By training Google Ads on the companies that visit a website, not just the few that fill in a form, we roughly doubled monthly B2B leads for one of our clients, two months in a row. Here’s how we use Leadinfo, a tool that identifies which companies visit your website, to give Google’s algorithm the data it’s missing.
B2B Google Ads: shooting in the dark
Google’s bidding algorithms are clever. But they learn from conversions, and B2B campaigns produce very few of them.
B2B buying cycles are long. The right decision-maker might click your ad today and fill in a form six weeks later, on a different device, via a different channel. Or never fill one in at all, while their company still ends up in your pipeline.
So we never expect someone who clicks through to fill in the form straight away. There are plenty of good prospects on a website who simply haven’t converted yet. That’s perfectly normal. But it does mean you’re working blind.
The result? Google optimises towards whoever converts cheapest, not the companies you actually want to sell to.
The idea: reward Google for finding the right companies
Our theory is simple: use external data to train Google Ads to work better.
Proxy conversions are nothing new. Plenty of PPC teams count a visit to the contact page as a signal of interest, to help new campaigns learn faster.
We use the same trick, but from the outside in. If we want a certain type of company to engage, why not reward Google Ads when it finds them in the first place, even if they’re not ready to fill in a form yet?
That’s where Leadinfo comes in. It identifies which companies visit the website. So instead of waiting for a form, we can tell Google: this click brought in exactly the kind of company we’re after.
How it works, step by step
- Customise the ad URLs. Google Ads adds the Google Click ID (GCLID) to every landing page URL.
- Identify the visiting company. Leadinfo recognises which company is behind the visit and captures the GCLID alongside it.
- Qualify the company. We check whether the visitor fits the client’s ideal customer profile. Universities and colleges, who seem to click on every B2B ad going, are filtered out (unless being targeted).
- Send an offline conversion. The GCLID plus the time of the visit is all Google needs. Good-fit companies go back into Google Ads as an offline conversion, with a value that reflects how good the fit is.
- Let the algorithm learn. Google starts bidding towards clicks that look like those high-value companies.
How strict we make the filter depends on volume. With plenty of data, we only reward exact matches, such as companies above a certain number of employees. With less, we reward companies in the right industry. The algorithm needs some data to guide it, so we use the best we have rather than going all or nothing.
It’ll never be 100% perfect. But it points the algorithm in the right direction far earlier than form fills alone ever could.
The results: leads roughly doubled, two months running
We’ve been running this setup with Leadinfo for a B2B client for several months. There were a few hurdles and mistakes along the way (more on that below). But for the last two months it’s been working well, and in both months the number of leads roughly doubled.
We’ll be upfront: this wasn’t an A/B test, so we can’t prove all of the uplift is down to this approach. Some of it might have happened anyway.
But we’ve seen this pattern before. In an earlier campaign for the same company, using a different (and frankly less reliable) company identification tool, the same approach coincided with a lead increase of around 500%. And across our other B2B accounts, campaigns without this kind of signal have been much harder to get going.
The algorithms are clever, but they need data. This is one of the data sources they need.
Proving the spend is worth it
Training the algorithm is only half the story. The other half is showing the client’s leadership that the ad budget pays off. And B2B attribution is notoriously tricky: people switch devices, work from home, or simply forward a link to a colleague.
So we combine form data, Google Click IDs and Leadinfo company identification into a traffic-light model with four levels of evidence. (Yes, we know traffic lights don’t have four lights.)
Level 3 covers a very real scenario. Someone clicks the ad, likes what they see, and shouts across the office: “Oi, this is the one, go and fill in the form.” The tracking is lost, but Leadinfo still shows the company arrived via an ad.
Level 4 matters because the same landing pages also feed our remarketing. A company identified there is being influenced by the campaign, even if it never clicked.
It’s soft attribution, and we’re clear about that. But it helps marketing directors show the board that the budget is working. And any company spending money on ads wants exactly that kind of evidence.
What we learned along the way
- Campaign type matters. Demand Gen brings in lots of mobile traffic, which rarely produces identifiable company visits, and it isn’t built for lead generation anyway. With the same message, Demand Gen alone delivered almost no leads. Performance Max, trained on company-visit conversions, delivered many more.
- Use the best data you have. Start with broad filters, such as industry, and tighten them as volume grows.
- Track where it counts. We only track the pages the campaign sends traffic to. That keeps the signal clean and the setup focused.
- Give it time. The algorithm needs a steady flow of conversions before it starts prioritising the right companies.
What’s next
We’re turning this approach into a dedicated service for clients who need leads, not just reach. The same setup would work just as well on Microsoft Ads. On LinkedIn it matters less, as the platform already knows which company a user works for.
We believe this is the way forward for B2B paid search. And it’s working.
Want more B2B leads from Google Ads?
If your Google Ads budget is bringing in clicks but not the right companies, let’s talk. Get in touch with our PPC team to see what this approach could do for your campaigns.
