Marketers need to create new predictive audiences away from Google and Facebook

Why outsourcing your audiencebuilding to the big tech companies is depriving you of a deeper understanding or your audience

It might seem convenient to rely on sophisticated, in-place systems, but are you giving away your competitive advantage asks Jana Jakovljevic, VP, Client Success at Cognitiv…

 Marketers have a big problem. They’ve become addicted to the ease of Facebook and Google – so much so that they’ve become completely reliant on the two platforms for the bulk of their digital advertising. Of course, that’s exactly what Google and Facebook want: marketers to spend as much time and money on their platforms as possible. You might think that’s fine, so long as marketers are getting the return on ad spend (ROAS) they want; they are actually making it more difficult for themselves in the long run. 

The crux of the issue is the use of predictive audiences. As any marketer who’s advertised on Facebook knows, the platform makes it incredibly easy to generate predictive audiences based on who it thinks will be the best target for conversion, as well as lookalike audiences that exhibit similar behaviors to your existing customers. The same goes for Google, which also automatically generates these audiences for quick and easy use. 

Understanding and reaching an audience across channels

So what’s the problem? Well, by outsourcing the work of predictive audience-building, marketers deprive themselves of the opportunity to truly understand their customers. This in turn affects how well they’re able to reach audiences across all channels, resulting in, at best, missed opportunities for conversion and, at worst, wasted advertising spend. 

This might manifest itself in several ways. First of all, the use of predictive audiences generated by Google and Facebook is limited to the platform they’re created on – so a Facebook look-alike audience can’t be ported over to, say, TikTok. As a result, marketers might find it more difficult to optimize their advertising on multiple channels, and instead consolidate to focus on just Facebook. 

Secondly, marketers’ own data is being absorbed into these platforms, and is being used to fuel algorithms not only for their own organizations, but for their competitors as well; so any competitive advantage they may have had by utilizing these algorithms is effectively wiped out in the extremely likely event that their competition is also using Facebook or Google to advertise. 

Privacy changes and third-party cookies

Finally, brands that rely on these predictive audiences quickly run into issues of scale, and are increasingly getting lower ROAS. As Facebook grapples with the impact of Apple’s iOS privacy changes and Google prepares to phase out the third-party cookie, advertisers will need to find alternatives that enable them to maintain addressability and reduce their reliance on walled gardens. 

How can they do this? By understanding alternative identity solutions and fully leveraging their first-party data to provide takeaways that they – and only they – can use. Brands are sitting on mounds of first-party data that hold incredibly useful insights on not only their ideal customers, but also the indicators that mark someone as likely to convert. 

Additionally, because this data contains a more holistic understanding of the customer journey, marketers can identify the key platforms and touchpoints they should be investing in, instead of putting all their eggs in one basket. By bringing the process of building predictive audiences in-house, brands can diversify their advertising strategy in a way that would prevent single points of failure. 

The power of Deep Learning

Of course, that begs the question: how can marketers turn this mass of raw data into actionable insights? The answer is, the same way Facebook does – Deep Learning powered by artificial neural networks can help marketers maximize the impact of their own data, improving targeting and lowering the cost of customer acquisition. What Deep Learning does better than other machine-learning methodologies is identify complex patterns; so by feeding it your deterministic, first-party one-to-one data, it will quickly identify patterns in your customer behavior and accurately predict new, incremental customers. 

Marketers need to harness the power of their first party data to create predictive models outside of the walled gardens of Google and Facebook. They can no longer leave it in the hands of tech giants.

By Jana Jakovljevic

VP, Client Success