Menu

Ecommerce Brands Should Take Measurement Notes from the TV Industry

Measuring advertising lift
Ever since the world’s first banner ad, in 1994, asked users to click “right here,” clicks have been the default measure of ad success.

But clicks, it turns out, are a poor indicator of whether an ad drove sales. A consumer may have seen 10 digital ads before purchasing, but 100% of credit goes to the last ad clicked. Multi-touch attribution — an alternative to “last click” that awards a value to each digital ad a consumer sees — is an advertiser’s dream but collecting the depth of data to measure every factor impacting a buyer’s decision is impossible to implement in reality.

However, lift measurement — a staple of TV and billboard ad measurement — is a much better gauge of online ad performance than any click-based attribution. While digital advertising may be a quantum leap over traditional advertising — advertisers can target with precision who they want to see their ads — lift measurements are one case in which TV and billboard advertising had it right.

How lift studies for TV and billboards work

The best way to prove that a drug treatment or ad campaign worked is via a randomized control trial (RCT). When applied to offline advertising, the preferred RCT is a geographic lift measurement, in which one regional area is put in a treatment group and shown TV commercials or billboard ads. Meanwhile, a similar regional area acts as a holdout group and isn’t shown any advertising.

Related post: Why Battle Over Advertising Attribution When Incrementality Can Win the War?

If sales jump for the treatment group compared to the baseline of the holdout group, then the ads were successful. The amount of the rise in sales, commonly known as incremental revenue, tells you if the advertising spend was worth it.

For example, imagine if McDonald’s ran TV commercials in Cincinnati for its McRib sandwich but didn’t run them in Cleveland (assuming customers in those cities show similar purchasing behavior). If McDonald’s found that sales were up 20% in Cincinnati versus Cleveland, then they can surmise that the commercials prompted that 20% increase.

Digital gave birth to flawed attribution approach

Lift measurements aren’t perfect. Other factors could have increased McRib sales in Cincinnati. Maybe an influential radio DJ talked up the McRib during the morning commute. Also, separating by geography means the groups may not be exactly comparable, due to regional differences in personal preferences, culture, weather, etc.

But lift studies are far superior to click-through rates. Clicks don’t tell you much about whether an ad drove incremental revenue. So why are we so focused on them?

“The advantage of doing lift measurements in the digital world is that audiences can be truly randomized and used as part of continuous testing and measurement.”

One reason is that search pioneers decided that pay-per-click was the best model for ad sales. It makes sense: clicks provide solid proof that a user responded to an ad. This model underscored the advantage search advertising had over display in identifying short-term intent. Even better, advertisers could analyze the on-site behavior of ad clickers.

In the ideal scenario for this model, a consumer sees an ad for a sweater, clicks on it, and buys the sweater online. While that certainly happens, a consumer can also view the sweater ad, think about it for a week, look at other sweater ads, and then visit the site and buy the sweater. Or maybe it was the combination of online ads and a billboard that sent the consumer down the purchase funnel.

It’s no secret TV and billboard ads have always had a big problem: the advertiser doesn’t know who sees the ads. After all, you can’t click on a billboard or TV commercial. With digital ads, the advertiser does know — at least on an anonymized basis — which opens the opportunity for much better measurement. But many advertisers falsely assume that because a consumer clicked on an ad, that the ad led to whatever behavior followed. It’s a correlation versus causation scenario.

In either case, there’s no solid proof the ad worked. Lift studies, on the other hand, are the best possible way to determine incremental revenue as the result of an ad.

Lift in a digital world

The advantage of doing lift measurements in the digital world is that audiences can be truly randomized and used as part of continuous testing and measurement — as opposed to finding a new city to hold out on billboards and TV commercials. It’s no big deal for L.L. Bean or Adidas to withhold digital ads from 10% of their online audiences.

Related post: Navigating the 3 Ecommerce Audience Types

The reason this needs to be done continuously is that the revenue from your holdout group provides your baseline (revenue you’d make with no advertising). And that baseline will vary over time due to seasonality, competition, and changes to your products. If your holdout baseline is high, then advertising is getting too much credit; if it’s too low then advertising isn’t getting enough credit. Continuous digital lift measurements provide the most accurate baseline, from which you can calculate the true incremental return on your advertising.

Unfortunately, many marketers still focus on click-through rates and can’t say whether their digital advertising helps their bottom line. But measuring and optimizing your ad spend based on lift measurement will finally directly tie back incremental revenue to specific ad campaigns.

While this approach isn’t new, it is hiding in plain sight for digital advertisers. Billboard, print and traditional TV advertising budgets may be in decline, but the tried-and-true method of lift testing to measure ad performance is timeless.

A version of this article originally appeared in Entrepreneur.

All Things IncrementalityAll Things Incrementality: A Guide to Growing Revenue Profitably

Are you optimizing your advertising toward the only performance metric that truly matters to your bottom line? Here’s everything you need to know about incrementality.

Download the Guide