One of the most common questions our team gets in sales calls: “Custom audiences just don’t work for acquisition anymore. So what makes Innkeepr different?”

The short answer: Innkeepr builds audiences around intent.

In this article, we’ll walk through how platforms like Google and Meta actually use CRM data - and how this translates into new strategies for modern targeting.

Why CRM Data Is Misaligned with Modern Targeting

Not too long ago, Meta was a place of micro-targeting - using CRM data to seed lookalikes based on purchase history, recent funnel activities or product affinities to ultimately find the right cohorts.

But the game has changed. With broad ad sets and ASC+ everywhere, Meta has replaced almost 96% of granular targeting, and most growth teams are experiencing that custom audiences just don’t perform anymore.

And they’re not wrong. But the issue isn’t that custom audiences don’t work anymore - it’s that most teams feed platforms with the wrong kind of data. In today’s intent-driven ecosystem, audience performance depends on signal quality and relevance. Without behavioral and intent-rich inputs, even the best targeting strategy can’t deliver effective results in acquisition anymore.

How Platforms Actually Use Customer Data

Before we begin, it’s worth understanding how platforms like Meta or Google actually use customer data.

Audiences Are Vectores

When uploading a customer list, each entry becomes a high-dimensional vector in a behavioral space - including past purchases, affinities, locations, devices, and more. The platform uses this data to identify similar users based on proximity within this abstract n-dimensional vector space.

You could say that your customer list becomes a blueprint and the richer and denser the signal, the better the algorithm can generalize and scale performance.

Why CRM Data Falls Short: A Data Science Perspective

Now that we have a picture of how ad networks interact with customer data, let’s take a closer look at the nature of CRM and customer data. From a data science perspective, it quickly becomes clear why it underperforms for acquisition and lookalike use cases:

• Intent: CRM data tells you who used to care. Not who’s ready to buy right now.

• Overfitting: Customer lists often contain purchase histories spanning months or even years, causing algorithms to over-index on outdated traits.

• Signal Quality: Customers represent only a small subset of your total audience - making the available dataset both limited and noisy, and therefore difficult for algorithms to generalize from.

• Loyalty Bias: Returning customers often look nothing like new prospects. Their behaviors, preferences, and conversion paths are shaped by past brand interactions - making them a poor foundation for predicting what drives first-time buyers.

• Engagement Mismatch: Many CRM lists include contacts who haven’t engaged in months - or even years. These users may still exist in your database, but their behaviors no longer signal interest or purchase readiness.

• Privacy Shifts: Due to privacy regulations (such as GDPR or CCPA) and technical restrictions (such as iOS 14.5), platforms like Meta and Google struggle to accurately match CRM records to real users. 

This is what we call the 8 CRM blind spots for Custom Audiences.

CRM lists send the wrong signals to modern ad algorithms. They’re outdated, sparse, and misaligned with how platforms find buyers today

The Alternative: Treatment Effects

So how can we improve audiences to help algorithms generate demand?

A good starting point is to redefine “demand.”

Demand = Impact

Demand describes a cohort or segment with a measurable likelihood to convert. The goal of demand generation, therefore, is to increase net conversion probabilities.

That means we don’t just want to find users who would buy anyway - we want to identify those who convert because of our demand generation activities. This is where causal analytics comes in.

Causal analytics help us understand those “treatment effects” - measuring how likely a user is to convert due to the incremental impact of e.g. an ad exposure. It’s the true north star for demand generation: maximizing lift on conversion likelihoods with least amount of effort.

With platforms like Innkeepr, treatment effects can be modeled across millions of anonymous sessions - long before a user logs in or joins a CRM list. Instead of guessing, it identifies most-relevant users and build audiences around them.

For a deeper dive into how treatment effects work, check out our dedicated blog article.

A Comparison Between CRM & Treatment Audiences

Focusing on treatment effects instead of CRM lists fundamentally changes how platforms can interact with data:

• Targeting focuses on users likely to convert due to ad exposure—not those who already converted.

• Demand expansion is based on finding new segments who are responsive to our signals, not just similar.

• Privacy restrictions become less critical since behavioral data doesn't rely on PII.

• Skewness & biases are minimized since every session suddently becomes a valuable data point, not just every checkout.

The following table outlines the key differences between CRM-based and treatment-based audience strategy.

CRM ≠ Future

TL;DR custom audiences are still a powerful tool for lookalikes and acquisition - they just need to be built differently. As platforms have shifted toward intent-based optimization over the past years, audience data strategies must evolve in parallel.

The future of data-driven targeting looks like this:

• Adopt treatment effect modeling to measure the true impact on demand.

• Build audiences around incremental impact, not identity.

• Use behavioral data to reclaim targeting precision — without relying on post-login data or personal identifiers.

This isn’t just a more effective way to target. It’s a mandatory path for growth.

We are excited about the future of causal analytics at Innkeepr. If you’re interested in learning more about how causal analytics can change the way you approach growth, book your demo now!