Skip to main content
Business

Marketing Might be Ruining Your Data Science

by: Zank, CEO of Bennett Data Science

Consistency in marketing matters. A lot!

Here’s a common scenario, exaggerated to make my point: Let’s say your company deviates from its usual marketing plan to go after a big new celebrity with a fantastic following. Sounds terrific, right? Here comes a 20% uptick in new leads, all from one strategic initiative. A tweet goes out and one of your products is shared seemingly all over the internet. It’s a fantastic buzz. You’re having trouble sleeping! This is too exciting. You sit down at the end of the day and poor over the results. Keeping vanity metrics out of it (such as total visits to your site or number of clicks on a link that doesn’t result in a purchase), you see worse than expected results. But there was so much traffic. And this celebrity is HUGE across the world.
We see this sort of thing all the time. What could have gone wrong? Lots, as it turns out! Let’s look at why this sort of thing happens, how to see it coming, and how to take the necessary steps to make the most out of new traffic or users.
Let’s define success for our new marketing initiative as an increase in product sales. With that in mind, let’s look at a few different scenarios we see over and over.

Different Scenarios

One of the chief assumptions essential to any data powered system is that the underlying data don’t change much. It’s close to common sense. Let’s think about users and products:
  • Over time, users don’t change that much (I’m not talking about slow evolution, but rather big unexpected jumps)
    This makes sense, right? Imagine if every one of your users won the lottery at the same time. Would their purchasing behaviors change? Probably. Would your company continue to serve their needs when they’re cruising town in their new Ferraris?
  • Over time, products don’t change that much
    This is pretty well established. Most products don’t change drastically. A pencil doesn’t become a smartphone (I’ve got to work on my examples, but you get the point; stuff changes iteratively, not BANG overnight into something completely different).

Scenario ONE:

In the best-case scenario, the new users are just like the old users and all the smart systems in place handle them work just fine. In this case, you’re set! The new traffic results in an upwards scaling of your business. Not much should change, unless something breaks due to all the new traffic!

Scenario TWO:

In this middle-case scenario, the new uses are somewhat different than your existing customers. In this case, it’s not that all your customers traded in their station wagons for Ferraris, but maybe a few are driving Maseratis (ok, time to drop the car analogy). What happens to your messaging if suddenly you have a whole group of users who wake up 14 hours before the rest of your users. All that research you did to optimize message timing likely won’t apply, even if you account for the new timezones. Shipping? Communication style? Similarity to other current users? You’ve likely taken on a new demographic; and that’s not bad at all. It can be beneficial, but in terms of accounting for them and providing personalization at scale, there’s a lot to think about and get right before we can expect to see the same ROI we expect from existing customers.

Scenario THREE:

This is the worst case. The new users came in pulled by their favorite celebrity. They click around once or twice, buy (almost) nothing and leave more quickly than they came. This is a horrible waste of resources and everyone’s time. The business suffers from handling all the new users who are just about to leave forever, the celebrity just lowered her/his cred and the IT department is not happy about the huge spike in pager duty calls over the weekend!
There’s a lot of good news, especially for the third case. This mismatch and revenue sink can be largely prevented by asking the right questions upfront. And, of course, avoiding being star-struck. There’s no good data science tool I’ve seen yet to keep a CEO from falling in love with the new celebrity de jour who can “absolutely send our business to the moon!!”. This one is hard to stop. I’ve seen it more than I’d care to admit!

Understand your current and future customers

When going after new customers, it pays dividends to have a relevant customer profile, generated by the data scientists who are helping with your AI or automation. Those are the people who will understand how to best describe your users. From there, you can work together to decide how far from that profile marketing can go before systems break. An easy example is age.

Let’s say you’ve been marketing and selling effectively to young adults for the past two years. Then someone in your company surfaces the idea of having celebrity X promote one of your products for a few days. This celebrity has a huge reach that will drive hundreds of thousands of potential buyers to your product. That alone is almost too exciting to ignore. But a quick perusal of the celebrity’s following shows that it’s mostly retirees in the later stages of their lives. I know this sounds like it would be an immediate no-go, but we have seen this; more than once!
There’s no doubt that these groups are going to respond to your messages and your products differently.

Prevent wasted marketing dollars:

You can prevent huge marketing mismatches from happing. Here are a few things I recommend:
  • Start by identifying segments of your current customers and summarize each one carefully  (something we do a lot of for our clients)
  • Using the segments and summaries, create data-driven customer avatars – these are incredibly informative, and if you don’t have these, you’re leaving money on the table in terms of personalization. You will have avatars that show greater LTV, some with higher purchase frequency, some who help spread the word about your business, etc.
  • Do some research on the demographics of the users your new marketing channel or celebrity will access. Often times this information is available upon request
  • Compare your customer avatars to the marketing demographics and make a decision about how valuable the new initiative really is to the bottom line of the company. If you built the avatars right, this comparison should give you an idea of what you’ll be adding to your customer base with the new marketing initiative. Or, you might find that there is no overlap – RUN!!!

In the case where there’s some overlap and the decision is difficult to make, here are a few more things to consider before moving forward:

  1. Will the boost in traffic/exposure be long-lived enough to justify the work required to learn how to accommodate them?
  2. Is it a good fit for your business?
  3. What will the lifetime value (LTV) of these new customers be?

Companies can do a lot to inform marketing. Data-driven approaches provide some sanity in this space where promises can be as big as the let down that often follows them.

If you’d like some help navigating this space. Please get in touch with us. We’re experts and we love to talk about this stuff!

Zank Bennett is CEO of Bennett Data Science, a group that works with companies from early-stage startups to the Fortune 500. BDS specializes in working with large volumes of data to solve complex business problems, finding novel ways for companies to grow their products and revenue using data, and maximizing the effectiveness of existing data science personnel. https://bennettdatascience.com/staging

Signup for our Newsletter

Tags: