By How do you earn budget if attribution is dead and traffic is vanity metric?
Lift-based experimentation. That’s when, rather than try to assign a weight to every potential channel or tactic you (or an algorithm) think contributed to a particular conversion, you measure the lift in conversions that happen after a particular tactic or channel begins. Weirdly, it’s less precise, but more accurate, because rather than rely on the few channels that are good at sending referral data, you know for certain if your sales have gone up vs. an expected baseline during a period of sustained investment.
Can other factors complicate this? Yes. Does that make the process invalid? Not compared to the only alternatives: attribution (which misleads in a thousand ways) or not measuring anything at all (which, unfortunately, isn’t an option for marketers who need to prove their worth to skeptical bosses, teams, or clients).
In this week’s 5-Minute Whiteboard, I’ll walk you through exactly how these lift-based measurement systems work, with clear examples 👇
Transcript
Friends, I bet you’ve been seeing a lot of these: The Alligator Graph.
It’s a graph where clicks and impressions diverge, and it’s because the age of zero click everything is here. The traffic that you used to get from Facebook, from Twitter, from Linkedin, from Reddit, from Pinterest, from YouTube, it’s all going down. Google of course as well, the AI tools are keeping all the traffic for themselves, zero click everything. And so you get this alligator graph and everybody is seeing it.
And what I’ve been hearing is, hey Rand, you keep talking about how in this world we have to measure influence rather than attribution.
Don’t tell me that, tell me exactly what I need to measure otherwise I will never get budget for anything again. Okay. So let’s let’s let’s go back for a second and I’m gonna walk you through that. I think it’s a fair question I wanna answer.
So this classic buyer journey right As imagined by marketers, it has like the top of the funnel where your audience learns about the problems that you solve and then potential customers are engaging and interacting with your brand learning more and then essentially they’re deciding whether they’re gonna buy from you or anybody or competitors.
I admit it’s it’s kind of a terrible metaphor today because buyer journeys don’t look like this. They look like a pinball machine. Right? You get shot out of there.
You’re like bouncing around social media. You ask a friend. You visit pricing pages and read news and skim blog posts and consume podcasts and YouTube channels and you read a white paper and get a demo and then there’s email newsletters, right, that bounce you back up. So pinball not funnel.
I agree. But where funnel is useful is for measurement. At the top of the funnel you can still measure things that are useful. Right?
Like whether somebody viewed your social media post. Right? The number of impressions that this video gets on social channels or on our website, it’ll be in both places. Whether people were exposed to a brand advertisement if you if you can measure that.
There’s some companies doing this right with like, billboard, you know, they they try and do eye tracking to see that. Whether they saw your brand in search results or AI tool results, the impressions number, right, that you’re getting from Google and these other places. And then there’s middle of funnel like replies and engagements and people who Google your brand. That’s that’s in the middle of the funnel most of the time.
People who clicked on your results, that’s in the middle of the funnel. And at the bottom, they they followed you. They subscribe to your newsletter. They added a product to the shopping cart.
Right?
So with the way that you measure in twenty twenty five is not trying to attribute every sale that you got to some number of contributing channels and then weighting them and using a machine learning system. No. That is that is dumb. It doesn’t work for a million reasons attribution it dead.
Don’t don’t forget. Right? The Apple cookie change, the anti tracking privacy laws, the ad blocker adoption which is at forty percent plus multi device journeys which are a hundred percent domination of apps which you can’t see data inside click consumption on every single platform way before people get to you and dark traffic that’s missing referral data and a ton of other things. You you’re never gonna get it.
The way you measure in twenty twenty five is you use lift based testing. You can do it with geography, reasonable time series testing, also reasonable. There’s lots of other simp similar segmentation. That’s the smart way to do it.
What do I mean when I say that? Lift based testing is here’s the simplest example. Right? Your Coca Cola, you’re gonna run two billboards in very demographically similar cities and then you’re gonna measure in a five, ten, twenty five mile radius whether same store sales lifted more in one region than another.
They’re all getting the same television ads and radio ads and the rest of the Coke branding series, but these two billboards are different. And you have measured that, you know, highway a in region one gets about as much traffic as highway two in region b. Great. Fantastic.
That’s lift based measurement and then you you take the winning one and you go verify, you run it in a bunch of other cities. Same thing you can do in digital marketing. This is not just an offline tactics, not just a tactic from nineteen sixty five, it’s a twenty twenty five tactic.
For example, we’re doing terribly. I’m embarrassed to show you this but look, I have started putting my five minute whiteboards and my other short videos onto YouTube. They have performed horribly for me on YouTube. I haven’t done much with them but but we can see as, you know, at at the top we can see like impressions for these YouTube things through the channel, metrics that YouTube provides and we can see views, then we can see, whether traffic to our website is lifted to people type in SparkToro more to Google and come to the site.
Does that brand lift happen? I hope in a year or two or three after these videos do well that we will see that it will perform, but it will not be because we can tell that they came from YouTube. The only way we’ll be able to see it is in our, brand numbers. Right?
In in those kinds of metrics. The way I show impact is test based measurement. Right? It is it is lift based.
I run tests. I see lift. Amanda does the same thing. So we have things like brand interest.
You know, over a three month trend. We can see our numbers. We know what historically it looks like. We know what the growth rate should look like.
We can see social engagement and subscribers. We can see website traffic. Yes. Traffic is a number.
It’s in the middle of the funnel for us, but it’s a real thing. Home page probably posts, not so much. That might be top of funnel for us. And then there’s subscribed audience.
People who subscribe to our email newsletter, followed us on a social channel, maybe subscribe to that, you know, YouTube channel and then sales at the bottom. So top, middle, bottom sales. You have these two. They’re gonna be different for every different one.
The metric you should use unfortunately is I cannot tell you. I can’t tell you, no one can tell you. Your company depending on your channels, your goals, your campaign, it’s gonna be unique.
Broadly, it’s gonna look like this. At the top, you’re gonna see metrics that show initial exposure to your brand.
Right? That could be things like, number of people who saw your LinkedIn posts. For us that’s a big one and we definitely look at LinkedIn impressions. It could be people who saw your YouTube views.
It could be people who drove by your billboard on the highway or listen to a radio station at a time. Maybe it’s share of market. Yep. All that kind of stuff.
Initial exposure.
Then it’s some kind of engagement. You know, for us that’s like a subscriber or a reply to us, or someone signing up for the the email the blog email. And then metrics that suggest a deeper interest like they they come to spark together. They, you know, are they’ve signed up for a free account, a free SparkToro account. You know, we have a forever free account system. And then at the bottom it’s sales and conversions.
Yours is gonna look different. I can’t tell you what yours is going to be. You have to build for yourself, but you can measure these things. You can show the trend over time. You can build the graphs and then you can do lift based investing and testing, which is the only way to really test to know whether a marketing channel or tactic is performing for you in twenty twenty five.