Does Your Business Rely on False Signals?

The marketing world is full of data. 

It comes from everywhere and begs for quick action to be taken against it. But which data is right to use? It turns out that very little data should be trusted to make business decisions, particularly in the world of marketing investment and measurement. 

Does your business use data from “platforms” like web analytics, search engines, social media, and CRM systems? Chances are good you’re using that data incorrectly. Each of those platforms can be very powerful when used to make decisions within those platforms. One of the biggest problems we see with marketing today is that those platforms somehow get a cushy seat at the data table to make investment decisions outside of those platforms. Point solution platforms are not to be trusted on their own with cross-channel or big picture investment decisions. 

Five main reasons why those platforms are horrible for making investment decisions on their own. 

 1. They Don’t Establish a “Base.” This is a huge problem, over-attributing 80-90% of total measured platform volume. Platforms are programmed to count (or these days to project) every sale they see as an attributed sale. I.e. one that is driven by marketing. The obvious issue is that it simply isn’t the reality. Most businesses have past customers, a digital presence, locations on well-traveled streets, etc. Even without paid marketing investment, 80-90%+ of people would have purchased anyway over a given timeframe (an attribution window). 
 
 2. They Don’t Account for the World Around Them. A compounding issue with the above problem is that people make purchases in context of the rest of their lives and the world around them. Platforms don’t account for that either. There are not embedded considerations for economic conditions, weather, construction, seasonality, competition, or many of the other variables that impact the business. 

 3. They Don’t See Each Other. Platforms are walled gardens. They don’t know if the same person who saw an ad on one platform also saw an ad on another platform. If that person goes on to purchase – both platforms likely take full credit. Partial credit would have been better, or none at all if that customer had purchased anyway. 

 4. They are Based on Clicks. A force working in the opposite direction of the above points is that a bulk of attribution is based on clicks. Certainly search, but also web analytics, CRM, and largely social media use click-focused measurement. The trouble is that most people don’t click on an ad. In the traditional media days, we knew that television drove business growth, but we didn’t rely on them to click their TV to measure that growth. The same problem exists in digital today. Most people don’t click on most ad types. Search inevitably ends up getting the lion’s share of credit because that is where people click. Search budgets balloon while the business results do not. 
 
 5. The Ads Aren’t Seen by People. In non-click platforms like online video and digital display, a scary-high percentage of ads are either served out of view of humans or not served to humans at all (because they’re served to bots). The strangest thing about these platforms is that they don’t disqualify impressions or ad exposures that weren’t seen. They take credit for them anyway with their head in the sand
 

What to do?

Use the platforms for their strengths, which is making in-platform decisions. For bigger decisions like how much to spend or where to spend it, then you need to look elsewhere to avoid applying data incorrectly. There are two powerful tools, which can be used together for even better long-term results.  

 

  • Experimental Design. Set up incremental lift studies to measure the impact of marketing on the business results you actually care about. If you run social ads in five cities, do your business results go up in those five cities above and beyond other similar cities’ business results? Geography is one lever to pull here, but so is audience. Many of the maligned platforms discussed above do also offer incremental lift measurement with audience-based test and control where they separate some of their platform users from the rest and measure results between the two groups. This is a much clearer view of reality, which the smart marketer can compare to the attribution natively reported in-platform to see how far off those two views of performance can be. Looking at multiple experimental results in aggregate is a sound approach to decision making. 

 

  • Marketing Mix Models. MMM is again the gold standard in measuring the impact of different variables on your business results, from media channels to PR to other marketing activities, even including external variables like competition, macroeconomics, and weather. MMM faded to the backseat for a while with the irrational exuberance around MTA (multi-touch attribution). MTA fell far short of its promised real-time perfection thanks in part to the many walled gardens that won’t share data and in another part to bad methodology (like not accounting for base volume). The best multitouch partners today are indeed MMM solutions with a small dose of click-based MTA logic built in as well. Again, with the clicks… 

 

We know this topic is often frustrating and can be overwhelming for clients. The best approach to break through the fog is to combine frequent experiments with ongoing MMM to measure incremental business results and help each tool complete one another. With those tools in your belt, you’ll be much more ready to build the business with good data-driven decisions.