Analysing Five Competitors in One Afternoon: The Complete Method with Semalt

In short: competitor analysis usually fails for one of two reasons — it never starts because it sounds enormous, or it produces a spreadsheet nobody acts on. This is a four-hour method with a decision at the end of each hour. The tracking and analysis side is free to open at semalt.com/authorize.

The purpose is not to admire competitors. It is to find the specific gaps you can occupy in the next quarter, and to stop wasting effort on ground that is already lost.

First, pick the right five

Most people list the five businesses they think about at conferences. Those are usually the wrong five. What matters is who occupies the search results you want.

Take the ten queries that genuinely produce customers for you — not the vanity term, the queries a buyer types — and record which sites appear in the top ten for each. Sites that appear repeatedly are your search competitors, whether or not you consider them rivals commercially. Expect surprises: a directory, a marketplace or a trade publication often outranks every actual business, and that changes your strategy more than any competitor does.

Pick five: two obvious direct rivals, two sites that keep appearing across many of your queries, and one that is clearly smaller than you but ranks anyway. The last one teaches the most, because whatever it is doing is achievable.

Hour one: where they beat you, and by how much

For each competitor, answer three questions and nothing else:

• Which of my target queries do they rank for that I do not? This is the gap list, and it is the only output of the hour that matters.
• Which page of theirs takes each one? Note whether it is a homepage, a service page, or an article — the answer tells you what kind of page the engine wants for that query.
• How far ahead are they? Position three is a different problem from position one.

Decision at the end of the hour: cross out every gap where the competitor is a national brand, a marketplace or a publication. You will not take those, and pretending otherwise consumes the whole quarter.

Hour two: read the pages, do not scan them

Take the five or six pages that beat you on queries still on your list after the cull, and actually read them.

Look for what the page does that yours does not:

• Does it answer a sub-question yours skips — price, timescale, what happens if something goes wrong?
• Is it structured for scanning, with the answer near the top rather than after four paragraphs of introduction?
• Does it show evidence — a worked example, a photograph of real work, a named author — where yours asserts?
• Does it cover a stage of the decision you have no page for at all?

Decision: for each page, write one sentence — "to beat this I must add X". If you cannot write that sentence, the query goes off the list.

Hour three: check the boring structural things

Now the mechanical comparison, which takes minutes per site and occasionally explains everything.

• Page speed on mobile, tested on a throttled connection. If they load in two seconds and you take six, that gap is doing work no content can undo.
• Site structure. How many clicks from their homepage to the page that beats you? If theirs is one and yours is four, your internal linking is the problem.
• Page inventory. Do they have thirty service pages where you have one? That is not a content advantage, it is an architecture advantage, and it is copyable.
• Local signals, where relevant: category choice, review count and recency, whether replies exist.

Decision: separate the findings into "fix this month" and "requires a project". Be ruthless — most sites have one structural defect worth more than everything else on the list.

Hour four: turn it into a plan you will actually execute

Anything that survives three hours of culling goes into one of three buckets:

Take now. Queries where a competitor ranks with a page clearly weaker than what you could publish this month. Usually three to six of these. Start here — they pay back fastest.

Build toward. Queries needing a page type you do not have — a comparison, a pricing explanation, a proper case study. Schedule them across the quarter.

Concede. Everything else, written down explicitly so nobody reopens the argument in six weeks.

A four-hour analysis that ends with six pages to write beats a two-week analysis that ends with a spreadsheet.

What to repeat, and how often

Full analysis twice a year is plenty. In between, watch position movement on your ten queries: a competitor climbing steadily on several at once has done something structural worth a look, while single-query noise means nothing.