ICP & Buying Signals ~1,300 words · 6 min read

Competitor Analysis for ICP — Why Current Customer Lists Mislead You

By Anna Orestova · Founder, LamparaLab · August 8, 2026

TL;DR: Building an ICP from competitors' customer pages gives you the buyers they can win today, not the ones you can reach now. Across 135 customer records from 15 B2B sellers, archived versions of the same sites told a different story: the early customers were smaller, closer to the seller, and won long before the proof that attracts today's logos. Which part of a competitor's history is worth copying depends on your stage, not their product. Rewind the customer page in the Wayback Machine and you get a second ICP — a segment that appears nowhere on the current site. Here's how to reconstruct it.

Everyone builds an ICP off competitors at some point:

Launching without closed-won data
Putting a new product into an established company
Entering a new segment
Moving upmarket
Or just checking whether the market looks like your pipeline says it does

The move is always the same: collect customer logos from competitors' websites, enrich the companies, look for patterns in industry, size, funding, geography. Clean clusters come out. Looks like an ICP.

It tells you who your competitors sell to today. It doesn't tell you whether those buyers are reachable from where you stand.

Competitor Analysis Shows Who Buys Now, Not How They Were Won

I analyzed 135 customer records across 15 B2B sellers. First pass: current logos, grouped by firmographics. Clear clusters, convincing ICP.

Then I opened archived versions of the same sites.

The customer lists weren't just shorter. Different companies entirely. The early ones were smaller, less recognizable, and usually connected to the seller somehow — same region, same professional community, a previous employer, an existing relationship.

The names on the site today showed up later. After results, references, case studies.

Same website. Two different answers to “who buys this?”

Which customers you can copy depends on your stage, not their product

A similar product gets a competitor onto your list. It doesn't make their buyers available to you.

What decides that is stage — where they were when they won those customers, and where you are now. Not calendar age: how much proof each of you had at the moment of the sale.

Sometimes the stages match. An established company moving into a new category already carries brand trust, buyer relationships, and distribution over from the business it's in today. Its competitors' current buyers may be reachable on day one. Nothing to rewind — go take them.

Most sellers aren't there. A two-year-old company reading a category leader's customer page is looking at buyers who took a decade of proof to earn.

Their current ICP is noise. The segment they sold to before that proof existed is signal.

And it isn't only a launch problem. Enter a new segment or move upmarket, and you're back to an early stage in that market no matter how long you've been selling elsewhere.

So: product similarity gets them on the list. Stage decides which part of their history is worth reading.

How to reconstruct a competitor's early customers with the Wayback Machine

It's all public.

1
Record who's on the customer page today.
Logos, case studies, named industries.
2
Pull the archive.
Drop the site URL into the Wayback Machine and open older snapshots. What's captured varies — the homepage almost always is, inner pages depend on what the crawler picked up that year.
3
Date each customer.
Which snapshot did each one first appear in?
4
Read the positioning next to them.
What could the seller prove at that moment — which results, which industries, which names?
5
Watch what came next.
Which segment showed up after that proof landed?

The difference it makes. Static conclusion:

Most customers are Series A B2B SaaS companies.

Chronological conclusion:

Series A customers started appearing after the seller published two projects with smaller SaaS companies in the same use case.

The first is their ICP. The second is a hypothesis about what it cost to get there.

A second ICP that isn't on anyone's website

Here's what made the hours worth it.

The current logos gave me one ICP — the firmographic clusters anyone can pull in an afternoon.

Lining the early customers up against the positioning that was live at the time gave me another: a consistent industry thread running through the earliest wins, and a visible shift in how the sellers described themselves as they moved toward the buyers they have now.

That's a second ICP. It appears nowhere on the current website.

Not a better one. A different one. Another segment to test, invisible to anyone reading only today's logos.

Your closed-won deals still outrank anything you find

If you have closed-won deals

Start there. Your own deals show which companies actually bought, which patterns already produce revenue, what was happening when they entered the pipeline. No competitor's website beats that. The full method for deriving an ICP from won deals rather than a workshop is in An Ideal Customer Profile Framework for B2B SaaS.

Dig further while you're in there: why did those buyers need the product then? What had just changed for them? That's the layer firmographics never shows — and it's what we build ICPs from at Lampara, because a buying situation you can watch for is worth more than a company type you can filter on.

Then hold the competitor hypothesis next to it. Overlaps with your closed-won pattern? Confirmation from two directions. Contradicts it? Cheap to test — you may be sitting on a segment you've simply never targeted.

If you don't have closed-won data yet

Then this is your starting hypothesis. Not a great one — but grounded in how comparable sellers actually entered, which beats a definition invented in a doc.

Compare their starting advantages against yours: expertise, relationships, distribution, reputation, access to a community. Sometimes their current ICP turns out to be within reach. More often you'll find the earlier, lower-proof segment — the one that produces the case studies that unlock everything after it.

💡

Then go test it. It's a hypothesis, not an ICP.

Competitor Analysis Gives You a Draft ICP, Not a Final One

The customer page shows who buys today. The history shows what made those buyers reachable, and from what starting position.

Read the logos and you have one hypothesis. Rewind and you have two. Neither is your ICP until the market says so — and if you'd rather test both against real buyers than sit on them, that's what we do: build the ICP and run the outbound that proves it.

Common questions

How do you find your competitor's customers?

Customer pages, case-study indexes, testimonials, review sites, and press releases cover the current ones. For earlier customers, work back through archived snapshots of their site in the Wayback Machine — whatever was captured — plus launch announcements, founder posts, and podcast interviews to date when each one appeared.

Can you copy a competitor's ICP?

Only if you're at the same stage they were when they won those customers. A similar product doesn't make their buyers reachable — the proof, references, and distribution behind those wins do. Copy the ICP of a seller ten years ahead of you and you're targeting buyers who won't take the call.

Can you use competitor analysis to define your ICP?

As market evidence, yes. As a finished ICP, no. The customer list tells you which companies buy this category. It doesn't tell you what proof and distribution the seller needed to win them — which is what decides whether you can win them too.

How do you use the Wayback Machine to research competitors?

Enter their site URL and work back through the archived snapshots. Coverage varies by year and by page, so take what's there — usually the homepage, often the customer page. Date each customer by the snapshot it first appears in, then read the positioning that was live alongside it. The sequence shows which segment they started with and which proof unlocked the next one.

How do you choose competitors for ICP research?

Product similarity gets them on the list. Stage decides which part of their history is relevant to you. A category leader ten years ahead is worth rewinding, not copying.

How do you define an ICP with no customers or closed-won data?

Build hypotheses instead of conclusions — from competitor timelines, early pilots, founder expertise, existing relationships, customer interviews. Then test them against the market rather than committing to one.