Facebook has made an old marketing measurement trick new again, getting many marketers excited that it could finally match online ad impressions to offline shopping behavior for better ROI calculations. The only question is, will it work?
First, here’s the old trick — a “matchback.” Say you were a marketer running a direct mail campaign, and wanted to know how many people really bought your product after you sent them postcards with “50% off our Gizmo!” messages. You could count phone calls or web visits if you printed unique numbers or URLs on the cards, but those always undercount response — for every 10 people who call in or click to your landing page, another 20 might show up at the store. If you can’t count all the behavior, you’ll never know how the campaign really worked. Ah, but being clever, you decide to match the list of people you sent mailers to with the names of buyers at the stores (where computer systems can count up those names and addresses via loyalty cards). If that mailing cost you $50,000 and you generated $200,000 in sales, voila! — you have a 4:1 return on investment!
This is called a “matchback,” because you match the people who buy back to the list you sent the direct mail to. Doing this, of course, requires that you have a centralized ID system for the people you marketed to that can be matched to the actual people who buy at the store.
Enter Facebook, replicating all of this online. Facebook has the largest data set in the world of social media users and their corresponding email addresses; it has entered a partnership with Datalogix, a massive data company that has detailed information on most U.S. consumers compiled from retail loyalty cards, which identify consumers at stores. Online tracking of your ID + offline tracking of your ID = hey, a new matchback!
In simple terms, Facebook is trying to solve the “real ROI question” on its advertising, by finally counting all possible sales — not just online clicks — resulting from FB ads. Marketers love this idea, because finding the real source of sales is extremely difficult when most digital advertising can only be counted to the online checkout and not the offline behavior.
See FB ad for suit. Buy suit at mall. Now, FB knows.
Let’s use a theoretical campaign for Men’s Wearhouse as an example:
1. You see an ad on Facebook for a Men’s Wearhouse suit. You click or don’t click on the ad, it doesn’t matter.
2. This weekend, you go to Men’s Wearhouse and buy a suit.
3. Men’s Wearhouse now gets a report from Facebook matching people exposed to their FB ads with people who actually make the product purchase.
4. Men’s Wearhouse marketers can now tally up all sales and revenue from those ads, and get a true picture on ROI.
To protect privacy and avoid freaking out FB users, individual personal information is disguised through a series of data hashing (which prohibits marketers from learning that exactly you, John Doe, bought the suit), but in aggregate marketers can get a count of exactly how many ad exposures match how many consumer sales.
While it all sounds sexy, Facebook’s matchback poses a question — could it over-count results? Facebook ads sure reach a lot of people; because the ads have low click-through rates (about 0.03%), a small $10,000 budget spent on Facebook at an average $1.50 cost per click would back out to a whopping 22.2 million ad impressions. If you hit the entire world with your ad message, and some buy, those people may be triggered by many other factors, and Facebook could be taking credit for sales that might have happened anyway. Maybe you just really needed a new suit.
No matter. With Facebook struggling to continue to scale revenues, a matchback service in the world’s largest digital ecosystem will be certain to lure marketers. Correlation may not be causation, but Facebook will sure look good taking credit for it.
Image: Toni Blay