In one of my favorite presentations from the past year (Unlocking Excellence – Analytics as the Key to Digital), I walk through some of the most important lessons I've learned when it comes to doing digital analytics well. One of those lessons, headlined “Don’t Look at Data – Use It," argues for the central importance of personalization when it comes to digital. It's central because the most important thing to understand about digital is this: digital isn’t a mass medium. Effectiveness in digital, regardless of your business model, is about delivering a maximally relevant and therefore highly personalized experience. Think about it. The companies that succeed in digital, whether in eCommerce, Hospitality and Travel, Social, or Media, are unique in their ability to drive and create a highly personalized experience.
This isn’t, fundamentally, a point about analytics. It’s about business model, and it applies to digital business models across the entire spectrum of applications. We tend to think of mass media and direct as alternative channels for marketing spend. That certainly holds in digital (it’s a direct response world when it comes to digital marketing spend), but it isn’t just limited to marketing spend.
Great digital eCommerce sites don’t deliver a monolithic experience. Out there in the physical world there are all sorts of storefronts – each with a distinct look and feel. At the bricks-and-mortar level, Walmart, Trader Joes, and Neiman Marcus are massively different but carefully crafted storefronts. Still, each store is the same for every visitor. The shelves inside a Trader Joes don’t re-arrange themselves when you walk in the door. But that’s exactly what happens in a great digital experience.
It’s the same story everywhere. In traditional publishing, targeting exists only at a mass level. The NY Times and Cosmopolitan presumably target a different audience. But the experience they deliver is the same for each member of that audience. Because of that, targeting is a very secondary variable to content in the traditional publishing world. The guy with the best content pretty much always wins.
In digital, that paradigm gets flipped. Content doesn’t go away in digital. In fact, it’s more important than targeting is in mass. Given remotely equal content, however, it’s the guy with the best targeting who wins. If you can deliver what people care about most to their eyeballs, you’ll get more share.
If you think about almost any of the digital experiences that get noticed, that we respond to, that define best of breed, this focus on personalized experience and maximal relevancy underpins them all.
Which brings me to the analytics story; because if relevancy underpins digital, analytics drives relevancy. You can’t deliver relevancy without analysis.
Mind you, there are all sorts of strategies for delivering relevancy. Variations on traditional market basket analysis underlie many of the most famous ecommerce experiences. Traditional segmentation and rule based personalization drive some of the most famous publishing and social experiences. Crowd-sourcing methods are used in many verticals but especially in travel and hospitality where they are often the focus of the entire experience.
You can't deploy these techniques without some form of analytics.
Indeed, it’s pretty much impossible to imagine personalization without analytics. That’s why analytics is fundamental to digital in a way that cuts deeper than in many other businesses. You can be good, even great, at many a business without analytics. Because digital is fundamentally direct and because direct places that premium on targeting and personalization, I just don’t think that’s true with digital. Great mass creative won’t beat decent targeting.
So why isn’t personalization more prevalent in digital? While we see leaders in digital chasing ever deeper levels of personalization, the rest of the market hasn’t caught up and isn’t even clearly in the hunt. I think there are a couple reasons driving this failure:
The technology to drive personalization isn’t trivial and off-the-shelf solutions are few and far between. Most of the companies that have done exceptional personalization have created it on systems that were entirely home-baked. Indeed, this is a rare instance where the existence of home-baked platforms embedded deep in the business have actually supported not restricted innovation.
The absence of commercial solutions is changing slowly. Testing tools are evolving into more fully-baked personalization tools. CMS systems have significantly expanded their capabilities to store customer profiles and provide rule-based personalization. Several fully-baked digital personalization solutions have hit the market. And analytics systems are evolving that better support real-time decisioning. None of these systems are quite at the level that provides an enterprise with a warm-fuzzy technology choice, but the situation has improved quite a bit and should get rapidly better.
Of course, technology trends mirror business demands. If business stakeholders had really understood the value of personalization in digital then all those brilliant folks writing software would have supported the need. Despite the many obvious examples among the leaders in the digital space, most enterprises have been disappointingly slow in learning the real lessons. They tend to think that those business models just don’t apply to them. That’s not without foundation – you can’t just assume that because an online retailer with millions of products makes great use of market-basket like analytics that it’s what you need too. It’s the deeper lesson – that driving personalized experience’s differentiates – that often goes under-appreciated.
Finally, methodology matters and this, too, has been a sore point. In my last post I talked about how Use-Case and Re-survey methodologies were particularly appropriate to support continuous cycles of improvement. This need to fit analytics method to problem is frequently missed. In digital, I’ve already mentioned three very popular methods for driving personalization. Market-basket techniques underlie most Next-Best-Offer situations. Crowd-sourcing underlies many filtering and rating situations. Segmentation drives most rule-based applications and I think it's the most common and important tool for digital personalization. I've written extensively about how 2-tiered segmentation works, but there are other aspects of digital segmentation worth discussing in the context of personalization.
Sometimes, you’ll need all three techniques together to drive appropriate personalization. There are, also, analytic methods around offer matching (finding the best offer-fit for a given consumer) that require modeling techniques that go well beyond market-basket to find the best combination of offer-to-consumer to optimize take-up, inventory, and margin.
Lacking significant experience in or understanding of these methodologies has crippled more than one personalization effort.
I hope, in the next few posts, to think through some of these issues around technology and methodology. Along the way, I hope to build enough of that crucial understanding to help drive an enterprise toward doing more with analytics than looking at numbers – to tackling the fundamental challenges and opportunities in digital and to deliver a truly differentiated experience.