A Swipe Right for Educated Guessing the Impact of Digital on Offline Activity

By Jason Carmel, Global SVP, Marketing Sciences, POSSIBLE

Digital is the best when it comes to data collection, isn’t it? Given the tools and a reasonably comfy chair, a digital analyst can tell you nearly anything a consumer does. Analytics can tell you when someone arrives, from where, what they do, what they liked or didn’t, when and where they left, and how much they spent before they did. Media, apps, and even social are catching up as well. While it can be gnarly to staple all of those things together, if there’s any shortage in what we know about someone online, it’s not the data’s fault.

But all of that awareness evaporates when we switch offline. And, despite what digital trending indicates, the overwhelming majority of buying still happens offline. Most companies have absolutely no idea how their digital drives someone to engage offline. After being exposed to digital advertising, there are a series of often competing paths that the consumer could take to either finalize a purchase, or not. All of this happens far away from their screens, and there’s rarely a way to quantify an end state, let alone the path that led to it. Offline, we are dealing with Schroedinger’s Customer.

So marketers have a few choices. The easiest choice is to not attribute any digital influence to offline purchases. This feels neater because it doesn’t contaminate any of the firm digital data we’ve grown so fond of with “likelihoods” to buy. It’s also dangerously misguided. A 2015 study indicates that $0.64 of every consumer dollar spent offline was influenced by interactions online. Not paying attention to how digital influences offline activity is a recipe for a laughably inaccurate marketing plan.

The best choice is to build a system that requires authentication and tracking across every channel, online and off. I love this choice, but it’s a myth. You will never know exactly what a customer does offline or why they do it. Pure data feedback of offline behavior is the tooth fairy of analytics.

Which leads to a third way. I’m a big proponent of it, but it involves a word that is much (and, I would argue, unfairly) maligned in the analytics world: guessing. Guessing is NOT about making things up. It’s about using the data we do have to understand the otherwise unknowable. Here’s an example.

How Many Unplanned Pregnancies Does Tinder Cause Per Day?

Perhaps unsurprisingly, a simple web search doesn’t return any hard data on this. It’s about as unknowable as it gets. So where do we start?

1. I like population as the first variable. It sets a top bracket, and is typically an easy figure to drum up. Tinder tells us that it has 26 million matches per day.
2. Of those people matched, the vast majority are 44 and under, which we will deem a capable of conceiving a child. But let’s knock off 5% conservatively to account for those unable to have kids, and those (uh-oh) who are already pregnant, leaving 24,700,000 relevant matches capable of producing a child.
3. Not everybody who matches on Tinder ever meets in person. In fact, it’s a rarity, and it varies greatly depending on gender. A crude user research exercise held over lunch put the match-to-meet ratio for men at 15-20% and the match-to-meet ratio for women between .25% and 1%. A simple average puts the universal match-to-meet ratio at 7.6%. So we are down to 1,877,200 people who actually meet in person via Tinder, every day.
4. Of those people who meet in person through Tinder, let’s continue with an assumption that 1% end up having sex. The kind that could produce a child (please don’t make me be more specific here). This is of course debatable. But I am sure that 100% of in-person Tinder meetings don’t result in sex. And I’d be surprised if it were zero percent. So I picked 1% to be extremely conservative. And again, because the math is easy. That’s 18,772 uglies bumped per day.
5. Also, Tinder has an option for same-sex meet ups, which should be excluded when talking about pregnancies. Let’s assume that Tinder maps to the US population of Gay, Lesbian and Bisexual people at 3.5%. So we’re down to 18,115 heterosexual hook-ups.
6. Now let’s pause to deal with an attribution issue. Of these 18K+ sexual encounters, some of them would have happened anyway, even if Tinder had never existed. There is definitely no available data on this, but I like to think that any app with a $5B valuation would be relatively successful at promoting causal sex if it put its mind to it. So let’s say that 50% of the hookups can be directly attributed to Tinder. That 9,057 boom-booms courtesy of the app itself.
7. Of those sexual encounters, I assumed that 1% would be unprotected, because people can be stupid. (Note: a casual search on unprotected sex rates puts the number way higher, but -all together now- 1% is conservative and the math is easier). This means that there are 91 unprotected liaisons, and 8,966 liaisons that use some sort of birth control.
8. Of the 8,966 hook ups where protection is used, we will consider a failure rate of 1%. That means of the hooks-ups theoretically playing it safe, there are still 89 daily cases when the proverbial dam breaks during intercourse.
9. Add those poor slobs to the 91 couples who just couldn’t be bothered, and you have a daily grand total of 180 couples at risk of procreating.
10. Now we need to know how many of those sessions resulted in actual conception. To understand this, I beg you to trust that I’ve come up with a conservative average for a likelihood to conceive on any given day of the month as 3%. So of the 180 Tinder matches where conception is possible, five will result in a pregnancy. Per day.

Now I will freely admit that that actual number of pregnancies from Tinder in real life is not precisely five per day. But guesses aren’t meant to be precise. We need only enough precision to form an understanding of the magnitude of risk (opportunity?). And that’s infinitely more useful than assuming it is unknowable or (even worse) that it doesn’t exist.

Look, I know this example is ridiculous. No reasonable person would “blame” Tinder for a pregnancy, any more than they would blame John Legend’s “All of Me” or a little too much champagne. But the point is that attributing pregnancies to Tinder usage is a far more difficult and “unknowable” problem than understanding, say, whether your Pinterest page is getting people to your physical store. You have access to more data than you do for Tinder. You have been exposed to regularly-used internal assumptions for your company. And (no disrespect to your mad swipe game) you are far more informed on your business than you are on Tinder usage and its consequences.

Go and guess. The effort brings greater insight into how your digital impacts the way consumers navigate the world, online and off.