The Pitfalls of Data Driven Marketing

I often hear techie marketers boast about how they are “data driven” so they don’t get caught up in the errors of human judgment – as if it is possible to make complex decisions based solely on what their computer model, formula or data shows them.

The idea is seductive because with large data sets over a large period of time it is possible to predict patterns, narrow focus and eliminate waste by optimizing marketing efforts and budgets to the predictions of the data model.

But the catch is, it will always take a thinking person to figure out what the data means and how to apply it to real world actions.

Too many “data driven” marketers begin relying so much on what the data says, that they ignore glaringly obvious real world indicators that what the data tells them is incorrect, misinterpreted, or just wrong.

In general consumer markets with huge volumes of searches, transactions, and conversions along with a relatively few possible search phrases or keywords used to find the product – such as in the t-shirt market, it may be possible to see from the data that the majority of sales come from women between 18-35 during the months of March through June, using the keywords “cute tee shirts”, “cute sayings tee shirts”, and “cut t shirts women’s” – then confidently limit spend to women of that age, during those months, and for those keywords to reduce waste and increase ROI.

However, in high cost, low volume markets such as behavioral health services it doesn’t work that way.

This is because the overall volume of searches is low, the cost of clicks is extremely high, and spread over nearly 3,000 keyword variations used to search for a solution for a child’s issues.

In the t-shirt market, there are only so many ways to describe what you are looking for (t shirts, tee shirts, t-shirts for x, x color t-shirts, etc.)

But in the behavioral health market a parent who is looking for help for a child may search for anything related to “help” and “child” including, rehabilitation, treatment, behavior, schools, boarding schools, military schools, therapeutic boarding schools, disorders, therapy, reform, camps, boot camps, therapists, wilderness, substances, anxiety, depression, adoptions, defiance, anger, PTSD, abuse and on and on with each of these general phrases widened by “for boys”, “for girls”, “for teens”, “for teenagers”, “for adolescents”, “near x”, “for x disorder”, “for y symptom” etc.

Someone selling t-shirts can generally focus on one area of t-shirts, a few styles, and colors. But with behavioral health marketing and residential treatment centers, a treatment center serves nearly all of the areas listed above (and more) with only a few exceptions.

So it is almost impossible to focus on just one “behavior” when the behavior may be caused by drugs, peers, PTSD, anxiety, abuse, alcohol, relationships, obesity, mental disorders, etc.

Additionally, since the search phrase used could be nearly anything of 3,000 variations, and conversions are spread out across all possible search phrases.

Actual intakes for an entire year to the treatment center would likely show up as one here, one there, etc. with only a few phrases racking up the huge number of 2 or 3 for the entire year.

With this kind of data, it becomes impossible to truly see the trends and patterns a “data driven” marketer wants to see.

But often they do see a trend – in the few phrases responsible for the “2” or “3” intakes or enrollments.

Then they limit everything down to those “key” phrases, and end up killing their business.

Another way I see “data driven” marketers get killed in the behavioral health marketing arena is to misinterpret unusual changes to their data, intakes, or conversions, and frantically begin “undoing” things they think may be the cause for the drop in conversions or intakes.

It happens every single election year without exception.

In “regular” years, parents look for help for their troubled teen whenever the need arises – the most urgent problem they are facing is the problems that their teen is causing or dealing with.

But in election years, especially for presidential elections, parents will hold off looking for help away from their home during the months leading up to November.

The most urgent or threatening problem they are facing in these times is the uncertainty of the world – and having their son or daughter far away from home when they are uncertain is something they don’t want to think about, so they wait.

One final example in behavioral health marketing.

It is interesting that when presented with actual data that DOES show a trend and a predictable pattern that should be watched for and addressed, it is usually overlooked or ignored.

Then panic sets in causing everyone to rush and “undo” everything that might have caused the change. Or worse, they fear that everything that has worked for years no longer works, so they dump everything and go looking for another “solution” or “new ideas.”

The predictable pattern I’m referring to is the seasonality of searches.

Behavioral health services for troubled teens follows a seasonal trend beginning with the start of a new school year having a high volume of searches and intakes, followed by a declining volume to a sharp dip into summer months, then back up to a high volume at the beginning of school again in the fall.

The big holidays always show a drop because parents usually think that a family event, trip, or vacation with gifts and fun will pull their family back together – but after it doesn’t pan out the search volume spikes after Thanksgiving and Christmas.

The sad thing about this is that it is entirely predictable.

Like Christmas, it never comes unexpectedly – it happens every single year.

But “data driven” marketers overlook this one almost every year, and almost without exception.

So yes, data is good and even imperative. But to rely on it blindly instead of looking at it as only a guide – like with using a GPS, you could easily drive right off a cliff because the GPS data says there is a road when there is a cliff.

But if you took the time to look up, look out the window, and compare the data to the real world you would see for yourself that the data sometimes leads you to do things that would be catastrophic when followed blindly.