Thin prospect websites get filled in from elsewhere

When a prospect's website turns out to be a landing page with almost nothing on it, research now looks elsewhere: their LinkedIn presence, recent company posts, and web search results.

What changed

Prospect research started at the company website and largely ended there. For most companies that works, because most companies describe what they do on their own site.

A meaningful share of good prospects do not. A consultancy that gets all its work by referral has a one page site with a contact form. A company mid rebrand has a holding page. A firm whose real presence is on LinkedIn has a website nobody has touched since 2019. None of these are bad prospects. They are prospects whose website is not where the information lives.

Research used to treat a thin site as a dead end. The lead would be marked as having nothing to work with, and either produce a weak draft or get held back.

Now a thin site triggers a second pass over other sources. The company's LinkedIn presence, recent posts, and general web search results are pulled in, and the draft is written from the combination rather than from the website alone.

How to use it

This runs automatically when the website comes back thin. There is no setting.

What you will see is fewer leads stuck with nothing to say about them, and drafts referencing things that are not on the prospect's website. A post they published, a role they are hiring for, coverage of them somewhere else.

If you are building lead lists, this changes what makes a viable prospect. Companies with weak websites used to be poor candidates for personalised outreach through the product. That is much less true now, which widens the pool of lists worth importing.

Why it matters

The cost of treating a thin website as a dead prospect is not evenly distributed. It falls hardest on smaller companies, on service businesses, and on anyone whose marketing happens on a platform rather than a site of their own. Those are often the best prospects for outreach, because they are reachable and they are not fielding fifty vendor emails a week.

Losing them to a technical limitation of where research looked was an expensive default. It meant the product worked best on companies with the most polished web presence, which is close to the opposite of who is easiest to sell to.

What gets used

The company's LinkedIn presence gives the description they wrote about themselves, their size, and their sector. This is often more current than their website, because it is easier to update.

Recent posts give something time bound and specific, which is the most useful kind of detail for a first email. A company posting about a new hire, an event, or a piece of work is giving you a concrete opening.

Web search fills the remaining gap. Coverage, directory listings, and mentions elsewhere. This is the weakest of the three and is used last, because search results about a small company are the most likely to be about a different company with a similar name.

How the sources are weighed

Not all of it is treated as equally reliable, and that matters for what ends up in the email.

Something the company published about itself is treated as strong. They wrote it, it is current, and it is about them. A recent post is the best single source a draft can have.

Something written about them by someone else is weaker, and gets used only when it can be tied to the right company with confidence. The risk with search results is name collision, and a claim about the wrong company is worse than no claim.

The evidence binding described elsewhere in this changelog still applies to all of it. Enrichment widens where research looks. It does not lower the bar for what can be asserted.

When it still comes up empty

Some prospects have almost no online presence anywhere. A small firm with a placeholder site, no LinkedIn activity, and no coverage is genuinely difficult to write a personalised email to.

Those leads are now handled by a separate change covering first touch emails written from what you offer rather than what was found. Between the two, a thin prospect is no longer a lead that quietly goes nowhere.

Effect on speed and credits

Enrichment runs only when the website comes back thin, so most leads are unaffected and research takes the same time it did.

For the leads that do trigger it, research takes longer, because more sources are being read. This happens in the background and does not delay anything you are waiting on directly. It does not consume additional lead credits.

What a thin site actually looks like

The judgement is about how much usable content the page carries, not how the page looks. A well designed one page site with a clear description of what the company does is not thin. A large site built entirely in a way that leaves no readable text behind is.

The common cases are a holding page during a rebrand, a contact form and a logo, a site that is a single image with no text, and a site built so that its content only appears after scripts run and never resolves into anything readable.

There is also the case of a site that loads but carries no information about the business: a page of stock photography and three words of tagline. Technically there is content. None of it tells you anything you could write an email about, and it is treated as thin for that reason.

When enrichment finds the wrong company

The main risk with searching beyond the company's own website is confusing them with a different company that shares a name. This is common with short or generic company names, and it is the reason web search is weighted lowest of the three sources.

If you see a draft referencing something that clearly belongs to a different business, that is worth reporting with the lead named. Those cases are useful because they usually point at a specific pattern in how the name was matched rather than a one off.

The evidence binding limits the damage here. A claim that cannot be tied confidently to the right company does not make it into the email, which means the usual outcome of a name collision is a plainer draft rather than a wrong one.