After iOS 14 privacy changes, lookalikes audiences native within Facebook/Meta Ads have become nearly worthless. All Machine Learning algorithms at their core require data. And lots of it. Bad data in = bad data out. Less data = less accurate models.
DO YOU REMEMBER WHEN FACEBOOK LOOKALIKE AUDIENCES GENERATED POSITIVE ROI? OR EVEN JUST BASIC FACEBOOK INTEREST AUDIENCES FOR THAT MATTER?
UNFORTUNATELY,CHANGED ALL OF THAT
To date, all lookalike models – including Facebooks – are built off of both cookie and pixel data.
If 96% of US user have opted out of tracking,
how well do you think Meta’s algorithms are able to
find new cold traffic that looks like your customers?
Both Safari and Firefox already block all 3rd party cookies by default.
And Google will soon kill 3rd party cookies as well!
Do you now get better results just targeting “broad”, untargeted audiences, and win by focusing on the best creative? Feels like running a billboard on the side of the highway, right?
Well, you’re not alone. But we have good news for you: by training a Machine Learning / Artificial Intelligence model with your existing 1st party data (read: cookie-less, iOS 14 proof) of existing buyers – you can unlock a “cold” audience of HOT prospects that are exactly like your ideal customer.
Ok, we know what you’re thinking - “AI” is the new buzz word of the day and everyone is suddenly an “AI” company/expert, right??
Well, as a matter of fact, Fortune 500’s have been using similar technology for years.
We know this because our CEO, Alex Herndon, has managed $5M/mo media buys for not one but two Fortune 500 clients. These companies pay well over 6 figures per year – with yearly contracts – for similar tech. But we’ve democratized this technology, bringing it to the SMB space for a fraction of the cost and no long term contracts.
Lookalike Audiences in
the 1st Party Data Era
We ingest a file of 1st party PII (Personally Identifying information) data of US Consumers.
We run these seed files against marketing databases, including credit bureaus, across 3,400 attributes. As these are mostly offline data attributes, these are cookie-less and immune to any future changes of 3rd party cookies.
After training a ML algorithmic model of what your existing buyers look like, we then apply this model to the US population and pre-score 96% of US adults on their likelihood to convert – BEFORE they have even heard of you.
This audience is then ingested into your ad account, via LiveRamp as a hashed email list, for you to run ads against.