Top news and emerging ideas in display advertising, delivered every Monday.
3 Simple Alternatives To Attribution Modeling For Search Marketers
In the recent articles Why Search Marketers Are The Future Media Planners and How to Maximize SEM Efforts with Search Retargeting, we discussed how search marketers are ideally positioned to own a larger portion of the total media budget because the skills needed to buy display media are becoming more and more quantitative, just like SEM.
Techniques like site retargeting and search retargeting are often the starting point for a search marketer to break into this space as they are the most familiar, and the common metrics used in search (CTR, CPA, CPC) are valid for this type of display too.
However, the smart search marketer will soon start to ask questions about the interplay between search and display and see that a more holistic approach offers greater insight.
Clients demand this level of understanding, but it can be confusing and time consuming, particularly without a media background, and so in this article we provide 3 simple measurement techniques that the search marketer can use to answer your client’s questions simply.
When adding display to the mix for the first time, clients’ frequently want to understand:
- Am I paying twice from SEM and display to convert to the same person?
- Should I count view-thru revenue from display?
- Is the quality of the traffic the same?
The first thought is often to look at attribution modeling and develop a method of examining the interplay between each channel in your mix, but this is often an over-engineered approach.
I have seen many brands go down this path only for them to discover that attribution modeling can be complex and distracting; there are technical barriers to overcome and internal models to discuss and communicate. I have seen organizations and agencies tie themselves up in knots for months, investing far more money in resources than the answer to the question could possibly earn them back.
I am actually a big fan of a true attribution model for the right brands, but frustratingly, it often reveals the answer that there often is no real answer, and the investment means it is suited only to the largest of budgets and the most complex of scenarios.
Alternatively, there are actually three very simple techniques you can use that will address each of these common questions.
1. The Media Overlap Report
When adding display to a marketing program, is it actually driving incremental revenue or are you now just paying twice to convert the same person?
This code then needs to be passed into your SEM and display tracking software as a hidden variable. For SEM this might be your bid management platform’s conversion pixel, or your Google Analytics tag, and if you are using DART it is a simple custom variable. When you pull your regular SEM and display reports, be sure to include these unique ID numbers, and using something like Excel, look for the quantity of IDs that are unique, and the number that are the same.
If this percentage of IDs that is the same is less than 10% the answer to your question is that yes, display is driving incremental revenue over the search investment. If the overlap is a higher percentage, then you must determine how to divide the revenue between the two channels. Most commonly, I have seen clients simply give 50% to each channel for the conversions that overlapped.
(Once solved, you have actually arrived at one of the most complex attribution model questions with minimal effort!)
2. The PSA Study
View-thru credit is typically the sticking point for most search marketers because they don’t have it in their world (although often wish they did!). In short, view-thru is when an individual is exposed to an ad, does not click on it, but goes on to convert.
If you are a large brand like Pottery Barn that people are likely to shop at anyway, then it means view-thru revenue is artificially high, but if you are a small or new brand then it is probably more accurate.
The problem is that 0% credit for view-thru is incorrect as exposing people to advertising clearly works, but 100% credit is also incorrect as in the Pottery Barn example. The solution is a study to benchmark the right view-thru percentage for your display investment.
A PSA (Public Service Announcement) takes two groups of your audience and exposes one to the client’s display ads and the other to a PSA announcement. Clearly, the client’s ads can influence the buying behavior of the audience, whereas the PSA ads cannot as it contains no brand messaging.
To calculate the percentage of credit that view-thru should be given, then simply look at the delta between the two sets of results.
- PSA ad group – 1m ad impressions – $10,000 in view-thru revenue
- Client ad group – 1m ad impressions – $100,000 in view-thru revenue
This tells us that the audience you are working with was going to spend $10,000 whether you exposed them to the client’s ads or not, but when you expose the ads to them, they actually spent 10x more as a group. The percentage of revenue that should be counted from view-thru is therefore 90%.
At Chango, we encourage our clients to run a PSA test for the search retargeting and site retargeting campaigns we provide so that the true value can be understood, and I encourage you to ask your vendors to do the same.
Note though that best practice says run a PSA twice a year for each vendor or type of display campaign as each audience group will have its own benchmark – the reality is, most clients run one study on a single vendor once a year, which is not ideal but is better than nothing at all.
3. Traffic quality – New Versus Existing & LTV
Assuming your overlap report doesn’t highlight any problems, and your PSA tells you the right amount of credit to give to a conversion, the last question to address is whether the traffic the program is generating is of comparable quality to your other programs.
If you are now capturing the order ID from the Media Overlap report, use that same ID and have it compared against your CRM database. Are the customers spending the same in a transaction? Are they shopping as frequently? What is their estimated LTV (Life Time Value)?
A good quality display program will show that the behaviors are very similar, but this is a good test to ensure you have no surprises waiting.
Moving into display is relatively easy for the search marketer as they already have all the tools and skills needed for these new ways of buying media like search retargeting, but as the programs become more sophisticated, unfamiliar questions can arise around measurement.
Whilst an attribution model might be the first thought to address some of these complex problems, you can look like a hero by bringing these simple measurement tricks to the table. With minimal engineering or integration effort, the most common questions about the interplay between search and display can be answered.
Some opinions expressed in this article may be those of a guest author and not necessarily Search Engine Land. Staff authors are listed here.