Use Big Data for Smart Facebook Targeting
Published: January 13, 2014
Author: Declan Dunn
Feeding frenzies can teach you a lot, especially during the first Christmas of Facebook’s Big Data.
When Facebook cut Page Reach 20 days before Christmas, lots of very big budgets began making expensive runs at their own Likes, while looking for needles in the haystack of Partner Categories, Interests, and demographics.
Big Data in the right hands, with the ultimate measure an in-person purchase, is serious gold for Facebook. The rest of us ad buyers have to find ways to make Big Data serve what we do, without throwing variable darts at demographics or Partner Categories, hoping something makes sense. Unless you sell offline, all that data may/may not have anything to do with your results, and you can waste tons of time mixing general, offline retail categories chasing the illusion that you’ll find what you want.
Many ignore the people they are already in front of – customers and repeat visitors – who are a better foundation for Facebook Big Data advertising than mixing together untested Categories and hoping to find a pattern.
Like all things digital marketing, it comes to focus, spreading your bets on a few variables, and watching where to scale your ad buy because you can isolate where the activity is coming from.
Let’s look through the 3 Facebook Big Data tools you can use to build on your existing audience to make small but real gains:
1. Partner Categories – FB targeting through Datalogix, Acxiom, BlueKai, and Epsilon.
Looking through all the Partner Categories and possibilities is much like your first visit to Costco.
So many things look cool and cheap, but they come in big amounts. Big numbers tend to mean less precise targeting, so look for the median number between the highest and lowest.
For example, here’s a Partner Category we tried – Online Buyers, because we sell online only.
Online buyers as a variable didn’t test well, so we got more specific and changed that one variable to Interests, specifically Amazon.com; after all, what online buyer doesn’t know or hasn’t bought from Amazon.com?
This interest became part of the control ad buy, likely because that segmentation was specific, not a general basket of Online Buyers, and a small basket that was highly relevant, measured by activity at the right cost.
Remember that the power of Partner Categories is in the offline purchase, so the measure of an online buyer in a world driven by offline buying may/may not impact results. Like most testing, the more specific we got, the better the result.
You can learn a ton about who your audience is without ever buying an ad. We created Custom Audiences and began to segment them, based on Partner Categories. The learning about the customer base was huge, not used for targeting on Facebook but for campaigns within our site, and retargeting in general. You don’t get specific numbers, but you do see powerful groupings.
The more data you target, the more variables come into play – and having all this data doesn’t always convert better; in fact often it tricks people into making risky ad buys, trying to find that proverbial needle in a Facebook Data haystack. Less is more in Facebook ad buying!
2. Custom Audiences is an excellent way to continually segment and focus your Facebook Ads.
While Big Budgets can afford to market to strangers, most of us would do better beginning with those we know – meaning emails, Facebook User IDs, App IDs, and/or phone numbers – to get to know them better, by applying Big Data to data you already have, and growing it.
By segmenting your Custom Audiences like crazy, you can set up many ad buys at lower overall budgets that collectively help you reach your audience. The goal is more, segmented lists that will help you target, as well as split your ad buys and track them by the qualities of your audience, not just the ad.
If you know Location, for example, using this knowledge and seeding your ads with statements related to that physical location helps make the ad relevant. Just make sure you follow the Facebook TOS on this and don’t call it out too dramatically.
Mentioning physical location in advertising has long been a powerful tool; knowing zip codes and location, in a mobile world driven by location, is smart.
If you don’t have an email list, join the many reasons you should collect these on your Facebook Page. So many social media experts never understood the power of email.
You can use your IDs if you have an App, or target Likes to your page of course.
Phone numbers can be the ultimate targeting for companies with deep phone databases – one of the places telemarketing will surely make itself known in this space. Much mobile activity involves the phone, so it makes sense phone number targeting could be powerful for some businesses.
3. Lookalike Audiences teaches you growth is a 1-5% game.
A Like as a measure of engagement is obviously waning, and with the low amount of Likes ever coming back to a page, those who Like are often strangers to the brand. Why market to strangers? Get people to come back to your Page, and business, as Repeat Visitors – not just Likes!
Take that same approach to your Custom Audiences, not just by throwing up a huge list of emails but segmenting where you can, and running them through Partner Categories to see any obvious groupings – you don’t have to buy an ad to do this, just Pause the campaign.
Segment Custom Audiences and begin testing the 2 types of Lookalike audiences you can target as well, with relations defined by Facebook:
Similarity: the 1% on Facebook related to your data;
Reach: the 5% of Facebook users related to your data.
Statistically you will see why big numbers really fuel Big Data; with millions these numbers could add up, but if you have a small audience to begin with, these numbers are going to be very small.
If you segment, you’ll also see the 1% and 5% related to your segment of your audience, and not your whole audience, while will improve results.
Yet even that small number begins to show you how to build your business; at 1%, the audiences are similar, based on a variety of factors that of course are not clearly spelled out for you. (As always, testing is advised.)
The 5% is a broader audience; it reminds me of the Six Degrees of Separation meme. When you succeed and begin adding the 1-5%, that growth should compound if you do it right, at a cost that is somewhat reasonable. Small, agile growth instead of huge leaps is a recommended strategy here; over time you’ll grow targets that perform based on similarity and not just Interests or demographic data, which is always tricky to master.
Looking for more growth than 1-5%? Look outside of Facebook to partners who have audiences matching the profile you develop within Facebook.
Broad targeting made a bit more specific is what Facebook Big Data offers, but without some big numbers to begin with, it’s hard to take advantage.
I love the idea of Big Data – connecting customers by always being there, mobile and tactile, and responding to where they are and what they are doing with an offer to help and connect with the person at the right time, right place, and right intention to buy.
The real Big Data opportunity in Facebook ads is likely right in front of your eyes, and you can’t see it because they are your existing customers and visitors. In an ad world driven by marketing to strangers, it may be smarter at Facebook to start small, with who you know, and grow, than to do high level, multi-variable testing to strangers.
Reach, attract, and convert the friends of people who know, Like, and trust you, and you’ll find the Big Data value for your business.