Follow the Money

The first net is Google's public advertiser index, searched by name. It works beautifully when the trade is in the name — search "piscine" and pool builders come out one after another, each with the number of ads it is running right now.

It does not work the other way. The index knows the advertiser's name, not what the company sells, so a clever query about an outcome returns nothing. Cast wide on the obvious terms and accept that this pass is about volume, not precision — precision is three steps away and cheap there.

LinkedIn's library is searched the opposite way: not by advertiser name but by the words inside the ads themselves. So the query has to be written the way an advertiser writes, not the way an industry is named — "sponsoring sportif" finds plenty, "lunettes de natation" finds nothing.

Each matching ad's detail page is opened to read the real advertiser behind it, and the impressions band is the dial that separates budgets from noise. Sweeping the impressions ladder turns one query into many, which is how a single search that returns eight companies becomes thirty-six.

Both discovery sources hand over an advertiser name and no website. Everything downstream — verifying the budget, reading the tags, finding the person — needs a live domain, so this is the step that turns a name into a company you can actually work.

It is dull and it is load-bearing. A name that never resolves to a real, live site is not a lead; it is a string, and it drops out here before it costs a single enrichment credit elsewhere.

With a domain in hand, each company is checked back against the ad libraries for a plain verdict: advertising, or absent. The Google check is worldwide, the LinkedIn check is confirmed against the advertiser's own page so a substring match on a common word can't fake a hit.

This is where the pull gets its confidence. A verified "advertising" is a company you can open a performance conversation with today, without guessing whether the budget exists.

The libraries have a blind spot: a brand can be invisible in every one of them and still carry a Meta Pixel, a Google Ads conversion tag, a TikTok Pixel or a LinkedIn Insight Tag on every single page. Reading the served HTML for those tags is what surfaces Profile B — the budget nobody else is pitching.

The reading has limits worth respecting. It sees the HTML as served, without executing it, so a pixel injected by a tag manager can read as organic — but the pass flags when it saw a tag manager, so a false negative announces itself rather than hiding.

Everything so far was proof of budget; this is the split that decides the message. Profile A advertises in the open — the pitch is about the spend they can already see reported: the waste, the plateau, the channel they haven't tried. Profile B is invisible in the libraries but pixelled on every page — the pitch is warmer and less contested, because nobody else knows they qualify.

Send the wrong one and it shows. Open with "we noticed your LinkedIn ads" to a company whose budget only ever surfaced as a pixel and the first line is already wrong. The segment is the opening line.

An explicit gate before any contact enrichment is paid for. Out go the private individuals that slipped the impressions floor, the agencies running ads on behalf of clients you actually want to reach directly, and the domains that resolved to something that isn't a company at all.

As in every recipe the discards are kept and read. If a whole keyword keeps importing agencies instead of advertisers, that is a fix to the discovery query, not a hundred individual rejections downstream.

Only now is a person attached. The contact is whoever is accountable for ad performance — a head of growth, a CMO, or in a small company the founder who still reads the reports themselves. Size decides which: past a certain headcount the founder has handed the channel over, and the email has to follow it.

The point of doing the person last is the same as everywhere else in this catalogue: it is the most expensive step, so it runs on the shortest list — the companies that survived proof of budget, a clean domain, and a segment that matches the pitch.