How to do competitor research, properly, in one afternoon

Most competitor research fails in the same way. Somebody spends two days building a spreadsheet, presents it, everyone nods, and nothing changes.

This method is built backwards from that. The output is three decisions. Everything you collect exists to produce them.

Before you start: what question are you answering

Write it down in one sentence. Genuinely one of these, not all of them.

• Why are we losing deals to them?
• Where is the demand they are capturing that we are not?
• Are we priced wrong?
• What should we publish next quarter?
• Are we invisible somewhere they are visible?

The research is the same either way. The filtering is not. Without a question you will collect everything and conclude nothing.

Step 1: pick the right competitors, 30 minutes

Not the ones you talk about internally. Those are usually the ones you admire or resent.

Three sources:

Sales. Ask whoever takes calls which names come up. Write down the top three. This list is the most accurate one you will get and it is free.

The search results. Search your five most commercial terms in an incognito window. Who is there? Some will be competitors you had not considered, some will be directories and listicles, which matter in their own way.

The assistants. Ask ChatGPT and Perplexity to recommend a company in your category and location. Write down every name that comes back. This list is often different from the search one, and it is the one your buyers are increasingly seeing.

Pick three to five. Note which of those three lists each came from, because a company that is strong in AI answers but invisible in search is a different threat from the reverse.

Step 2: what they sell and for how much, 45 minutes

For each competitor, save:

• Their pricing page as a screenshot, with today's date in the filename.
• Their tiers, what is included, and what changed recently. Check the Wayback Machine for last year's version, which takes two minutes and is often the most revealing thing you will find.
• Their positioning sentence, taken from the homepage hero, word for word.

That last one matters more than people expect. Write the three competitor sentences and your own under each other. If they are interchangeable, you have found a problem that no amount of channel work will fix.

Step 3: what they publish, 45 minutes

Go to their domain and add /sitemap.xml. You get their complete page list, usually with last modified dates.

Save the file. Note:

• How many pages total.
• How many are blog or resource pages.
• How many carry a lastmod date in the past 90 days.
• Which page types repeat. Fifty pages of the shape service plus city tells you their whole SEO strategy in one glance.

If you do nothing else in this guide, do this. It is their own published index of everything they have, and almost nobody looks.

Step 4: what they advertise, 45 minutes

Three free official libraries:

• Google Ads Transparency Center. Search their domain. Every ad, with first and last shown dates.
• Meta Ad Library. Search their page. Every live Facebook and Instagram ad.
• TikTok Commercial Content Library. Same for TikTok.

Screenshot the ads. Note how many are running, how long the oldest has been live, and where the ads point.

The landing pages are the valuable part. An ad that has been running six months is working. The page behind it is what they believe converts, tested with real money. That is free research.

Step 5: what people say, 30 minutes

Their Google reviews, their G2 or Capterra profile, their Trustpilot.

Read the three star reviews specifically. Five stars are marketing and one star is usually a delivery incident. Three stars say things like it works but the reporting is limited, which is a positioning gap you can use tomorrow.

Also search Reddit for their brand name and for their domain with the url: operator. What people say when the company is not in the room.

Step 6: whether AI recommends them, 30 minutes

Write ten questions a buyer would ask. No brand names. Ask each in ChatGPT and Perplexity with web search on.

Record two columns: who was named, and what was cited.

The second column is the useful one. Those cited sources are the pages deciding your category's shortlist, and getting into them is usually faster than trying to outrank anybody.

Step 7: the part that matters, 30 minutes

Close the tabs. Open a blank page. Write:

Three things they do better than us. Specific, not vague. Not "better brand". Something like: they publish twice a week and we publish monthly.

Three things we do better. Same standard.

Three decisions. What you will change, who owns it, and by when.

If you cannot write the three decisions, the research did not work and adding more data will not help.

Keeping it current

The first pass is the easy part. The failure is always month three.

Options, honestly:

Manual, ninety minutes a month. Re download their sitemaps and diff. Re check the ad libraries. Re run the AI questions. Works perfectly and gets skipped by month three in most companies, which is not a discipline problem, it is what happens to any recurring task that is nobody's actual job.

Free alerts. Google Alerts on their brand names, a change monitor on their pricing pages. Costs nothing and catches some of it.

A tool. From $12 a month for page watching up to $49 for something that checks the whole set weekly and writes you a summary. Ours is Figo and it does the steps 3 to 6 loop every week, but the honest point is that any tool that runs without you is better than a perfect manual process that stops.

Do the manual pass first either way. You cannot tell whether a tool is showing you the right things until you have done it once yourself.