woman with blonde hair giving a presentation


Is My Advertising Working? (And Why That is a Hard Question to Answer)

Advertising is an investment to generate sales, and return on investment is the key metric in determining whether advertising is “working.”

That was the message from Elea Feit, Assistant Professor of Marketing, LeBow School of Business at Drexel University at the 5th Annual Digital Analytics Philadelphia Symposium, held on October 29, 2015 at Drexel.

Calculating ROI with purely online sales is relatively easy. But, As feit posited, the real challenge involves calculating ROI for online and offline sales. Feit discussed how brand loyalty programs and ad tracking are increasingly solving the “data problem” by allowing savvy companies to match up online and offline user behavior.

Professor Feit described a mixed modeling process, (also known as algorithmic attribution) that will use logistic regression to properly attribute sales lift across multiple channels. For more details on this process, see her article, Fusing Aggregate and Disaggregate Data with an Application to Multi-Platform Media Consumption, in the Journal of Marketing Research, in June 2013. From the abstract, the article: “provide[s] a method that firms can use, based on readily available data, to gauge and monitor multiplatform media usage. The key innovation in the method is a Bayesian data-fusion approach that enables researchers to combine individual-level usage data (readily available for most digital platforms) with aggregated data on usage over time (typically available for traditional platforms). This method enables the authors to disentangle the intraday correlations between platforms (i.e., the usage of one platform vs. another on a given day) from longer-term correlations across users (i.e., heavy/light usage of multiple platforms over time).

Feit admitted that this challenge is one of attribution issue as well. She warns against making a direct attribution between exposure to one advertisement and a sale and declaring that calculation to be the end of the story. Feit instead advocated that the much better approach is to analyze the behavior of customers who have seen advertisements and customers who haven’t.

Feit also cautioned against the assumption that advertisements only affect customer behavior on the same day they were shown. Instead, she described the ad-stock model, which assumes that the effect of advertising is initially very high, but slowly decays like a leaky bucket. Again, a regression analysis can determine the rate of this decay and determine sales lift over time.

Apples to Apples Is Key

Next, determining whether advertising works requires an apples-to-apples recognition that seems obvious in hindsight: customers who don’t see an advertisement are different people than ones that do. The issue in this assumption is in over-attribution – believing that advertisements work better than they really do.

For example, Feit discussed television advertisements on Animal Planet. Animal Planet viewers who would see that advertisement are different people than ones that don’t watch Animal Planet, and don’t see the advertisement – for one thing, Animal Planet viewers are much more likely to purchase dog food. Or to put it in the digital arena, subscribers to an email mailing list are likely different than those who aren’t on the mailing list.

The Solution: Test, Test, Test

Feit’s solution to over-attribution is with an ROI experiment (and a data model). From a targeted audience, select a random subset of customers to not expose to advertising. This allows for a control group with which to measure the impact of your advertising.

Feit created an in-depth model that can interpret this type of experiment and predict sales lift figures (both initial and over time) across multiple campaigns. This model can accurately show whether advertising works.

Key Takeaway: Every digital marketing initiative should be founded on, and backed by, solid analytics. But it’s important to make sure that the data provided by those analytics doesn’t ignore important variables that can affect attribution and results. A/B testing is the easiest, most scientific way to cut through variables and answer the key question: Is my advertising working?