In my last post in the series, I introduced the concept of a Two-Tiered Segmentation. It’s a distinctively digital approach to segmentation. The first tier is a classic Database Marketing segmentation by visitor-type. The second tier is distinctly digital – it’s a segmentation by visit type/intent. The beauty of the Two-Tiered Segmentation is that it frames every metric, KPI, and use-case in terms of the two most important business dimensions: who are we talking about, and why were they here. In doing this, it creates a powerful framework for generating KPIs, testing scenarios, doing analysis, and driving personalization and outbound marketing.
So how do you do it?
In concept, the Two-Tiered segmentation is strikingly easy but it can be challenging to implement.
We’ve done quite a number of these now, and the concept applies to nearly any type of Website, but I thought I’d start by showing a Financial Services example – in this case, a non-trading brokerage site.
We start by defining the first-tier (visitor-type) of the segmentation. For Financial Services, this first tier nearly always consists of AT LEAST two levels. The first level is the business relationship. For this site, the high-level Customer/Prospect segmentation was broken up into four major sub-types : Advisors, Plan Managers, High-Wealth Investors and General Investors.
It’s extremely common in Financial Services to further qualify segments by either Potential or Actual relationship value. The split between High-Wealth and General Investors already captures this to some extent, but we’d probably include it as a further sub-segmentation. For existing customers, we might choose to have both an Actual Relationship Value and an Opportunity Value segmentation since both are useful in different contexts:
As you can see, the tree gets fairly complex in short-order. Is it too complex? That’s a difficult question to answer in the abstract. For a few uses, Customer/Prospect may be appropriate. For more uses, the next division (Advisor/Plan Manager/etc.) is probably the right one. For many DB Marketing purposes and a fair number of reporting purposes, qualifying a segmentation with at least one value dimension is appropriate. The right level of qualification depends on the business use.
In my last post, I said that if someone comes to me and says “Our traffic is up,” the first question I want to ask is “With whom?”
The segmentation I’ve been describing is designed to provide a framework for the right answer. In most cases, the right answer isn’t “Customers” – that’s not specific enough to really help me understand the change. “Current Customers who are High-Wealth Investors” is much more likely to be a good answer. If I’m primarily focused on Customer Support, then the answer I’m looking for might be something like “High-Wealth Investors who are High-Value Current Customers” or “All Current Customers who are High-Wealth Investors”. If I’m focused on marketing, the right answer is probably something like “High-Wealth Investors with High Opportunity.”
I don’t think this is too complex. If you know you’re business, then this type of answer is what you should be expecting. In trying to dumb down and simplify our numbers, we’ve often stripped out the meat not the fat. A segmentation adds apparent complexity to reporting and analysis but by tying the numbers to the business it creates real understanding.
This is clarity, not complexity.
So far, we’ve only addressed the first tier (Visitor Type) of the segmentation. Now it’s time to add the 2nd-Tier – Visit Type.
If I know that traffic is up with "Customers who are High-Value Advisors," I’m probably going to be pleased. Unless I find out that they are using the site more because of an operational problem that’s making them work harder to get their job done.
Now I’m worried.
It isn’t enough, in the digital world, to understand who someone is. When you look at a customer touchpoint (and every Web visit is a touchpoint), you need to understand what the touch was for to understand the business meaning and whether or not the touch was successful.
This is so obvious that it’s hard to believe that designers of Web Report Sets and KPIs have managed to get it almost universally wrong.
A successful touch for an Advisor seeking customer support isn’t captured by an acquisition, a registration, or even a buy order. None of these are remotely relevant. For a current Advisor seeking information on a specific fund, a buy order (if it’s possible on site), may indeed be a measure of success.
Personalization opportunities, KPIs, and metrics are all relative to Visit Intent as well as Visitor Type. In other words, segment-wide KPIs don’t exist – any more than site-wide KPIs - and every success rate on your site needs to be Visit-Type specific.
Creating a Visit-Type Segmentation is usually harder than a Visitor Segmentation. Visitor Types are deeply embedded in the business consciousness. Everybody understands them. Some Visit Types are equally obvious – but others can be obscure. Here’s a sample (somewhat cut down) Visit-Type segmentation for a Financial Services Brokerage Site:
This is a very high-level view, but as it stands it provides powerful context around any question when combined with the Visitor Type Segmentation.
Q: “Why is traffic up?”
A: “Traffic is up because we are getting more visits from High-Wealth Prospects who are doing Research.”
That’s an answer!
For high-level reporting, this might well be sufficient as a visit-type segmentation. For the Acquisition Manager or the Advisor Support manager, however, it isn’t going to be enough detail. We’d typically create at least one more sub-dimension under each high-level visit-type for more sophisticated reporting or analysis:
Once you’ve got this two-tiered segmentation in place, it becomes the basis for nearly everything you do. It anchors your KPIs:
I hope, at this point, that every Visitor/Visit Type having a distinct success metric feels inevitable not surprising.
It's also the framework for all your testing and personalization:
Testing strategies should emerge and be specifically tailored for target segments in specific types of visits. Site-wide tests should be almost as rare as site-wide KPIs!
It should be at the heart of your Management Reporting - and I'll show examples of this in a subsequent post dedicated specifically to this topic. And, finally, it’s the framework for analysis. A huge chunk of the analysis that we deliver these days begins with a Two-Tiered Segmentation and focuses on problems insider or across those segments.
I hope that at this point, you’re thinking to yourself that this makes enormous sense but you’re worried about whether or not it can actually be done. It isn’t immediately obvious how you can categorize visits on a Web site by Visit Intent – meaning that the all important 2nd Tier is in jeopardy.
In my next post, I’ll show how we at Semphonic go about building this Two-Tiered Segmentation and then incorporating it into your Digital Analytics. After that, I’ll go on to show how the same Two-Tier Segmentation concept works in a variety of other industries and applications – including places like Media, Public Sector, and Customer Support where you might wonder if it is applicable.
Hello, as an avid follower of your posts I don't remember being quite so excited about a series of posts as I am now, and I have been quite excited before.
I work for a media website so looking forward to a breakdown of how you have done those. I remember a comment of yours in a post a couple of years ago where you mentioned that for media sites you would segment visitors for different levels of activity. I am doing this, but the problems start when you try to identify the forces that are moving people up or down this engagement ladder.
Best, Chico
Posted by: Chicoli | April 27, 2011 at 09:04 AM
very pecisely made schematic graphs and very useful information . .
Jamie Dingman Shipston Group
Posted by: Mary Bush | July 20, 2011 at 02:01 AM
When creating a strategy it should be like this every analysis required basis. A skillful way in managing will lead to direct success.
Jamie Dingman
Posted by: Suzanne Twitchell | August 01, 2011 at 11:31 PM