How do you measure what's working?

No doubt you’ve heard the quote famously attributed to the noted retail giant and marketing pioneer John Wanamaker back in the early 1900’s, “Half the money I spend on advertising is wasted; the trouble is, I don’t know which half.”

Late into the 20th Century the answer to this conundrum was barely improved. Marketers relied primarily on survey data to determine whether their advertising was moving the needle in any substantive manner. Ad campaigns were measured by consumer surveys that reported back statistically significant changes in brand awareness, perception and usage. If there was a lift, the campaign was deemed a success.

For consumer packaged goods companies things got slightly better in the early-90s. With the advent of UPCs, modernized grocery stores and mass merchandisers began scanning products at checkout and that data became available for marketers to measure the actual lift in product sales. Agencies could open and close the spigot of advertising and promotion to determine directionally, which tactics were actually moving product off the shelves.

The 21^st Century brought a greater promise of advertising accountability, especially with the Internet’s claim of delivering 1:1 marketing. Would agencies finally be able to report back to their clients the actual ROI for their advertising efforts? The short answer is, not exactly. As hard as agencies tried, they could not fully isolate the impact of one single tactic unless they were only using one tactic. If a person was exposed to an Instagram post, podcast, email, banner ad, and then did a Google search for a certain product how could you define which was responsible for that particular sale? Or which combination of tactics? Well…you couldn’t.

So largely marketers are still left guessing which tactic contributed most to that sale. There are all kinds of attribution models that are used today to try to determine the relative value of each interaction that a customer encounters with a brand.

Last Interaction Attribution is where the last interaction before the sale gets full credit for the sale. Then there is First Interaction Attribution where the first interaction gets credit for the final sale. Some marketers use Last Non-Direct Click Model which assigns 100% value to the single interaction you can isolate before a person clicked through to visit the website to make the final purchase. There is also a Linear Attribution Model which assigns equal credit for each of the interactions with the customer. In this scenario, a person who was exposed to a Facebook ad, email, Instagram post and banner ad for a product and then went to the brand site directly to make a purchase, each of the tactics would be assigned 25% of that sale.

You can see the problem with each of these models…they don’t solve John Wanamaker’s challenge with identifying what absolutely works. Last Interaction, First Interaction and Last Non-Direct models are all flawed because they ignore all of the media touch points in between. The Linear Attribution model seems to be the most “fair” of these tools because it assigns equal importance to everything. However, you can never isolate the ROI and identify an individual or grouping of tactics that have more impact than others.

If John Wanamaker were alive today he’d still be asking the same question. Only now, the question might go like this “I know that a very small portion of my advertising is wasted, but the trouble is I cannot tell which portion because these attribution models can’t pinpoint exactly which tactics are contributing to the biggest ROI.”

So I guess that’s progress.