A Few Thoughts After Another Awesome eMetrics….
Back from another excellent eMetrics. I’m a very big fan of the eMetrics Marketing Optimization Summit… Props go to Jim Sterne for growing this event from a little seed into an incredible, blogworthy blossom. How involved is Jim in eMetrics? I’d say he’s completely immersed in every little piece - he even came up to me at the SF WAW (way to go June D!) to find out about the renegade AV work I did in one of the sessions, and to get my take on how it could have been avoided. He’s that intimately connected to what’s going on. Macro and micro, micro and macro. And when you have one of the best Internet Marketers in the world, keeping a tight rein on the Clydesdale of conferences, you know you’re in for one heck of fun ride.
And so it was for about 500+ of the top web analytics in the beautiful Palace hotel. Props to consummate conference organizers Matt Finlay and his crew at Rising Media for keeping the road smooth as we all trotted on it as well. Fanny, you are one helpful polyglot of a marketing manager! I never knew German keyboards were so wild… Thanks.
The eMetrics sessions were informative and actionable. The lobby bar and after-hours parties fun and enlightening. You really can’t ask for more out of a conference. As I flew home thinking back on it all, there was a lot to blog about, including:
- It’s all about attitude, dude – as in attitudinal data. Like my father says “it’s all about your attitude.” And so it is on the Internet in 2008. From ForeSeeResults, to iPerceptions, to OpinionLab, to CRMMetrix, the often missing link in customer analytics is attitudinal data. I’m talking here about Voice of Customer (VOC) technology that allows you to ask a question set to site visitors and then apply some sort of algorithm or model to express the meaningfulness of the data in quantifiable terms. From the American Customer Satisfaction Index to 4Q. VOC technology enables you to participate in a continuous, automated dialog with your customers in order to identify problem points on your web site and enable you to measure purpose and success of your most valuable segments. Expect to see some of the big players gobble up these smaller companies. Omniture, Unica, WebTrends, and CoreMetrics should be thinking about acquisition in this space to round out their offerings.
- Testing, 123… as in multivariate, MVT. The rage is site optimization technologies beyond the simple A/B, champion challenger, test. In this category you find folks like SiteSpect (the only non-intrusive multivariate testing solution!). I’m a big fan of these guys (and was in 2006 long before they ever sponsored a WAW, thanks to a nice demo from Larry at my old job). Eric Hansen and his crew have specialized software that you install in your data center. No futzing with damned tags. Swap out your variations, create different recipes, determine what’s statistically significant in giving you a lift to your macro or micro conversion goal, and you’re off to the races. The good folks at Google are doing it and doing it well with Google Site Optimizer (thanks for the t-shirts!). Interwoven is baking in Optimost to the CMS, and Omniture has their Test and Target integrated with the Business Optimization Suite. Accenture has Memetrics. Kefta too. And what ever happened to Verster?
In a nutshell, these technologies enable you to test variations of content themes, colors, creative, calls to action, points of resolution, buttons, navigational elements, –whatever you want to call the stuff on the screen—to determine what combination performs best against your goals. But of course, this is all just software, so don’t get too excited. The tests are about as good as the people creating them… And complex tests that take a long time to execute may not finish. Imagine 1-800-Flowers starting a test in January and not finishing until March, missing Valentine’s Day. Or Intuit running a test beyond April 15th for a tax product. Go humbly and carefully into this space, my friends, or you may end up optimizing for everyone and appealing to none.
- Tying it all back to the dollar for profit-generating sites and to the mission of non-profit generating sites… It seems like a “no, duh” moment but metrics for the sake of metrics can be a big waste of time. If you can’t tie metrics or visitor actions back to value on a revenue-producing site or to the betterment of a non-profit site’s core mission, then what’s really the point of the measurement… That’s why I’m a big fan of the stuff ZaaZ does. They totally get the fact of how actionable metrics turn the wheel of Internet commerce and ad-based models, and they can model it all to prove it out the ROI. Folks like newly elected WAA Director Alex Langshur’s company Public InSite do similar stuff for content driven sites. That is they know how to use metrics to optimize the channel to goals, not to just puke confusing data, like most web analytics tools do. Again, it’s all about the people you hire, not the tools you use… My good friend Avinash, right again!
- The emergence and rise of deeply psychological and neuro-behavioral methods for automating persuasion and conversion. Anyone who knows my good friend Joseph Carrabis, over at NextStage Evolution, knows that besides being one heck of giant kite flying, music master, he’s also got the models and the patents to help target and respond to human behavior across programmable devices. We’re already seeing some companies, like Seven Billion Joe’s, er People, taking what he’s been saying for years and going to market with it. The idea here being that if you can identify the affective, behavior, and motivational drivers of site visitors, you can maximize cognition in elements on the site (like pictures, text, informational flow) to appeal to target segments and persuade/provoke desired behavior. It’s like a higher rung on the optimization ladder. It’s not test what they see, it’s figure out how they think, then make the site better because of it. Cool stuff. Blows my mind.
- Integrated, multichannel marketing. Just ask my good friend Akin Arikan, author of the newly released Multichannel Marketing. (Disclaimer: I was a technical editor on the book. It’s easy to do when you edit brilliance). Make sure to check it out! Marketing in general will become more Internet-centric, but will continue to clutch the roots of broadcast and print. You will have the database marketer and statistical modelers working with a union of web channel and offline data. What’s preventing it now? A unified marketing database. You see companies like Salford Systems circulating in this space. And take a look at Unica’s blend of Enterprise Marketing Management… I’d stay tuned to see what Unica has up their sleeve for bringing together online and offline. When you can segment and target across online and offline campaigns, if I were pure web channel player only, like Omniture or CoreMetrics, I’d be a bit concerned that people are waking up to open systems, not closed black boxes. WebTrends is already moving in this direction… But they all remain far behind Unica when it comes to multichannel marketing.
And that’s just a few of the things the phenomenal eMetrics got me thinking about… I hope to see you in Washington DC in October!
Greg Moore added the following ...
Another trend at eMetrics I noticed was discussion of the “last click” problem, where your conversion is credited to the last thing a user did, and what went before is ignored.
JP, do you think this is new, or is it something people have talked about for a while? How “solved” do you think the problem is? Is it significant?
Wondering what you think….
Greg Moore
Jim Novo added the following ...
Akin’s book is great, you should be proud of your participation in that project!
I’d advise all the “average attitude” is just as dangerous a piece of data as the “average visit” - make sure you know the behavior of those giving you ‘tude. Are they best visitors / customers? Where did they come from / what did they do / what happened to them that generated this attitude? The real power of attitudinal data is not in the data itself, but answering the question “why do they think this way?”
Personally, I prefer “first click” to last click because it’s more predictive and the leverage is higher in terms of optimizing campaigns - the first click is the hardest to get, right? Because every click after the first benefits from the first…
Judah, see you in DC!
Judah added the following ...
Tom: Thank you. I take that especially to hear since you work for WebTrends! 
Greg: Yeah, the whole “attribution” issue is a big deal, especially across sessions. There’s this idea of “pan session analysis” going around right now because the last click generally isn’t what compelled you to purchase (but not always).
To answer your questions, no it is not new at all. Maybe to web analytics folks it is, but direct marketers have been dealing with this for years. Direct marketers have this concept called “matchback reponse attribution” which has an analogy in online advertising called “view through.” Essentially a company attributes the response (i.e. conversion event) to some previous customer touchpoint/contact, like an ad they sent you last week or some direct mail offer. They match your purchase back to the ad, if they can, using some method, perhaps a coupon you use or a loyalty card at checkout instead of just thinking you walked in and bought the item randomly.
For example, you were mailed an ad for “buy 1 get one free Brita water filters” so you go to buy it and use your customer loyalty card in the checkout transaction. In the backend, the company matches back your purchase to the ad they sent via the customer loyalty card. They match it back. That’s a pretty simple example.
In the online world, where I live right now, “view through” (a function of ad servers) takes into account the display ads (i.e. banners, skyscrapers, boom boxes etc) you saw before you made the purchase, and the ad server attributes the conversion event back to the ad you first saw, not the last ad you saw.
In web analytics, certain tools, like Unica NetInsight and I think CoreMetrics, allow you measure the concept of “initial referrer” or the referrer that you came from when you first visited the site. Of course, all this online stuff requires persistent cookies, third party persistent cookies in the case of ad servers, so it’s a tough marketing science right now in the online world. Even persistent cookies are way less persistent than the loyalty card you carry in your wallet for your local supermarket.
So to continue my answer to your question, I don’t think the problem is at all solved in the online world, just a somewhat solved under the right conditions for certain segments when the company uses the right tools. And yes it is a very significant issue that needs more attention. The company that solves it will be printing money.
Hope that helps clarify things a bit (and my quick typing made sense!). Thanks for reading my blog and for your comment.
Jim: I am honored you would leave a comment on my blog. Your last comment was like a year ago! I wholeheartedly agree with you. Very smart, savvy, and wise advice. Your point being to segment the attitudinal data and give more weight to value-generating segments and/or the customer segments you are looking to target/appeal to on your site.
The challenge with these “attitudinal” tools is they seem to be completely divorced from the web analytics database and other customer databases. Hence, the need for a unified marketing database that joins online and offline data, so we can draw these segments and analyze them across all the currently disconnected data points we have in the online world.
Sage advice about “first click” too, and I agree. But don’t you think “last click” analysis has some advantages too? Take for example Amazon. My first click was more than 10 years ago (and tens of thousands of cookies ago), but my last click, a few weeks ago, via paid search led me to buy a book. While my first click created my loyalty and could be used to predict my likelihood to purchase in the future, it was the last click that sealed the deal. Which gets the attribution?
Maybe we should ask Akin! I am very honored to have been asked to participate in his project. He’s a very smart (and super nice) guy, and it was all my pleasure to do so.
Here’s to drilling down, and let’s make sure to swing back a few in DC! UBC materials are coming soon too.
Thanks again for commenting and reading!
Jim Novo added the following ...
Re: surveys - Yes, can’t believe we’re not firing surveys against known behavior yet. I suppose that’s “hard” to do, but in the end, the data is much more actionable and you can trend it over time with the same base. Undoubtedly, lack of focus in this area is driven by lack of focus on current customers as opposed to new customers. That will change…
Re: Last click, sure, nothing in a vacuum, right? Depends on what you’re trying to accomplish. I was generally referring to “campaign cycles” as opposed to “forever”, though if you’re interested in LifeTime Value prediction, saving the first click for a couple of years can help figure out where long-term ROI is highest. 10 years is probably a stretch :O
Judah added the following ...
Jim: Re: “hard” to do, yeah agreed. I guess that depends on the software you use or technology you create. There are some companies out there who have modules to “detect” behavior on the web based on “events” and respond in a variety of ways. In theory, “response” could be a survey (but will more likely be an “offer”)…
That’s an interesting comment about lack of focus on current customers. Perhaps the purchase cycle has something to do with that… businesses with longer purchase cycles where the majority of customer LTV is in the first purchase are probably pretty guilty of this, ay?
Re: LTV. Yeah, I was putting myself in that context with Amazon. Speaking of “hard” to do, first click attribution dependent on cookies…
Tom Strachan added the following ...
Great commentary and thanks for your blog. I agree with your comments regarding on the open systems and everything you said around attitudinal data. Thanks.