ROI is very important, but it’s also a bit of a myth.
It’s tempting to think that ROI should be easy on the internet. After all, websites run on computers, and e-stores use databases, and we can track links and clicks and such, so that when you boil it all down it’s just math. We should be able to link A to B to C with mathematical certainty.
Not really.
ROI matters. You can’t go investing in marketing efforts without knowing that they’re doing any good. At the same time, you’re kidding yourself if you think you can tie costs to results with very much precision. Despite the best technologies, there’s a whole lot of fuzziness to the equation.
Consider retargeting as an example.
Retargeting works by displaying your ads to people who have visited your site. It’s what causes ads for cordless circular saws to follow you around the internet after you did some online browsing about cordless circular saws. Yes, it can get annoying, but the idea is that the constant reminders will make you come back and buy.
The trouble is, linking a specific purchase to your retargeting ad campaign isn’t easy. In an ideal world, the sequence would go like this.
- A visitor comes to your cordless circular saw page and gets a cookie that will identify him across a broad ad network.
- Your non-buyer goes hither and yon on the internet, sees your retargeting ads, and thinks, “You know, I need to finish that fence in the backyard, and a cordless circular saw would make it easier.”
- Your now re-engaged prospect clicks on the ad and goes to your landing page.
- Your reclaimed visitor completes the purchase.
That sequence is very trackable. The trouble is, that’s not what happens in the real world, mostly because people don’t click on ads. What they often do is see the ad, then google the thing all over again. If you’re running an Adwords campaign, you may now have two marketing efforts that want to take credit for the same transaction.
There are other common practices that mess up retargeting, like when people delete cookies and when they use different devices. E.g., I may casually search for cordless circular saws on my tablet, so the ads might follow me around there, but I might complete the purchase on my desktop (because it has a real keyboard).
The messiness of actual human behavior can foul up your ROI calculation.
There’s another thing to bear in mind with retargeting, which is that some portion of the people who visit your site are going to come back and buy anyway. So how do find out how much of a lift you get from the retargeting ads?
A few years ago I tried to answer that with an A-B test. We split our website visitors into two groups. The first group got our retargeting ad while the second group got a public service announcement. We then compared the behavior of the two groups.
The retargeting people want you to credit the retargeting effort with every example of someone coming back to the site and buying. But that’s nonsense.
We found that the retargeting ad provided a 33% boost in conversions over the control (people who got the PSA). That is, some of the people in each cookie pool came back and purchased, but 33% more of the people in the retargeting pool did.
The retargeting campaign did have an effect, but the number of conversions reported by the retargeting campaign were wildly exaggerated, since only a third of them should have counted.
That’s just one example of how ROI can get messed up. With careful experimentation and planning you can make your ROI calculation better, but you’re never going to get it exactly right. It is, at best, a good estimate.