Tag Archives: marketing

Connecting Marketers to Machine Learning: A Traveler’s Guide Through Two Utterly Dissimilar Worlds

Artificial Intelligence for Marketing by Jim Sterne

There are people in the world who work with and understand AI and machine learning. And there are people in the world who work with and understand marketing. The intersection of those two groups is a vanishingly tiny population.

Until recently the fact of that nearly empty set didn’t much matter. But with the dramatic growth in machine learning penetration into key marketing activities, that’s changed. If you don’t understand enough about these technologies to use them effectively…well…chances are some of your competitors do.

AI for Marketing, Jim Sterne’s new book,  is targeted specifically toward widening that narrow intersection of two populations into something more like a broad union. It’s not an introduction to machine learning for the data scientist or technologist (though there’s certainly a use and a need for that). It’s not an introduction to marketing (though it does an absolutely admirable job introducing practical marketing concepts). It’s a primer on how to move between those two worlds.

Interestingly, in AI for Marketing, that isn’t a one way street. I probably would have written this book on the assumption that the core task was to get marketing folks to understand machine learning. But AI for Marketing makes the not unreasonable assumption that as challenged as marketing folks are when it comes to AI, machine learning folks are often every bit as ignorant when it comes to marketing. Of course, that first audience is much larger – there’s probably 1000 marketing folks for every machine learner. But if you are an enterprise wanting two teams to collaborate or a technology company wanting to fuse your machine learning smarts to marketing problems, it makes sense to treat this as a two-way street.

Here’s how the book lays out.

Chapter 1 just sets the table on AI and machine learning. It’s a big chapter and it’s a bit of grab bag, with everything from why you should be worried about AI to where you might look for data to feed it. It’s a sweeping introduction to an admittedly huge topic, but it doesn’t do a lot of real work in the broader organization of the book.

That real work starts in Chapter 2 with the introduction to machine learning. This chapter is essential for Marketers. It covers a range of analytic concepts: an excellent introduction into the basics of how to think about models (a surprisingly important and misunderstood topic), a host of common analytics problems (like high cardinality) and then introduces core techniques in machine learning. If you’ve ever sat through data scientists or technology vendors babbling on about support vector machines and random forests, and wondered if you’d been airlifted into an incredibly confusing episode of Game of Drones, this chapter will be a godsend. The explanations are given in the author’s trademark style: simple, straightforward and surprisingly enjoyable given the subject matter. You just won’t find a better more straightforward introduction to these methods for the interested but not enthralled businessperson.

In Chapter 3, Jim walks the other way down the street – introducing modern marketing to the data scientist. After a long career explaining analytics to business and marketing folks, Jim has absorbed an immense amount of marketing knowledge. He has this stuff down cold and he’s every bit as good (maybe even better) taking marketing concepts back to analysts as he is working in the other direction.  From a basic intro into the evolution of modern marketing to a survey of the key problems folks are always trying to solve (attribution, mix, lifetime value, and personalization), this chapter nails it. If you subscribe to the theory (and I do) that any book on Marketing could more appropriately have been delivered as a single chapter, then just think of this as the rare book on Marketing delivered at the right length.

If you accept the idea that bridging these two worlds needs movement in both directions, the structure to this point is obvious. Introduce one. Introduce the other. But then what?

Here’s where I think the structure of the book really sings. To me, the heart of the book is in Chapters 4, 5 and 6 (which I know sounds like an old Elvis Costello song). Each chapter tackles one part of the marketing funnel and shows how AI and machine learning can be used to solve problems.

Chapter 4 looks at up-funnel activities around market research, public relations, social awareness, and mass advertising. Chapter 5 walks through persuasion and selling including the in-store journey (yeah!), shopping assistants, UX, and remarketing. Chapter 6 covers (you should be able to guess) issues around retention and churn including customer service and returns. Chapter 7 is a kind of “one ring to rule them all”, covering the emergence of integrated, “intelligent” marketing platforms that do everything. Well….maybe. Call me skeptical on this front.

Anyway, these chapters are similar in tone and rich in content. You get the core issues explained, a discussion of how AI and machine learning can be used, and brief introductions into the vendors and people who are doing the work. For the marketer, that means you can find the problems that concern you, get a sense of where the state of machine learning stands vis-à-vis your actual problem set, and almost certainly pick-up a couple of ideas about who to talk to and what to think about next.

If you’re into this stuff at all, these four chapters will probably get you pretty excited about the possibilities. So think of Chapter 8 as a cautionary shot across the bow. From being too good for your own good to issues around privacy, hidden biases and, repeat after me, “correlation is not causation” this is Pandora’s little chapter of analytics and machine learning troubles.

So what’s left? Think about having a baby. The first part is exciting and fun. The next part is long and tedious. And labor – the last part – is incredibly painful. It’s pretty much the same when it comes to analytics. Operationalizing analytics is that last, painful step. It comes at the end of the process and nobody thinks it’s any fun. Like the introduction to marketing, the section on operationalizing AI bears all the hallmarks of long, deep familiarity with the issues and opportunities in enterprise adoption of analytics and technology. There’s tons of good, sound advice that can help you actually get some of this stuff done.

Jim wraps up with the seemingly obligatory look into the future. Now, I’m pretty confident that none of us have the faintest idea how the future of AI is going to unfold. And if I really had to choose, I guess I prefer my crystal ball to be in science fiction form where I don’t have to take anything but the plot too seriously. But there’s probably a clause in every publisher’s AI book contract that an author must speculate on the how wonderful/dangerous the future will be. Jim keeps it short, light, and highly speculative. Mission accomplished.

 

Summing Up

I think of AI for Marketing as a handy guidebook into two very different, neighboring lands. For most of us, the gap between the two is an un-navigable chasm. AI for Marketing takes you into each locale and introduces you to the things you really must know about them. It’s a fine introduction not just into AI and Machine Learning but into modern marketing practice as well. Best of all, it guides you across the narrow bridges that connect the two and makes it easier to navigate for yourself.  You couldn’t ask for a wiser, more entertaining guide to walk you around and over that bridge between two utterly dissimilar worlds that grow everyday more necessarily connected.

 

Full Disclosure: I know and like the author – Jim Sterne – of AI for Marketing. Indeed, with Jim the verbs know and like are largely synonymous. Nor will I pretend that this doesn’t impact my thoughts on the work. When you can almost hear someone’s voice as you read their words, it’s bound to impact your enjoyment and interpretation. So absolutely no claim to be unbiased here!

 

Segmentation is the Key to Marketing Analytics

The equation in retail today is simple. Evolve or die. But if analytics is one of the core tools to drive successful  evolution, we have a problem. From an analytics perspective, we’re used to a certain view of the store. We know how many shoppers we get (door counting) and we know what we sold. We know how many Associates we had. We (may) know what they sold. This isn’t dog food. If you had to pick a very small set of metrics to work with to optimize the store, most of these would belong. But we’re missing a lot, too. We’re missing almost any analytic detail around the customer journey in the store. That’s a particularly acute lack (as I noted in my last post) in a world where we’re increasingly focused on delivering (and measuring) better store experiences. In a transaction-focused world, transactions are the key measures. In an experience world? Not so much. So journey measurement is a critical component of today’s store optimization. And there’s the problem. Because the in-store measurement systems we have available are tragically limited. DM1, our new platform, is designed to fix that problem.

People like to talk about analytics as if it just falls out of data. As if analysts can take any data set and any tool and somehow make a tasty concoction. It isn’t true. Analytics is hard work. A really great analyst can work wonders, but some data sets are too poor to use. Some tools lock away the data or munge it beyond recognition.  And remember, the most expensive part of analytics is the human component. Why arm those folks with tools that make their job slow and hard? Believe me, when it comes to getting value out of analytics, it’s hard enough with good tools and good data. You can kid yourself that it’s okay to get by with less. But at some point you’re just flushing your investment and your time away. In two previous posts, I called out a set of problems with the current generation of store customer measurement systems. Sure, every system has problems – no analytics tool is perfect. But some problems are much worse than others. And some problems cripple or severely limit our ability to use journey data to drive real improvement.

When it comes to store measurement tools, here are the killers: lack of segmentation, lack of store context, inappropriate analytics tools, inability to integrate Associate data and interactions, inability to integrate into the broader analytics ecosystem and an unwillingness to provide cleaned, event-level data that might let analysts get around these other issues.

Those are the problems we set out to solve when we built DM1.

Let’s start with Segmentation. Segmentation can sound like a fancy add-on. A nice to have. Important maybe, but not critical.

That isn’t right. Marketing analytics just is segmentation. There is no such thing as an average customer. And when it comes to customer journey’s, trying to average them makes them meaningless. One customer walks in the door, turns around and leaves. Another lingers for twenty minutes shopping intensively in two departments. Averaging the two? It means nothing.

Almost every analysis you’ll do, every question you’ll try to answer about store layout, store merchandising, promotion performance, or experience will require you to segment. To be able to look at the just the customers who DID THIS. Just the customers who experienced THAT.

Think about it. When you build a new experience, and want to know how it changed behavior you need to segment. When you change a greeting script or adjust a presentation and want to know if it improved store performance you need segmentation. When you change Associate interaction strategies and want to see how it’s impacting customer behavior you need segmentation. When you add a store event and want to see how it impacted key sections, you need segmentation. When you want to know what other stuff shoppers interested in a category cared about, you need segmentation. When you want to know how successful journeys differed from unsuccessful ones, you need segmentation. When you want to know what happens with people who do store pickup or returns, you need segmentation.

In other words, if you want to use customer journey tracking tools for tracking customer journeys, you need segmentation.

If your tool doesn’t provide segmentation and it doesn’t give the analyst access to the data outside it’s interface, you’re stuck. It doesn’t matter how brilliant you are. How clever. Or how skilled. You can’t manufacture segmentation.

Why don’t most tools deliver segmentation?

If it’s so important, why isn’t it there? Supporting segmentation is actually kind of hard. Most reporting systems work by aggregating the data. They add it up by various dimensions so that it can be collapsed into easily accessible chunks delivered up into reports. But when you add segmentation into the mix, you have to chunk every metric by every possible combination of segments. It’s messy and it often expands the data so much that reports take forever to run. That’s not good either.

We engineered DM1 differently. In DM1, all the data is stored in memory. What does that mean? You know how on your PC, when you save something to disk or first load it from the hard drive it takes a decent chunk of time? But once it’s loaded everything goes along just fine? That’s because memory is much faster than disk. So once your PowerPoint or spreadsheet is loaded into memory, things run much faster. With DM1, your entire data set is stored in-memory. Every record. Every journey. And because it’s in-memory, we can pass all your data for every query, really fast. But we didn’t stop there. When you run a query on DM1, that query is split up into lots of chunks (called threads) each of which process its own little range of data – usually a day or two. Then they combine all the answers together and deliver them back to you.

That means that not only does DM1 deliver reports almost instantaneously, it means we can run even pretty complex queries without pre-aggregating anything and without having to worry about the performance. Things like…segmentation.

Segmentation and DM1

In DM1, you can segment on quite a few different things. You can segment on where in the store the shopper spent time. You can segment on how much time they spent. You can segment on their total time in the store. You can segment on when they shopped (both by day of week and time of day). You can segment on whether they purchased or not. And even whether they interacted with an Associate.

If, for example, you want to understand potential cross-sells, you can apply a segment that selects only visitors who spent a significant amount of time shopping in a section or department. Actually, this undersells the capability because it’s in no way limited to any specific type of store area. You can segment on any store area down to the level of accuracy achieved by the collection architecture.

What’s more, DM1 keeps track of historical meta-data for every area of the store. Meaning that even if you changed, moved or re-sized an area of the store, DM1 still tracks and segments on it appropriately.

So if you want to see what else shoppers who looked at, for example, Jackets also considered, you can simply apply the segmentation. It will work correctly no matter how many times the area was re-defined. It will work even in store roll-ups with fundamentally different store types. And with the segment applied, you can view any DM1 visualization, chart or table. So you can look at where else Jacket Shoppers passed through, where they lingered, where they engaged more deeply, what else they were likely to buy, where they exited from, where they went first, where they spent the most time, etc. etc. You can even answer questions such as whether shoppers in Jackets were more or less likely to interact with Sales Associates in that section or another.

Want to see if Jacket shoppers are different on weekdays and weekends? If transactors are different from browsers? If having an Associate interaction significantly increases browse time? Well, DM1 let’s you stack segments. So you can choose any other filter type and apply it as well. I think the Day and Time part segmentation’s are particularly cool (and unusual). They let you seamlessly focus on morning shoppers or late afternoon, weekend shoppers or even just shoppers who come in over lunchtime. Sure, with door-counting you know your overall store volume. But with day and time-part segmentation you know volume, interest, consideration, and attribution for every measured area of the store and every type of customer for every hour and day of week.

DM1’s segmentation capability makes it easy to see whether merchandise is grouped appropriately. How different types of visitor journeys play out. Where promotional opportunities exist. And how and where the flow of traffic contradicts the overall store layout or associate plan. For identified shoppers, it also means you can create extraordinarily rich behavioral profiles that capture in near real-time what a shopper cares about right now.

It comes down to this. Without segmentation, analytics solutions are just baby toys. Segmentation is what makes them real marketing tools.

The Roadmap

DM1 certainly delivers far more segmentation than any other product in this space. But it’s still quite a bit short of what I’d like to deliver. I mean it when I say that segmentation is the heart and soul of marketing analytics. A segmentation capability can never be too robust.

Not only do we plan to add even more basic segmentation options to DM1, we’ve also roadmapped a full segmentation builder (of the sort that the more recent generation of digital analytics tools include). Our current segmentation interface is simple. Implied “ors” within a category and implied “ands” across segmentation types. That’s by far the most common type of segmentation analysts use. But it’s not the only kind that’s valuable. Being able to apply more advanced logic and groupings, customized thresholds, and time based concepts (visited before / after) are all valuable for certain types of analysis.

I’ve also roadmapped basic machine learning to create data-driven segmentations and a UI that provides a more persona-based approach to understanding visitor types and tracking them as cohorts.

The beauty of our underlying data structures is that none of this is architecturally a challenge. Creating a good UI for building segmentations is hard. But if you can count on high performance processing event level detail in your queries (and by high-performance I mean sub-second – check out my demos if you don’t believe me), you can support really robust segmentation without having to worry about the data engine or the basic performance of queries. That’s a luxury I plan to take full advantage of in delivering a product that segments. And segments. And segments again.