How to optimize your campaign for sales, not just leads

The cost per lead dropped, the paid media report shows an ever-better number, and even so the sales team keeps complaining that sales are not coming in the same proportion. This is the most common complaint among marketing coordinators and managers when they hit the CPL (cost per lead) goal and still hear from the sales team that the base is weak, a sign of marketing and sales goals that do not meet. Lowering the cost per lead looks like a win on the spreadsheet, but optimizing a campaign for sales requires sending the algorithm a different signal from the one most accounts are sending today.

Why the ad platform brings in more than you asked for

The bidding algorithm of ad platforms, Google Ads and Meta Ads, cannot tell on its own the difference between a lead who buys and a lead who just fills out a form and disappears. It learns by observing the conversion event configured in the account. If that event is “filled out the form,” the system goes after more people with the profile of those who fill out forms.

There is no guarantee that this profile matches the profile of those who sign a contract at the end of the funnel. The target it pursues remains the form, not the sale that comes after it. With paid media more expensive in 2026, paying for a lead who never converts weighs even more on the budget.

The signal most accounts send today is shallow

In practice, this means that the pixel or conversion tag fires at the exact moment the form is submitted. That event carries little information about the quality of the lead: it does not say whether they had the right profile, whether they moved on to a meeting, whether they became a proposal, or whether they bought.

The algorithm receives a shallow signal and optimizes within the limits of that signal. The shallower the event, the easier and cheaper it becomes to generate volume, and the further that volume gets from revenue.

How to optimize a campaign for sales in practice

The way to fix this goes through the data sent to the platform, not through the creative or the targeting. The steps:

• Map the sales funnel stages that already exist in the CRM: qualification, visit or meeting scheduled, proposal sent, sale closed.
• Choose which of these stages best represents the quality of the lead for your sales cycle. In a longer sale, proposal sent is usually a better signal than qualification, because it has already filtered out those who did not have fit.
• Send that event back to the ad platform as a conversion, in place of the form event or alongside it.

It is this sending back that changes the game. The platform starts repeating the pattern of those who reached the proposal or the sale, not the pattern of those who only filled in a field.

The bridge between CRM and paid media is called offline conversion

This sending from the CRM to the ad platform has a technical name: offline conversion, an event that happened outside the website, within the sales process, and that needs to be imported manually or through an integration. It is the basis of optimization by offline conversion, and on Google Ads this flow has changed in recent months.

A good part of the traditional import is migrating to the Data Manager API, a more structured way of sending this data. We have already covered here on the blog how the GCLID, the Google Ads click identifier, supports this import and what breaks when it is not captured correctly. It is a worthwhile complementary read for anyone setting up this flow now.

What changes when the algorithm learns from those who buy

When the event sent better reflects who advances in the funnel, the algorithm starts looking for an audience more similar to that profile, not just a cheaper one. One way to decide which CRM stage is worth sending as a conversion is to look at lead scoring, which already exists scoring leads by their profile and behavior.

This change of signal also interacts with the attribution model used in the account: if the credit for the sale always falls on the last click, the algorithm learns from a partial slice of what actually generated the sale. The effect also shows up in the calculation of CAC (customer acquisition cost), which is already more delicate in sales with a longer cycle and becomes even more sensitive to the type of event feeding the funnel.

A warning before switching the signal

Switching the optimization event is not a switch that flips on its own. Ad platforms need a minimum volume of conversions on the new event to leave the learning period, and that usually takes a few weeks. In that interval, the cost per conversion of the new event tends to look worse than the form's cost per lead, because more advanced funnel stages happen less often than a simple form submission.

This should not lead you to revert the change in the first week: this slower period is the price of teaching the algorithm to aim at a rarer and more valuable target. Keeping both events active for a while, the form one and the advanced-stage one, helps to compare without losing top-of-funnel volume while whoever is trying to optimize a campaign for sales waits for the new signal to mature.

Frequently asked questions about optimizing a campaign for sales

What is offline conversion in paid media?

Offline conversion is an event that happens outside the website, within the sales process, and that is sent back to the ad platform as an optimization signal. Stages such as proposal sent or sale closed, recorded in the CRM, are examples of this type of event, imported manually or through an integration.

Why does cost per lead drop, but sales do not follow?

Because the algorithm is optimizing for the event you configured, and not necessarily for those who buy. If the conversion event is the form submission, the platform looks for more people willing to fill out forms, not people with the profile of those who close deals.

Which funnel stage should I send as a conversion to Google Ads or Meta Ads?

The stage that best represents the quality of the lead for your sales cycle. In longer sales, proposal sent usually works better than qualification, because it has already filtered out those who did not have fit with the product or service.

How long does it take for the algorithm to learn a new conversion event?

Usually a few weeks, because the platform needs a minimum volume of conversions on the new event to leave the learning period. In that interval the cost per conversion tends to look worse, which should not lead you to revert the change ahead of time.

An adjustment that changes the campaign's yardstick of success

A low CPL was never the problem. The problem is the shallow signal hidden behind it. Optimizing a campaign for sales means teaching the ad platform to recognize the profile of those who reach the end of the funnel, with real CRM stages feeding the bid, and not just those who fill out a form. This requires consistent naming between CRM and media, standardized UTMs, and a well-designed offline conversion process, without depending on manual adjustment every week. That is exactly the alignment that Laiki's Data Management Playbook documents, field by field.

The official Google Ads documentation on importing offline conversions details how this event enters the account and starts feeding the bidding algorithm, for anyone who wants to set up this flow with the technical team.