What I do every time I find someone selling something like Savia

The textbook version of competitor analysis goes like this: you sit down for a day, look at eight companies, build a nice table and draw conclusions.

I've done it. The problem is that it expires. Three months later the document describes a place nobody is standing in any more, and redoing the whole thing is such a chore that you don't.

What I do now is something else, and it's a good deal less glamorous.

It isn't an exercise, it's a reflex

Every time I come across a company or a person selling something like Savia — an ad, a post, someone showing up on LinkedIn — that triggers the process.

I don't wait until I've collected eight. One is enough.

The advantage is that the work accumulates instead of repeating. After a few months I have the whole map, built in pieces, and each piece is dated to the day that competitor existed exactly as I looked at it.

Step 1: the report

I use the Claude Code extension. I give it that company's site and ask for the same thing every time, in the same format:

1. The main promise. The big line on the homepage, verbatim.
2. Who they say it's for. Company size, sector, role.
3. What's included and what isn't. From their product or pricing page.
4. How they charge. Per user, per usage, licence, custom quote.
5. What proof they show. Case studies, logos, numbers, reviews. Named or unnamed.
6. Which objection they answer first. What sits right at the top is usually their customer's most common fear.

That last point is worth more than it looks. If three competitors give space to explaining where the data is stored, the customer is asking about that before they ask about price.

The instruction that always goes in: direct quote and link, or leave it out. Without that you end up comparing summaries that were invented very confidently.

Having the report come out identical every time isn't fussiness. It's what lets you set them side by side six months later.

Step 2: check it against mine

Here's the part that doesn't appear in any "how to analyse your competition" article, and it's the part that makes this worth doing.

I keep a project folder with all my own material in it: my content, my buyer persona, and the marketing files the agency I hired put together for me. It's what defines what I say, who I say it to, and why.

The competitor report goes in there. And the question I ask isn't "what does this company do".

It's this: given this, what am I doing well and what am I doing badly?

That's a big difference. A normal competitor analysis ends in a portrait of somebody else, which is information that doesn't oblige you to anything. This one ends in a list of corrections to my own work, which does.

Why this beats an afternoon

Because every new competitor is a free outside opinion on my own positioning.

If a company with real customers has decided to put data security in the first block of their homepage, that's a signal somebody else paid money to discover. I get it for free by reading their site carefully.

Stack up enough reports and you end up with feedback from a lot of different people — big companies, solo operators, agencies — without having spoken to any of them. It doesn't replace talking to customers. But it takes twenty minutes and they already paid for being wrong.

What came out of it, in my case

After a handful of reports, something repeated: almost everyone talked about searching faster.

Very few talked about what happens the day the person who knew the answer leaves.

That gap is what ended up being Savia's angle. I didn't find it with a comparison table: I found it because the same silence showed up report after report.

What the AI does and what it doesn't

The AI does what used to eat my time: reading, extracting, and laying it out in the same format.

The deciding part is still mine, and it's the short one. No model is going to tell you which of the gaps it found actually matters to your customer. You learn that by talking to people.

And one rule with no exceptions: no link, no data point. AI is good at structuring and very good at inventing something that sounds perfectly reasonable.

Mistakes I've made

Copying whoever is in front. The leader can afford a generic message because they already have a brand. You can't.

Only looking at the big ones. The small competitor moving fast teaches you more about where the market is going than the big one that hasn't changed in five years.

Filing the report and never coming back. A report that never goes into the folder with my own material is just another PDF. The comparison is the work; the report is only raw material.

What actually changes

It isn't that I know more about my competitors.

It's that when a new one turns up, instead of a jolt of "these people are ahead of me", I have twenty minutes of concrete work and one or two things to fix this week.