LinkedIn Lookalike Audiences To Be Phased Out: Predictive Modeling Is the Future
LinkedIn has recently announced plans to retire its lookalike audience targeting feature. For institutions and those in their marketing department who heavily rely on this functionality, this change will require significant adjustments to their campaigns.
Insights into LinkedIn's Decision
LinkedIn has specified that the lookalike audience feature will only be available until February 29, 2024. What happens to your current lookalike audiences:
After 30 days, if a lookalike audience isn’t used in a draft or active campaign, it’ll be archived.
If you add an existing archived lookalike audience to a campaign, it’ll show Building status and will use static data.
If an archived lookalike audience is expired, it’ll be deleted and won’t be accessible.
Alongside this announcement, LinkedIn has proposed two alternative targeting methods for marketers: predictive audiences and audience expansion.
Exploring Alternative Targeting Methods
Predictive Audiences: Predictive audiences, similar to lookalike audiences, draw data from lead gen forms, LinkedIn contact lists, and conversion data generated via the Insight Tag. Additionally, LinkedIn combines the marketer's data source with its AI capabilities to automatically generate a new custom audience for campaigns.
Audience Expansion: Audience expansion works to broaden the reach of advertisements by targeting users with attributes similar to the original audience. For example, if a campaign targets adult learners interested in higher education opportunities, such as "Continuing Education" or "Online Learning," audience expansion may extend the audience to members who list similar keywords on their profiles, thereby uncovering new quality prospects.
Venturing Beyond LinkedIn: While LinkedIn's alternatives offer valuable targeting options, they lack a crucial element: intent. This gap can be filled by what us here at CollegeAPP have to offer. We connect institutions with our database of over 240 million adults who have been scored on their likelihood to say they have intent as it pertains to enrolling in continuing education programs. This presents a powerful opportunity, particularly for those working in higher education. Adult student recruitment is challenging because 75% of adults don't intend to enroll.
CollegeApp helped one university struggling to get qualified leads and keep cost per lead down by focusing on audiences with specific intent. The university was able to generate 310 leads with an average of $30.09 cost per lead in the first month of the campaign. This is just one of many examples of successes we have found.
Evaluating Pros and Cons
Benefits: Predictive audiences leverage LinkedIn's internal systems and machine learning, offering potential improvements in results over time. Moreover, for institutions lacking sufficient data, Audience expansion provides a way to reach potential learners by selecting attributes that match the target audience.
Along with this change comes the migration to a single API, meaning Ads Lead Sync and Events Lead Sync will be deprecated, which may make buildouts easier than having to hook up several API connections in the future.
Drawbacks: The retirement of lookalike audiences may disrupt established campaign strategies and create silos in ad processes because current API connections involving LinkedIn lookalike audiences blended with more down-funnel metrics in comprehensive marketing or CRM platforms will no longer work. Here are a few platforms we know support LinkedIn integrations:
HubSpot
Salesforce
Zendesk
Zoho CRM
Pipedrive
While integration types may vary and perhaps not all these platforms are integrated in a way where the retirement of lookalike audiences has a direct impact, it is worth checking with your tech team just in case.
Audience expansion isn’t available for dynamic ad formats or predictive audiences. But just being left with predictive audiences presents some concerns. For smaller institutions, data sources must have 300 or more members to create a predictive audience. For those who are targeting super niche audiences, it may be hard to hit these numbers. For larger institutions or those managing several institutions with one account, there is a limitation of 30 predictive audiences per ad account, and they cannot be shared between accounts.
The models LinkedIn is using to make their predictions come with a lot of unknowns. We don’t know what attributes are getting weighed more heavily. The new LinkedIn Predictive Audience measures various activities only on LinkedIn, so the algorithm could include and weigh more so the criteria the advertiser doesn’t see as important. Additionally, even if LinkedIn were to recognize that people who typically have a certain characteristic engage your institution's ads more, that doesn’t tell us if the leads that come in are truly qualified leads because that’s off-platform data. All this ambiguity leaves advertisers asking, “how will I know if I am creating the right message for the individual seeing the ad?”
Moving Forward
Don’t worry, there is a path forward. This change might make a custom audience more powerful because it gives the control to the advertiser as to who to add to that list. And if that custom list is even more specific by including those who have shown some indication they might be more likely to enroll, that’s even more powerful.
There’s a number of reasons people can engage with an ad and they may come with different intent levels or senses of urgency. This is why working with CollegeAPP, who can provide these types of predictive lists based on criteria you choose and a model that is built with higher education in mind, will be a good alternative to consider.
With less than a month until the retirement of lookalike audiences, it's crucial to use this time wisely. Here is a preparation checklist you can use to get started:
Experiment with new campaigns using predictive audiences or audience expansion, and consider exploring intent-focused databases relevant to your industry. A/B testing can help determine the most effective strategy for your specific needs.
Make sure any insight tags, conversion APIs, lead gen forms, and contact lists are set up. The contact lists require a status of “Ready” to be used with predictive audiences.
Due to expired lookalike audiences being deleted, for your records to match with historical performance, you might want to note some details about those audiences and the results they drove in a spreadsheet.
If currently have an API in place, migrate to the new Lead Sync API to collect leads from ads, events, and other organic sources via a single API.
In CRM, data visualization platforms, and marketing automation systems that currently were pulling in data from lookalike audiences, one might want to add a note of date of change so if reviewing historical data along with data coming in from these new options, there’s an understanding of why valleys and peaks might be showing up on your graphs where unexpected.
For additional support with your ad campaigns, don't hesitate to reach out for assistance. Our team is here to serve you.
