Analytics Vendors – is the feature race over?

The feature war is over?

Today I re-discovered this blog post by John Rodkin, the interim (turn-around) CEO for WebTrends, from early last year on the impact culture had on WebTrends declining performance (despite what was and today still is considered a strong product). It’s simply very interesting to see how leadership style and employee morale can make or break a product’s success. And it’s only more magnified when you’re in a feature race, as WebTrends “was” (keep reading) and Yahoo! is. I don’t want to draw too close a parallel between WebTrends and Yahoo!, but as the 4th CEO in 3 years, Yahoo’s Carol Bartz certainly recognizes that Yahoo!’s business model has placed it in multiple feature races, and that reforming the culture is an important piece of the puzzle to even stay competitive in many of these markets.

What’s also interesting is that I just learned today that despite this amazing graphic showing intense intra-industry competition, the analytics feature war is over.

In short, John Lovett, lead author of the Jupiter Research Web Analytics Buyers Guide finds that we are no longer in a fierce feature competition beyond a few low impact areas like video tracking. Rather the new differentiators in this highly competitive market are pricing and flexibility, and the analyst concludes that a big part of this “flexibility” is the ability to integrate different sets of enterprise data together, such as online and offline data. And he even goes so far as to say that one reason we need data integration is to unite all of the multiple ways a marketer touches a customer with a marketing message, and all of the revenue and costs related to each transaction, so that marketers can get a holistic picture of the TRUE value of each marketing channel, message, product, search term, etc.

This is what I refer to as Attribution – attributing the correct value of a transaction (not just a sale btw) across all of the elements of a marketing campaign.

Though he has taken a strong stance with this release, I pretty much agree with Mr. Lovett. I’ve been doing competitive evaluations in the analytics and SEM management space and I’ve had difficulty finding many KEY differentiators. Depending on your prioritized needs, there are multiple choices depending on the level of sophistication you need, the tools you need to integrate with, and how much you’re willing to pay. Most vendors can now provide you more than you need or will actually use. And Attribution is certainly an area the team at Yahoo! Web Analytics has been thinking about as the next frontier for analytics.

The first step in attribution is the ability to track individual visitors, not simply individual sessions, which is how analytics data is currently collected and reported by almost all major vendors (congrats to Coremetrics LIVE Profiles which seems to me to be the truest visitor-level tracking available, though there are still obstacles to be overcome as I will discuss).

Though Yahoo! Web Analytics (and others) can already identify that a visitor is a returning visitor and that today’s visit is their 7^th visit to the site (at least since the user’s cookies were last deleted), no vendor with the possible exception of Coremetrics is currently able to capture data for each of those visits – at the individual visitor level – to allow you to view all of the touch points with a specific visitor before or after a specific visit or action (ie Sale), in order to track back from the purchase all of the things that affected the purchase.

Furthermore, no vendor (even Coremetrics) has the visibility to KNOW all of the touchpoints that occurred on all of the places that visitor has gone on the Internet. Sure, we know clicks, but even display ad impressions running on ad networks are hard to track at the visitor level. This is a sticking point, and will probably be best solved as publisher networks get large enough to actually be able to follow visitors for a larger % of their total online experience and thus have better insight into all of the ads, articles, widgets, and perhaps even conversations they have been exposed to. And don’t forget offline advertising exposure. We will need a way to integrate offline panel information on TV ad viewing, for instance, to get an even more robust picture of each visitor’s true exposure.

Example

Take an Internet user who eventually purchases something on a website. It would be interesting to see all of the ways that user was influenced before making the purchase.

User X sees banner ad1

User X sees banner ad2

User X is an average TV watcher for his zipcode and demographic and has seen TV ad1 twice

User X sees that a Facebook friend was talking about the product

User X interacts with rich media ad1 for 45 seconds

User X searches on Yahoo, clicks a search ad, and looks around on the website

User X sees banner ad1 again

User X searches Google and arrives on the site and completes a purchase

Today: Google receives 100% of the credit for the purchase because all the website owner knows is that the referring source for the visit that led to the purchase was a Google search and of course the query that was searched. There is no past visibility into that user, prior visits, or prior advertising touch points, such as ad views or clicks.

Future: Advertiser enters some rules (perhaps suggested by the analytics vendor based on industry standards as well as that vendor’s customer base, and perhaps customizable by the marketer to reflect their own experience). There are many possible ways to divide the credit up amongst all the ads/touchpoints. Here is simply a list of options:

• First banner ad should get some credit for raising awareness. The value of that impression though should perhaps decline over time or expire after 30 or 45 days.
• Or perhaps split the credit between all banner ads since it’s a well known advertising doctrine that most people must see an ad 5-7 times before a message registers, so perhaps 1 on its own is not worth much.
• A banner viewed the same day as a search seems like it should receive special credit since it very likely influenced the user to search.
• What is a TV ad view worth relative to a banner ad view?
• What is a brand impression from a Facebook friend worth? It is surely much more impactful than a corporate banner ad.
• What would a banner ad containing a user review or the quote of a trusted third party be worth relative to a corporate banner ad?
• Interaction with any kind of ad is a sign of engagement and should receive credit. What is a video play worth relative to a rich media interaction? Does the length of the engagement you created matter?
• Some will argue that the last search should receive a great deal of the credit because you cannot deny that the search helped enable the actual sale, however others might argue that many people do a Google search and click an SEO listing as a way to navigate to a website after they are already “sold” on the idea in their head, so how one actually navigates to the site is of little importance at all.
• Or for simplicity sake, an advertiser might just average the value of the transaction out over all the impressions and all the clicks that occurred prior to the last click. For it may not be perfect, but at least it’s more accurate than giving the last click 100% of the credit.

These are all questions and issues that individual analysts and the industry itself are waiting to tackle, once true visitor-based tracking is available. At that point, we will finally come closer to understanding the true value of different formats of ads, publishers, audiences, creatives, offers, etc. These answers will have a huge impact on where the money flows in the $500B per year global advertising industry. So yes, attribution is kind of a big deal. And data integration a necessary feature? to help solve the attribution question.

\o/

Advertisement

Tags: Attribution, Coremetrics, corporate culture, google, SEM, SEO, Webtrends, Yahoo, Yahoo! Web Analytics