Lookalike modelling in a cookieless world
Back to Blog

Why partnership-led lookalike modelling thrives in a cookieless world

  • 4 min read

Marketing addressability tactics have a choice: adapt to an increasingly cookieless world or perish.

As third-party cookies crumble, it’s fallen on marketers to leverage other data types (namely first- and second-party) through reliable and secure data sources.

That’s why connected, collaborative and consented consumer data is at a premium.

It’s why trust-based data partnerships are becoming the cornerstone of post-cookie addressability.

And it’s why lookalike modelling is starting to look a lot different (and, dare we say, better).

Because there’s a new opportunity for brands to target audiences at scale: by leveraging partners’ data, marketers can bring greater reach and relevance to their seed audiences.

Here’s how the right data partnership enables lookalike modelling to thrive in a cookieless world.

1. Drive richer insights

Data partnerships enable marketers to enrich audience segments within their own first-party data.

For instance, if you had a segment of buyers with particular behaviours and wanted to create a new, potentially insight-laden segment by combining your own first-party data with a partner’s.

Say you ran an app-based food delivery service with reams of first-party data on your regular customers – but you also wanted to target people who use rival services.

You know your own first-party data is strongly representative of the behaviours of people who use a food delivery app (yours, in this case).

You could start a data collaboration with a relevant partner—say a supermarket chain—to share insights that illuminate consumer behaviour on other delivery apps (in areas like cuisine preferences, buying cadence and discount sensitivity inferred from basket data). 

By partnering with a fellow brand holding related industry data, you can enrich your information via a modelling extension and create a fresh lookalike model off the back of that combination.

You can then use your partner’s relevant data to clarify the common behaviours around people who use similar services.

So you’re not just reaching people you know, or your own customers based on your own first-party CRM data… 

You’re making your own data more targeted to the wider ‘takeaway’ audience, while expanding your reach to prospects more likely to engage with your marketing efforts.

Using secure partner data as a seed within your own first-party data, you can create a hyper-relevant model.

With your own first-party data augmented with your partner’s audience signals, you can add new dimensions of accuracy, relevance and reach to your lookalike model.

Crucially, both parties collaborate under the assurance that consumer privacy and the value of the data are upheld; as your CRM includes data from a variety of delivery apps, you set the technical and legal conditions that determine partners’ access to individual apps.

2. Challenge and validate your audience assumptions

Marketers tend to approach lookalike modelling with audience activation as the end goal.

But there’s also potential for a learning and validation exercise as a peripheral (but valuable) by-product.

Let’s take the same food delivery service. 

This time you have your own CRM first-party data, and hold assumptions around common behaviours your customers share.

With an aligned partner’s consumer data, you can challenge those assumptions and build more accurate audience seeds as the basis of your lookalike model.

It may turn out that people who ordered from that delivery service tend to be couples rather than singles, or order on certain weekdays rather than weekends.

And from that, you can work with your partner to build seeds specifically off the back of those insights.

You can also pit your first-party data assumptions against your partner’s model to find any hidden overlaps: if your assumption is right, most of your segments should find its place within matching or similar segments within your partner’s data.

If you’re wrong, you’ll find a far lower number than you assumed – either way, you unlock fresh clarity over your own data.

Beyond audience activation, you can continually refine and re-evaluate every model against your partner’s – you can pull different levers and understand different variables across your first-party data and profile that against your partner’s data.

Now you can work with your data partner to build a specific seed that’s relevant to your customers – not beholden to your assumptions.

3. Unhook from the cookie train

In the world of programmatic, lookalike models are based on finding cookies who have visited the same websites and then mirroring their behaviours.This then forms the basis of your initial seed audience. 

Now as cookies decline, you’d be forgiven for thinking lookalike models face redundancy. But with a partnership-based approach, lookalike models work and become more valuable in the cookieless world.

Today, partnerships are becoming a new driver behind data activation.

Ready to start?

The death of third-party cookies isn’t the end of lookalike modelling – in fact, it’s opened the door to more accurate and relevant models.

With the right data partnership, marketers can become attuned to their customers’ attributes and attitudes in ways becoming more widely adopted since the death of cookies forced a change of tack.

This is an opportunity to increase customer lifetime value and average order value off the back of truly understanding customer behaviours.

And it starts when you find the right data partner.