Reading a SERP: A Walkthrough

Experienced SEOs open a search results page and know within twenty seconds whether a keyword is worth writing. Nobody ever writes down how. This page takes one query apart element by element, and ends with a checklist you can run in two minutes.

The query, and who is asking

Say you run a four-van appliance repair company. You fix washing machines and dishwashers in people's kitchens, you charge a call-out fee, and you have a blog with nine posts on it that somebody set up in 2021 and abandoned.

The keyword on the table is why is my dishwasher not draining.

Your instinct says no. It is a free-advice query, the people searching it are trying to avoid calling you, and there is no obvious money in it. Hold that thought, because the results page is about to disagree.

Start at the top: the ads

Two or three paid results sit above everything, and they are from national repair chains and home-warranty companies bidding on a question about a broken appliance.

That single fact is worth more than any keyword tool's estimate.

Somebody with a budget and a spreadsheet has decided that people searching this phrase are worth paying for. They are not paying for readers. They are paying because a meaningful share of people whose dishwasher is full of standing water end up booking an engineer within 48 hours.

So the query is informational in form and commercial in consequence, which is the most underrated combination in SEO. Ads at the top of a how-to query are a signal that the audience converts downstream.

Read it this way: no ads at all on a question query usually means genuine hobbyist traffic with no commercial tail. Ads from your actual competitors means the traffic buys something.

The local pack

Below the ads, a map with three business listings and a review count each.

A local pack tells you two things at once. Google believes a meaningful share of these searchers want somebody to come to their house, and there is a whole ranking surface here that has almost nothing to do with your blog post.

You cannot win the local pack with an article. You win it with a Google Business Profile, service-area settings, and reviews.

It still matters to your decision, though, because it confirms the money and it compresses the page. Every feature above the ten blue links pushes the first organic result further down, which lowers the click-through rate of position 1 before you have written a word.

The featured snippet

A boxed answer near the top, probably a numbered list, probably pulled from a manufacturer support page or a large home-repair site. Something like: check the filter, check the drain hose for kinks, check the air gap, check the garbage disposal knockout plug.

Featured snippets are lifted from a page that is already ranking, usually in the top ten, and they reward one specific writing behavior: answering the question in a compact, self-contained block near the top of the page.

A snippet that is a numbered list is telling you the format it wants. A snippet that is one paragraph of 40 to 50 words is telling you to write a 40-to-50-word definition high on the page and then explain underneath it.

There is a catch worth stating plainly. Winning a snippet on a query like this can reduce your clicks, because the reader gets the four checks and never needs your page. The snippet is worth chasing when your answer needs the visit to be useful, and worth ignoring when it does not.

People Also Ask

Four expandable questions, and they expand into more as you click. On this query you would expect roughly: how do you force a dishwasher to drain, what is the most common cause of a dishwasher not draining, can I fix a dishwasher that won't drain myself, how much does it cost to fix a dishwasher that won't drain.

This block is the single most useful thing on the page and most people scroll past it.

It is a free, continuously updated list of what real people ask next, phrased the way they phrase it. Three of those four are subheadings for your article. The fourth one, the cost question, is a separate article and probably a better commercial bet than the one you started with.

Video, and what it means when it shows up high

A row of three video thumbnails, probably somewhere around position four or five.

A video block sitting that high means a substantial share of searchers wanted to be shown, not told. That is normal for anything physical: a hose behind an appliance, a filter that twists out, a part that looks like three other parts.

The practical read for a text article is not "give up". It is that your article needs images of the actual components, clearly labelled, or it will lose to people who did the showing.

Where a video block should worry you is when it appears above the first organic result and occupies half the screen. At that point the query has effectively become a video query, and a written page is competing for what is left.

The ten blue links: who is actually here

Now the part that decides everything. Ignore the titles for a moment and classify the results by what kind of site each one is.

What you see	What it tells you
Manufacturer support pages (Bosch, Whirlpool)	Strong domains, but generic pages that answer for their brand only. Beatable on depth and neutrality.
Two national repair chains	Well-resourced competitors. Read their pages: they are often thin, written to book a call rather than to answer.
A large home-improvement publisher	Hard to outrank on authority. Almost always beatable on specificity, because they wrote it once for everyone.
A Reddit thread and a DIY forum	The opening. Google put user-generated content on page one because it could not find enough good pages.
One independent repair company's blog post	The proof. A business exactly your size is already there, which means the door is not locked.

That last row is the one to look for on every SERP you assess. A page from a site with no more authority than yours, ranking in the top ten, is direct evidence that the query is winnable by a site with no more authority than yours.

The forum results are the second-best evidence. Reddit on page one is Google telling you it could not find a good enough article.

How long are the top results?

Open the top three and get a rough word count. Select all, paste into a word counter, or use any reader-mode extension that reports one. You are not looking for a number to hit.

You are looking for the shape of the answer, and there are only three outcomes.

• They are all short, 600 to 900 words. The question has a compact answer and padding will hurt you. Write the compact answer better, with clearer images and one thing the others left out.
• They are all long, 2,500 words and up. The query is a container for many sub-questions, and a 700-word post will read as incomplete next to them. Match the coverage, not the word count.
• They are wildly inconsistent. This is the best case. Google has no settled format, which means the SERP is unstable and a genuinely good page can jump in.

Word count is a symptom of coverage, never a target. A 3,000-word article that answers the same six things as a 900-word one is just slower.

What this particular page adds up to

Put the evidence side by side and the verdict writes itself.

Paid competition says commercial value. A local pack says the buyers are local, which is exactly who you are. A forum result and a small independent blog in the top ten say the field is soft. Ten PAA questions say there is a whole cluster here, not one post.

Against that: a featured snippet you may not want, a video block that means you need real photographs, and a page so crowded with features that position 3 will not perform like position 3 used to.

The decision is yes, and the article is not the one you were going to write. It is a diagnostic walkthrough with photographs of the actual parts, the four checks in a compact block near the top, a section answering each PAA question, and a paragraph that says plainly when to stop and call somebody.

The checklist

Two minutes per keyword, in this order.

1. Set your location and go incognito. A SERP read from the wrong country is worse than no read at all.
2. Count the features above the first organic result. Ads, snippet, AI answer, map, video. The more there are, the less position 1 is worth.
3. Ask who is bidding. Advertisers on an informational query mean the traffic converts eventually.
4. Classify the ten results by site type. Brands, publishers, forums, small independents. Write the tally down.
5. Find the weakest thing in the top ten. A forum thread, a 2018 post, a page that does not really answer. That is your entry point.
6. Look for somebody your size. If a peer is ranking, the query is winnable. If the whole page is national brands, pick a longer query.
7. Read the dominant format. Lists, comparison tables, step-by-steps, tools. This is not a stylistic choice you get to make.
8. Harvest the PAA questions. Expand three, read the ones that appear, and keep them. They are your subheadings and your next four articles.
9. Spot-check length on the top three. Short, long, or inconsistent, and nothing more precise than that.
10. Write the one-line verdict. "Winnable, needs photos, cluster of five." If you cannot write that line, you have not finished reading the page.

How this works in RankSpiral

Everything above is a manual version of what the authority map build does per keyword, which is why the two agree so often.

During a build, RankSpiral pulls a live results page for each keyword and keeps a snapshot: the top ten organic URLs, the top thirty domains, the People Also Ask questions, and four features (AI Overview, featured snippet, video, shopping). Snapshots are reused for 30 days.

Three decisions come out of that snapshot, and none of them are made by the language model.

• Clustering is decided by shared URLs. If two keywords share four or more of their top ten results, Google is treating them as the same page and they merge as variants. Two or three shared results means same cluster, different articles. One or none means separate topics. The model only names the clusters and arranges the pillars afterwards.
• Winnability is computed from evidence, not guessed. It starts as a band from keyword difficulty, then moves: up one when two or more of the top ten are forum or user-generated domains, down one when an AI Overview is present on a non-transactional query, down one when the results page returns no organic results at all. If your own site already ranks between 11 and 30, it is pinned to the top and flagged as a quick win.
• Real PAA questions beat invented ones. The questions captured from that live SERP are merged into the article's research, ahead of the questions the research model came up with, capped at eight.

Two honest limits. Keywords where you already rank in the top ten are dropped from the map on purpose, because a second article there is cannibalization rather than opportunity. And winnability is not a column you can see in the app; it sets the order keywords appear in within a cluster, so the top of each cluster is the "start here".

Topical Map→ How Rankspiral Decides What to Write Next→

Going deeper

• Personalization is smaller than people think, but location is not. Incognito removes most account personalization; it does not change the country Google thinks you are in. Set that explicitly if you serve a market you do not live in.
• Check mobile separately for anything local. The map pack and the ads take a far larger share of a phone screen, and for service queries most of the searches are on phones.
• An AI answer at the top changes the maths, mostly for definitions. Queries whose whole value was a one-line fact lose clicks to it. Queries that need judgment, comparison, or a decision hold up better, and being cited in the answer is now its own goal.
• Read the dates, not just the titles. A top ten where four results are three years old is an invitation. Nothing you can do about a competitor updating theirs next week, but you get a head start.
• Zero organic results is not a signal, it is a failure. If a search returns nothing meaningful, treat it as a broken measurement rather than as an empty field waiting for you.
• Where this method breaks: for brand-new or very rare queries there is no settled SERP to read, and the ten results are Google guessing too. Fall back on judgment about what the asker actually wants.

Every keyword tool in existence is an attempt to summarise this page into a number.

The page itself is free, updated continuously, and takes two minutes.

Search Intent: The Four Types→ Keyword Clusters and Pillars, Explained→ SEO Terms, Explained→