I’m chunking up my reflections on last week’s eMetrics conference in San Francisco into several posts. I’ve got a list of eight possible topics, but I seriously doubt I’ll managed to cover all of them.
One of the first sessions I attended at last week’s eMetrics was Jim Novo’s session titled “The Evolution of an Attribution Resolution.” We’ll (maybe) get to the “attribution” piece in a separate post (because Jim turned on a light bulb for me there), but, for now, we’ll set that aside and focus on a sub-theme of his talk.
Later at the conference, Jennifer Veesenmeyer from Merkle hooked me up with a teaser copy of an upcoming book that she co-authored with others at Merkle called It Only Looks Like Magic: The Power of Big Data and Customer-Centric Digital Analytics. (It wasn’t like I got some sort of super-special hookup. They had a table set up in the exhibit hall and were handing copies out to anyone who was interested. But I still made Jennifer sign my copy!) Due to timing and (lack of) internet availability on one of the legs of my trip, I managed to read the book before landing back in Columbus.
A Long-Coming Shift Is About to Hit
We’ve been talking about being “customer-centric” for years. It seems like eons, really. But, almost always, when I’ve hear marketers bandy about the phrase, they mean, “We need to stop thinking about ‘our campaigns’ and ‘our site’ and ‘our content’ and, instead, start focusing on the customer’s needs, interests, and experiences.” That’s all well and good. Lots of marketers still struggle to actually do this, but it’s a good start.
What I took away from Jim’s points, the book, and a number of experiences with clients over the past couple of years is this:
Customer-centricity can be made much more tangible…and much more tactically applicable when it comes to effective and business-impacting analytics.
This post covers a lot of concepts that, I think, are all different sides of the same coin.
Visitors Trump Visits
Cross-session tracking matters. A visitor who did nothing of apparent importance on their first visit to the site may do nothing of apparent importance across multiple visits over multiple weeks or months. But…that doesn’t mean what they do and when they do it isn’t leading to something of high value to the company.
Caveat (defended) to that:
Does this means visits are dead? No. Really, unless you’re prepared to answer every new analytics question with, “I’ll have an answer in 3-6 months once I see how visitors play out,” you still need to look at intra-session results.
When I asked Jim about this, his response totally made sense. Paraphrasing heavily: “Answering a question with a visit-driven response is fine. But, if there’s a chance that things may play out differently from a visitor view, make sure you check back in later and see if your analysis still holds over the longer term.”
Cohort analysis is nothing more than a visitor-based segment. Now, a crap-ton of marketers have been smoking the Lean Startup Hookah Pipe, and, in the feel-good haze that filled the room, have gotten pretty enamored with the concept. Many analysts, myself included, have asked, “Isn’t that just a cross-session segment?” But “cross-session segment” isn’t nearly as fun to say.
Here’s the deal with cohort analysis:
- It is nothing more than an analysis based around segments that span multiple sessions
- It’s a visitor-based concept
- It’s something that we should be doing more (because it’s more customer-centric!)
The problem? Mainstream web analytics tools capture visitors cross-session, and they report cross-session “unique visitors,” but this is only in aggregate. You can dig into Adobe Discover to get cross-session detail, or, I imagine, into Adobe Insight, but that is unsatisfactory. Google has been hinting that this is a fundamental pivot they’re making — to get more foundationally visitor-based in their interface. But, Jim asked the same question many analysts are:
Having started using and recommending visitor-scope custom variables more and more often, I’m starting to salivate at the prospect of “visitor” criteria coming to GA segments!
Surely, You’ve Heard of “Customer Lifetime Value?”
“Customer Lifetime Value” is another topic that gets tossed around with reckless abandon. Successful retailers, actually, have tackled the data challenges behind this for years. Both Jim and the Merkle book brought the concept back to the forefront of my brain.
It’s part and parcel to everything else in this post: getting beyond, “What value did you (the customer) deliver to me today?” to “What value have you (or will you) deliver to me over the entire duration of our relationship” (with an eye to the time value of money so that we’re not just “hoping for a payoff wayyyy down the road” and congratulating ourselves on a win every time we get an eyeball).
Digital data is actually becoming more “lifetime-capable:”
- Web traffic — web analytics platforms are evolving to be more visitor-based than visit-based, enabling cross-session tracking and analysis
- Social media — we may not know much about a user (see the next section), but, on Twitter, we can watch a username’s activity over time, and even the most locked down Facebook account still exposes a Facebook ID (and, I think, a name)…which also allows tracking (available/public) behavior over time
- Mobile — mobile devices have a fixed ID. There are privacy concerns (and regulations) with using this to actually track a user over time, but the data is there. So, with appropriate permissions, the trick is just handling the handoff when a user replaces their device
And…Finally…Customer Data Integration
Another “something old is new again” is customer data integration — the “customer” angle of of the world of Master Data Management. In the Merkle book, the authors pointed out that the illusive “master key” that is the Achilles heel of many customer data integration efforts is getting both easier and more complicated to work around.
One obvious-once-I-read-it concept was that there are fundamentally two different classes of “user IDs:”
- A strong identifier is “specifically identifiable to a customer and is easily available for matching within the marketing database.”
- A weak identifier is “critical in linking online activity to the same user, although they cannot be used to directly identify the user.”
Cookie IDs are a great example of a weak identifier. As is a Twitter username. And a Facebook user ID.
The idea here is that a sophisticated map of IDs — strong identifiers augmented with a slew of weak identifiers — starts to get us to a much richer view of “the customer.” It holds the promise of enabling us to be more customer-centric. As an example:
- An email or marketing automation system has a strong identifier for each user
- Those platforms can attach a subscriber ID to every link back to the site in the emails they send
- That subscriber ID can be picked up by the web analytics platform (as a weak identifier) and linked to the visitor ID (cookie-based — also a weak identifier)
- Now, you have the ability to link the email database to on-site visitor behavior
This example is not a new concept by any means. But, in my experience, the way each of the platforms involved in a scenario like this has preferred to work is that they set their own strong and weak identifiers. What I took away from the Merkle book is that we’re getting a lot closer to being able to have those identifiers flow between systems.
Again…privacy concerns cannot be ignored. They have to be faced head on, and permission has to be granted where permission would be expected.
Lotta’ Buzzwords…All the Same Thing?
Nothing in this post is really “new.” They’re not even “new to me.” The dots I hadn’t connected was that they are all largely the same thing.
That, I think, is exciting!