs are in when renting attention beats owning it, and the prices are on the rankings board.

Common questions

How many results should I read? Five properly beats ten superficially.

Does this work for every query? It works best for queries with commercial or informational intent. For navigational ones there is nothing to analyse.

Should I check more than one phrasing? Yes. Two or three variants of the same question often return noticeably different pages, and the differences are informative.

How often should I redo this? For pages that matter, annually, and immediately if a page loses position for no visible reason.