Programmatic branding is like using Tinder.

It’s not DR, but it’s still about the numbers game.

Advertisers that rely on offline sales face a dilemma — “How can we measure the effectiveness of our online advertising?”

I started my career in media on a CPG account that had all sorts of products in its portfolio, and I soon became intimately acquainted with cat litter, gummy vitamins, and women’s vibrators (don’t ask). Since clicks and site visits could hardly tell us how many containers of laundry detergent our campaigns moved off the shelves, the proxy we often defaulted to was the brand study.

Suspending our disbelief that people will answer random online surveys just for kicks, the brand study is a straightforward concept and a handy tool. By comparing baseline levels of awareness/intent to those of impression-exposed users, we can measure the amount of lift that a certain campaign element generated. Performance of individual campaign elements or media partners is judged by the relative increase in lift percentage. This was all fine and dandy, and while I was a planner I had the opportunity to plan some fun campaigns with sponsored content, high-impact units, and the like.

Fast forward to this past year, when I became a programmatic buyer on an agency trading desk. This time I was the vendor under scrutiny, and to be honest, it sucked. You can try all sorts of things, but brand lift is a rather fickle thing — there is no direct optimization lever for it, just proxies like frequency and viewability. Here’s what’s hard about programmatic branding:

1. Viewability. As a baseline, programmatic generally has a 30% in-view rate, which looks like shit when compared to site-direct benchmarks of 60%+. Be prepared to either shell out on the CPM or cut out a ton of inventory to bump up your figures.
2. Site Quality. Let’s face it, unless you’re buying programmatic direct, it’s highly unlikely that all your impressions are coming from Conde Nast & friends. Buying impressions through DSPs on the open exchange is quite like buying a case of Coors Light — it’ll get the job done cheaply, but let’s not pretend that we’re getting Macallan 25 (no offense to Coors, I’d love your business). To get the best results, it’s definitely helpful to involve PMP deals.
3. Ad Formats. Video is probably the best format you can get in programmatic for branding, outside of some billboard inventory you might be able to call dibs on. Programmatic video has a wide range of quality though, and the relative cost per viewable impression may or may not be worth it. If you’re like me, you’re stuck with plain Jane standard banner sizes unless you’re lucky and can get Triplelift assets.

Over the course of my first year, I kept mulling over how to perform on par with other partners on the various plans that got thrown my way. I didn’t have any flashy units to use, and it’s hard to compare banner ads to funny Buzzfeed cat videos. But then I thought, “We don’t need to make people love this product… we just need them to like it enough to try it.” Sure, it’s awesome to have superfans and brand advocates, but that’s not what programmatic is good for. Takeovers, rising star units, and native work great for driving awareness, but even a superfan can only buy so much {insert arbitrary CPG product here} in a given year. On the other hand, I can extend a brand’s reach with programmatic in a manner that maximizes the total amount of people willing to try the product. Maybe they’ll like it, maybe they won’t — but at least I got them to try, and a lot of them at that.

It really is sort of like Tinder. Unless they’re delusional, pretty much no one Tinder would fall in love with someone else from first glance or initial conversation. You both don’t know that much about each other, but you find yourself mildly attracted enough to strike up a conversation. Maybe you’ll go on a date or two and if it doesn’t work out, there’s plenty of others available. And so it is with programmatic branding — I can’t make people fall in love with your brand, but that’s ok. All I need to do is wingman for you until a consumer is willing to try, and if it fails, it wasn’t meant to be anyway. We’ll move on to the next million people who might work out, and I can hold your hand through the heartbreak.

With that said, let’s get down to the geeky stuff on why this philosophy works. When you run a brand study, your lift percentage is the proportion of the population that became more aware or engaged. 100% lift does not mean that everyone you reached is 100% more enthusiastic! 10% lift above a 10% baseline just means that a general 10% of the population you reached is more inclined towards your brand/product. If we reached 1 million people, that’d be 100 thousand consumers. It’s more relevant for business objectives to maximize the total number of affected individuals given a fixed budget, since that increases your total sales potential.

Consider the following example:

Partner A:

5% Control, 15% exposed = 10 lift points/200% lift

Unique exposed reach: 5MM

Spend: $200k

Partner B:

5% Control, 7% Exposed = 2 Lift Points/40% lift

Unique exposed Reach: 25MM

Spend: $175k

From first glance it might seem that partner A vastly outperformed on this campaign. However, the story is different if we look at the total affected consumers and the relative cost per consumer.

Partner A Incremental Consumers= 10% x 5MM = 500k
Partner A CPIC = $200k/500k = $.40

Partner B Incremental Consumers = 2% x 25MM = 500k
Partner B CPIC = $175k/500k = $.35

We can see here that both partners impacted the same effective amount of consumers, and that Partner B actually did achieved their results in a more cost-effective manner. What’s more important for branding campaigns? Not the flash/glamour, but the results.

I manage my campaigns a quarter mile at a time, and I’ll trust Dominic Torretto on that.

Disclaimer — these are my views and do not reflect those of my agency!