Offline conversions and CAPI for B2B companies: why your campaigns optimise for the form and not for the sale
What offline conversions and CAPI are, what a B2B company loses without them, and how the signal is sent from the CRM to Google and Meta.
Google and Meta do not know what a sale is. They know what they are told. And most B2B companies tell them only one thing: that somebody submitted a form.
From there, the algorithms do their job. They go looking for more people who submit forms. Students, competitors, browsers, job applicants. All of them count exactly as much as a head of procurement with an approved budget.
Offline conversions and CAPI exist to fix this. This article explains what they are, what they recover, and how they are implemented in a B2B company.
Two different problems with the same solution
Problem 1: the platform learns from the wrong signal
A campaign optimised for “form submitted” gets better over time. It brings more forms for the same money. But there is no guarantee whatsoever that it brings more sales. Often it brings fewer, because the algorithm has found a cheap segment that fills in forms and does not buy.
Problem 2: the platform cannot see part of what happens
The browser pixel loses conversions. Ad blockers, iOS restrictions, browsers that clear cookies, consent banners that get rejected. Each one takes a slice. The platform thinks the campaign is performing worse than it is, and allocates budget accordingly.
The fix for both is the same: send the conversion from the server, with the data from the CRM, instead of from the browser.
What they actually are
Offline conversions (Google Ads). A mechanism by which the company tells Google, after the fact, which clicks ended in a qualified lead or a sale. Google stores a click identifier (GCLID). The CRM keeps it alongside the lead. When the lead changes status, the identifier is sent back with the new status and the value. Google assigns it to the campaign, ad group and keyword that generated it.
Conversions API, or CAPI (Meta). The equivalent on Meta. Instead of relying on the pixel, the company sends events from its own server. It can send the form, but also “qualified lead” or “purchase”, days or weeks later, with hashed contact details so Meta can match them to the user who saw the ad.
Both allow the same thing: telling the platform which leads were worth having.
What they recover, in numbers
Industry data is consistent on this:
| Effect | Typical range |
|---|---|
| Conversions recovered with server-side sending / CAPI | 20–30% |
| Conversions recovered in lead gen with iOS traffic | 30–40% |
| Match rate (conversions matched to users) before | 35–45% |
| Match rate after | 75–92% |
Match rate is the least discussed part and the most important. If Meta only matches 40% of conversions to a real user, it learns with 40% of the information. If it matches 85%, it learns with twice as much.
Why it matters more in B2B than in e-commerce
In an online shop, the purchase happens in the browser, minutes after the click. The pixel sees it. The problem is data loss, not definition.
In B2B, the purchase happens in a meeting, an email or a signature, weeks later. The pixel never sees it. The only way for the platform to know there was a sale is for somebody to tell it, from the CRM.
Without offline conversions, a B2B company is asking Google and Meta to optimise towards the point furthest from the sale. It is like judging a salesperson by the calls they receive rather than the contracts they sign.
What it takes to implement
It is not an afternoon’s configuration. There are five pieces, and all of them have to work.
1. Capture the click identifier
The form has to store the GCLID (Google), the fbclid or the contact details (Meta) and the source campaign. If the form does not do that, there is nothing to send later.
2. Keep it in the CRM
The identifier travels with the lead. If the CRM loses it in an import or a stage change, the chain breaks. Nexus keeps it natively, both with HubSpot and Zoho and with its own built-in CRM.
3. Decide what gets sent
This is the business decision. Do you send “qualified lead”? “Meeting held”? “Sale”? With a value? Each company decides based on its volume and its cycle. In many cases two signals are sent: qualified (more volume, faster learning) and sale (less volume, more value).
4. Send the signal
When the lead changes status in the CRM, it is sent to Google and Meta with its identifier, its date and its value. This has to happen without anyone doing it by hand.
5. Change what the campaigns optimise for
Once the signal arrives with enough volume, the campaigns are reconfigured to optimise towards it. This is the part most people forget. Sending the signal and carrying on optimising for form fills changes nothing.
What happens afterwards
The first few weeks are a learning period. The platform is getting a new signal and needs time to build volume. After that, cost per lead usually goes up. That is expected: the campaign has stopped hunting for cheap forms and started hunting for buyers. What goes down is cost per qualified lead and cost per sale.
Anyone judging the campaign on cost per lead will see it getting worse. Anyone judging it on sales will see the opposite. That is why measuring in sales has to be running before the signal is switched on, not after.
What changes in 2026
Meta has said it intends for any advertiser to be able to create and target ads entirely with artificial intelligence during 2026. Performance Max, on Google, has gone from 60% to 71% adoption. Both trends point the same way: less manual control over who sees the ad, more dependence on the signal the algorithm receives.
In that scenario, the signal is the only lever left. A company that sends real sales steers the machine. A company that sends form fills lets it decide on its own.
Before switching it on
The signal amplifies whatever it receives. If the definition of qualified is loose, the platform will learn to bring loose leads. If the CRM has unused statuses or salespeople who do not update it, the signal will be noise.
That is why the order matters: first the definition, then the CRM, then the signal. The diagnostic reviews all three in your company and tells you what is ready and what is not.