What happens on my site? This is what we web analysts call a core question. It’s harder to answer than it looks, even though site analytics programs are supposedly designed to answer it. Traditional analytics tools answer it with descriptive numbers, such as:
- 20% of your site’s visits looked at the About Us page at some point, and
- the average visit was about five pages long
That doesn’t really tell us what people are “doing.” It doesn’t tell a story. It doesn’t get you inside the visitor’s head at the moment of the visit. A much better answer with story-like overtones might be:
We looked at all your site’s recent visits and we found basically five kinds of visits.
- One type wants to know where your stores are because these visitors already have settled on your brand.
- Another wants to know about your products.
- Another type of visit isn’t checking products; instead these visits involve curiosity about your product category and are looking for ideas or tips.
- Another visit type is simply checking to see if you have anything new.
- And yet another is focused entirely on deals and discounts.
The rest of your visits wander all over the place and don’t seem to have a focus or mission.
Now, that’s a story! A very useful story.
If you can come into possession of that wonderful list of five kinds of visits, custom-extracted from your own site data, you can use it as a checklist:
- Does your site’s navigation and content work well for each visit type?
- Do you have a marketing message that will resonate with each one?
- Do you place your marketing messages where each one will likely be seen by—and resonate with—each visit type?
Going a little further, you can estimate the number of visits of each type and do some metrics:
- Evaluate marketing actions by whether they bring in a predominance of the most desired visit type more than other marketing actions do.
- Evaluate site changes by whether the proportions change the way you want them to – whether you are successful at encouraging one type of visit to become another.
And, if you really want to take it far, you can ask yourself these kinds of questions:
- What about those wandering visits? Are they potentially one of the focused visit types? Are they happy experiences of footloose visitors, or evidence of frustrated and confused experiences?
- Can you attribute value to each type? Can you arrange them into a maturity model, or a purchase-decision model?
<pause for breath>
I can guess what you’re thinking —
- “We already created visit types for our site (or marketing) strategy, so we can ignore all this”
- Good. Incorporating visit types into your guiding vision is absolutely essential to a strategy. But I have to ask — have you since checked those visit types with actual data? Will the data confirm that you visualized them correctly, and that you identified all of them?
- “We already have personas, which are demographic-lifestyle-lifestage descriptions of likely visitor groups, so this doesn’t apply to us”
- Visit types are a different animal, neither more granular nor less granular than personas. The key point of visit typing is that a given persona-person will use a spectrum of visit types over the course of their relationship with you. Your site and your marketing should be ready for as many of those instances as possible.
- “But one person could then have five visits, each a different type!”
- Yes. Isn’t that great? You get five ways to reach that person!
- “I get pages and pages of site statistics and trendlines every month, including side-by-side testing results and so on. Where are these ‘types’ in those reports?”
- The mainstream site analytics tools don’t provide this kind of discovered information. Adobe SiteCatalyst by Omniture, WebTrends Analytics, and Google Analytics have what they call segmenting, but although the word “segmenting” resembles “visit types” the resemblance is superficial. Segments as used by the analytics tools are filters, where you divvy up visits according to definitions you already have in mind, and then the analytics tools provide statistics for those divisions. Instead of divisions you assume are worth looking at (first time visits versus return visits, or search visits versus email visits) I am talking about discovering groups of similar-looking visits from the data.
- “That sounds like cluster analysis, maybe?”
- Yes, more or less. The gist of cluster analysis is getting the data to show you whether granular data can be clumped into groups of similar-ish items. In this case the granular data is your set of site visits, and similar-ish refers to “looking at the similar content on my site.”
That brings us back to what we’re doing in our Marketing Intelligence group. We really enjoy this particular analysis. We take raw site visit behavior – namely, thousands and thousands of individual visits with their individual lists of pages-viewed – and we work to get the data to tell us what the visit types are. We sometimes end up presenting crazy diagrams like the one below. If possible, we also host a retreat or some really concentrated group discussion. The assumptions and beliefs (about customers as well as about the web site) that are uncovered in those discussions are almost always eye-opening. All in all, we get more a-ha’s and useful insights than we usually get from plain cross-section statistics. And, we hope, “what happens on my site” starts getting a real answer.
