You are also correct that any company that want to maximize conversion - macro or micro - can do it now. But I think some companies have it much easier. Newer, smaller companies, for example. Larger, older companies can have extreme challenges to measuring conversion, whether technical, resource-driven, organizational alignment, management who understands/support conversion measurement, and so on. Thus, as you correctly, put it requires “a structured plan and educated marketer,” and that plan may/will touch many facets of the business (from product development to IT to sales).

The data models in web analytics are all over the map, and most of the major vendors have more than one (some intentionally, like CoreMetrics, and others as result of acquisition, Omniture, or new product innovation, WebTrends). I agree we are seeing vendors move in a “unified” direction, but not fast enough for me, so I will continue to talk about it.

I should also add that several of these constructs are also available for targeting ads too, in addition to content, depending on the ad server/network.

Thanks for commenting, and keep up the awesome work and blogging at SiteBrand!

Judah

I also spend a lot of time thinking about targeting, as I build and refine the business support programs for Sitebrand, a web personalization company. I wanted to drop a few points into the dialog around segment examples, and the usefulness of analytics.

Segment Examples: All the examples you used above are spot on for in-site targeting (rather than advertising targeting), but are much more powerful when used together as part of a solid messaging plan. For example, if you use your Language example and choose to personalize messaging based on french language speakers, look at goal outcomes, behavioral metrics, geographic data etc. This will allow for a more complete segment understanding, and more powerful targeted messaging.

Targeting and Analytics: Again, you are correct in stating that analytics vendors and targeting vendors have a ways to go when it comes to working off a single dataset, although the industry is definitely moving quickly in that direction. That said, even with shared data, targeting and personalization can’t be successful without a structured plan and an educated marketer. Any innovative company that wants to maximize their conversion rate can do so now. Find traffic segments and baselines in your historical data, and start testing changes in your site content and layout.

The more discussions like this take place, the more the discipline behind conversion optimization moves forward, the faster vendors will unify their data.

Cheers,

Jim

Helen: Thanks for the excellent feedback! Your example makes me think BI would do this. A few WA tools could expose this data if configured to do so, but only at the site level, which would be good for understanding audience in the context of all those dimensions on the site itself, but not across sites. Perhaps Quantcast is the closest to being able to measure segment outcomes/demo/behavior/reach/frequency across sites. You have identified the issue with conversion in a nutshell. Parts of the conversion equation are outside of the site’s control and subject to the agencies creative input, which of course should be provided in a framework that allow for testing to ensure the prediction has a better chance of fulfilling itself during execution. Thanks for commenting!

Helen and Toby: Apologies it’s taken so long to respond, I’ve been on vacay!

Sniffing is an interesting technique, but it can be difficult to deploy if you can’t “sit” in the request stream in front of the webserver - like if you are hosted somewhere for example.

Considering the very specific elements of this segment, Bostonian-Manga-Loving-Guys, I wonder how this data could be distributed and who would manipulate it to create the segment…Maybe I should have chosen a broader target =)

Regardless, I can see how other elements beyond the above 4 segmentation attributes can affect outcomes and produce dubious stats. Mainly, elements that the advertiser controls—messaging, calls-to-action, landing pages, design, etc. So a vendor may tell me that this segment can generate $xx ROI, but is that really the case in the context of execution?

Further, if such a pool existed, it effectively makes the publisher, not the client or its agency, responsible for ensuring that the creative running in all ad spots on its site(s) is of sufficient quality to engender direct response at a rate which would not reflect badly on the site(s) public performance data in the long term.

Finally, this type of performance data still somewhat fails to take into account the effect of ads where brand or product exposure/awareness/perception etc rather than direct response is the campaign objective.

I obviously agree with Judah in regard to getting at the raw data from a web analytic’s perspective. Omniture is the least flexible compared to the other vendors in the marketplace.