Attribution: What to Expect When You’re Expecting ROI


stork_smallADOTAS – When my wife was first pregnant, she bought the book, “What to Expect When You’re Expecting.” We were new to this journey, and wanted to know exactly what this answered for over 15 million people: what to expect.

And as online advertisers, agencies and publishers leap into this year with an attribution model on their list of goals, they often ask the same question: What should I expect?

How will moving from last-click attribution model to a full-funnel attribution model affect how I buy media?

What should I expect in search?

What should I expect in display?

What should I expect in affiliate?

With news from the longest running attribution study of its kind (two years), expectant media buyers will be pleased. But before we deliver the results, let’s reveal how our industry first came to ask these pregnant questions.

Imagine you’re responsible for a multimillion dollar online ad budget. And you learn that today’s outdated online ad tracking systems give 100% credit to the very last clicked or last viewed ad before an online transaction.

Example: if seven Internet ads contribute to a transaction, today’s outdated systems allocate entire credit to the very last ad, ignoring the prior ads that created your revenue.

Zero credit to revenue drivers — 100% credit to the last ad. It must be a bad dream, but it isn’t. For Internet advertisers spending $70 billion worldwide each year, we’re slowly waking up to the problem.

Now some coffee — a full-funnel media attribution model. Robust attribution modeling systems recognize credit should be assigned to a team of Internet ads versus the last ad.

Teamwork in attribution modeling arrives through assigning credit to Originators, Assists, and Converters within a transaction. Every transaction.

An attribution model should capture all online media sources involved from the top of the funnel where sales originate… down to the very bottom of the funnel. So in a $100 transaction, an Originator would receive a fraction of $100 credit–and the Assist and Converter would also receive fractional credit of the $100 attributed to them respectively.

One hundred percent of revenue credit is attributed and split among Originators, Assists, and Converters–accounting for the actual drivers of revenue. Then revenue and respective costs from paid media sources converge in a single, elegant number: Attributed Revenue-to-Spend Ratio (ARSR).

It’s a simple ratio any marketer can grasp. If you have a 4.0 ratio for a specific keyword, or specific Display campaign–you’re getting $4.00 in revenue for every dollar spent on that particular media source. Conversely, if you have an Attributed Revenue-to-Spend Ratio (ARSR) of 1.25 for a particular media buy—you’re getting $1.25 in revenue for every dollar spent there.

Scale paid media sources with high numbers, and cut or improve media sources with low numbers. A quick, actionable and fully attributed metric. The special sauce–is the numerator of the ratio (attributed revenue), derived from full funnel attribution.

Your newborn attribution model can indeed be a bundle of joy (not every attribution model is…read here for seven key details).

But what should you expect in search?

With category keywords (e.g. “Crossover SUV”) you’ll discover good news. These upper funnel keywords are definitely expensive on a PPC basis. Perhaps they kept you up at night because the CPAs were horrible, but blended with brand term efficiency, it came out in the wash.

Now you can sleep. Since a full funnel attribution model measures every PPC term, you attribute not only originations, but assists—tracking beyond current arbitrary time limits, and event caps deleting every view and click beyond the 10 most recent.

So with an accurate picture of the entire funnel matched to costs, some of the most expensive non-brand terms are definitely worth the price per click, and some will not to be. But finally, an attribution model brings full vision.

Competitor keywords–will simply shock you with the knowledge from attribution modeling. As protective parents of our campaigns, we typically view any competitor as a threat, and bid on competitor terms “just because.”

Let’s say you have four competitors. What you’ll discover with full funnel attribution is that you may have a positive correlation to Competitor A and Competitor C. Meaning, you actually want them to succeed–because when Competitors A and C are searched, you actually generate revenue.

So for these two competitors, you’ll root for them. Suprising. But for Competitor B and Competitor D–you may discover a negative correlation. Meaning, when they are searched, all you’re doing is spending money without returns. Essentially, not worth the getting worked up over, and not worth the bid. Surprising.

That’s what to expect in search. And after two years of analyzing and reallocating based on this attribution model–search ROI increased 98% vs. baseline.

What should you expect in display?

Display is a puzzling baby for most people. Sometimes you just don’t know what to expect. What happens in display with a robust attribution model is full revelation. Here’s the tale of two networks:

One small network labored with a small five-figure monthly budget, and wasn’t even on the advertiser’s radar screen. In the first 60 days of attribution modeling, they finally received the credit they truly deserved–a ratio of 4.1. The agency doubled budget with them every month to see if they could scale. Six months later, they enjoyed a healthy seven figure annual budget while holding a solid ARSR.

The flip side. Another display network was discovered to be producing a ratio of 0.9 (less revenue than media spent). The network failed to improve, and spend was cut–improving profitability of the overall ad budget instantly. Their budget was reallocated to media sources producing higher attributed revenue-to-spend ratios.

Intelligent caring and feeding of display delivered 160% higher ROI for the channel in the two-year attribution study. Great expectations.

Now… what should you expect in affiliate?

For current affiliates, nothing much changes. Most often, affiliates are your best performing partners, and there’s a risk the slightest change could cause them to jump ship to your competitors (and revenue runs out the door). Be prudent in the reality of the marketplace. You will, however, have a more accurate picture of your affiliate channel, typically crowding the bottom part of the funnel.

But attribution enables you to recruit new kinds of affiliates…ones who originate revenue vs. competing at the funnel’s crowded bottom. Because, now, you have a model to compensate affiliates for originating and assisting. Without upsetting the affiliate ecosystem and crowding it even more—you plant crops of new affiliates growing new revenue vs. harvesting the existing revenue.

What can you expect when you’re expecting ROI? To summarize the results of longest running attribution study of its kind:

  1. Seven-figure efficiency in the advertiser’s online ad budget
  2. Display ROI improvement of 160%
  3. Search ROI improvement of 98%
  4. Accurate economic model to grow non-competing affiliates

What can you expect? The fountain of new knowledge. Sounder sleep at night. And yes, amazing ROI. Welcome to the new world.


Please enter your comment!
Please enter your name here