Dossier 4.2 – Science of Marketing

Watch George Michie as he explores where machines and algorithms fit in marketing today, and how smart marketers can use this technology.

To download the full Dossier, including George’s article, “Big Data is a Tool for Smart Marketers, Not a Panacea”, click the following link:


George Michie: Hi. I’m George Michie. I’m Co-Founder and Chief Marketing Scientist at RKG. Today, I want to talk a little bit about the science of online marketing and why we scientists are going to be necessary, not just now, but in the future as well.

I was at a conference earlier this year, and the keynote speaker was talking about how the advanced algorithms and software packages that were coming out were going to make all of us number crunchers irrelevant in the near future. Fortunately, I didn’t have any milk in my mouth to blow out of my nose. It was just such an absurd comment, and it jives with a lot of what I’ve learned over the course of many, many years which is oftentimes the people who understand numbers the least trust them the most. If you don’t have somebody who really gets how the algorithms work and how numbers fit together, you end up making bad decisions. A really bad decision would be turning over marketing decisions to a machine.

Let me give you a couple of examples of why we think this way. Bear in mind, RKG is an algorithm developing company. We’ve built a lot of systems for our paid search offering, for our attribution systems, for the way we do comparison shopping engines and display advertising to make use of machines, algorithms, and higher end statistics. There are limits to what those things can do, and let me give you some examples.

In normal business, you want to optimize towards some metric. In the retail environment, you might be optimizing towards orders online. That’s great, and a machine can do that really well. What it can’t do, though, is then understand what the impact of sales value is; that some orders are small-dollar orders and some are large-dollar orders. It’s much better to optimize towards sales rather than orders. That’s great, we can do that. There’s also this notion that some $100 sales don’t have as much margin in them as others do. Some items have very thin margins; so really, it’s even better to optimize towards a margin.

The smart marketer knows that we can optimize towards a margin, but there are times when margin doesn’t tell us about lifetime value. Some low-margin items might have an incredibly high lifetime value that again if you just let the machine optimize to margin, you’re going to miss that. Some things that may not drive a ton of margin will drive a ton of new customers, rather than selling to existing customers which can be hugely valuable, again, from a lifetime value consideration.

There are times when losing money is the smartest thing you can do. Think about perishable merchandise: You have a hotel room that is going to go unsold. Selling it for $50 loses money because it costs more to do that transaction than you would like on average. But if the room is going to go unused, $50 is better than nothing, right? A lot of thought needs to go into how you use the tools. The tools are absolutely essential for doing this stuff well, but you have to have a smart person in charge.

Let’s talk about attribution along those same lines. Facebook ads; fantastic vehicle, but if you’re advertising only to your fan base, you’re advertising to the people who are most likely to buy from you again anyway. If you give credit for every ad on Facebook that somebody saw and placed an order, lo and behold, those ads are going to drive all of your repeat business from your most loyal fans. Loyalty programs in general can create a false sense of productivity. Understanding that difference between what is driving your business, what is incremental to your business versus simply being tracked through a program requires some thought.

I think automation platforms are incredibly important and powerful tools, but just like surgical instruments, very, very good surgical instruments still need very good surgeons to use them. You wouldn’t want to turn them loose on a patient without a surgeon in charge.