The top 8 things to do at eMetrics Toronto

eMetrics is this week in Toronto and I'm pretty excited!

8. Come out on Wednesday night to catch a panel with Jim Sterne and Mark Dykeman on Convergence and the State of Audience Measurement. Mark deals with radio, tv, social, and web analytics in his role at the CBC, and you'll learn a few of his secrets for dealing with media.

7. Corner Stephane Hamel at the eMetrics meet and greet that Wednesday night. Ask him about W.A.S.P., and then turn around and ask me how I use the product!

6. On Thursday, during the first break, visit Jim Novo at the WAA desk, and ask him three questions about analytics, certification, and professionalization.

5. Attend Stephane Hamel's Web Analytics Maturity Model session.

4. Attend the "How decision makers consume what analysts produce". Of all the Thursday sessions, this is the most critical. You need to hear what other decision makers think and need, as these are your primary consumers.

3. Attend a Web Analytics Wednesday on a Thursday, sponsored by Syncapse. It's the most heavy-weight group we've ever had in Toronto - Novo, Sterne, and Peterson - together on stage. I really want your experience what has been my experience with all three of them over the years. And we'll be doing it in WAWTO tradition.

2. Hemash Bhatti, a measurement scientist from Syncapse, contributed to the WAA Code of Ethics, and was there as dozens of issues arose. He knows a lot about what happened, the implications, and EU privacy considerations. He's on the front lines of this. The panel on privacy is extremely important.

1. Attend the Social Media Metrics session. Simon Austin and myself are presenting on Social Media Metrics. Simon Austin is one of the best minds in SMM, and his presentation is easily the $2000 session. I will be dovetailing with him, presenting findings on SMM from the WAA Research Committee.

There are a few other tips for getting the most value.

• Friday morning's sessions are excellent - if you have no mobile measurement, attend Patrick Glinski's session, otherwise, do mobile.

• The lobby is where I met Ned Kumar last year. I've learned a lot from Ned, and I continue to get deeper into the analytics of innovation as a result. I recommend spending some time in the lobby over the course of the conference.

• Shake hands and mix. eMetrics attracts 60% introverts who continue being introverted. You have to shake hands if you're to get to know others. Step out of the comfort zone, well outside the comfort zone, and really go to town.

I'd like to see you there.

Seven Answers.

I learned a lot from the discussions with many of you. Here are my proposed answers.

1. What is a valid model for monetizing a web page, regardless of whether monetary transactions are possible?

Yes. There are two conservative paradigms and multiple progressive ones. The most conservative one is to assign a monetary value to the conversion event, and then attribute the monetary value back through the tree using a decay curve. This method would automatically consider abandonment rates and discount those pages accordingly. The other conservative way is to assign a monetary value to each page based on paid media received or valued.

A progressive way is to examine how much content upon each page is copied and shared. A less progressive way is to examine call center deflection, paper deflection, and other cost deflection and assign that cost savings onto FAQ and transactional zones of the site.

A more progressive model, still, would be to incorporate all 4 models into a single unified one. At the risk of overcomplicating the model.

2. What is a valid model for costing a web page?

There are several valid models. The first, most simple, would be to take the entire digital budget and divide it by the number of pages.

The second, more complex, would be to distinguish the difference between working and non-working dollars, assigning the non-working dollars evenly across the entire site, and then assigning the working dollars (special copy, advanced creative, multi-media assets, so on), to only those pages containing the assets.

The first method is the most pragmatic and adequate for perhaps 80% of the websites out there.

There exist other models. Those are the two most communicable ones.

3. What is the relationship between the complexity of a page and the monetary value of a page?

Possible operational definition of complexity: word count, image area, ad count, video count, div count, link count.

Accepting the most progressive, complete monetization model.

Hypothesis: The relationship between the complexity of a page and the monetary value of a page is statistically insignificant.

4. What is the relationship between (hierarchical navigation) buried depth and the monetary value of a page?

Accepting the most progressive, complete monetization model.

Hypothesis: The relationship between the depth of a page and the monetary value of a page is significant, except in the instance of a hierarchical monetary valuation (where you couldn't answer this question with a straight face!).

This is because the conversion page would always have the highest monetary value. The immediate pages subsequent to it would have higher monetary values still.


5. What is the relationship between indegree connectivity and the monetary value of a page?

Accepting the most progressive, complete monetization model.

Hypothesis: Pages which have a very large number of inbound links are likely to be worth more to a firm than those with few.

Inevitable question: what about search?

Exactly. Search is the final frontier of good design.

6. What is the relationship between an audience segmentation and the monetary value of a page?

Accepting the most progressive, complete monetization model.

Hypothesis: Areas of the website dedicated to customer retention, service, and loyalty are more likely to have higher monetary value than those purely from an acquisition focus.

7. What is the relationship between customer affinity and the monetary value of a page?

Accepting the most progressive, complete monetization model.

Hypothesis: Customer segments with high customer affinity are to be associated with pages that have higher monetary values.

----

This discussion wasn't so much about deriving an universal model, applicable to all websites. It was simply asking for valid models, and under which circumstances would some be worth more than others.

I learned a lot over the past few days listening about it. Many web analysts are frustrated because they understand their firms business model better than most. Understanding how the business model of a firm ties in with the website, and then fully operationalizing it, it would appear, is a pretty tall order.

I've only asked, here, that analysts consider segmenting webpages the same way they segment customers, (using money) and then remap those two sets together.

What do you think of the method?

Seven Questions

Seven questions.

1. What is a valid model for monetizing a web page, regardless of whether monetary transactions are possible?
2. What is a valid model for costing a web page?
3. What is the relationship between the complexity of a page and the monetary value of a page?
4. What is the relationship between (hierarchical navigation) buried depth and the monetary value of a page?
5. What is the relationship between indegree connectivity and the monetary value of a page?
6. What is the relationship between an audience segmentation and the monetary value of a page?
7. What is the relationship between customer affinity and the monetary value of a page?

This is quite dangerous, isn't? But how fun this will be!

If your answer endogenizes of 6 and 7 into 1, please propose how you'd do this.

Please understand that I'm asking these questions with models in mind.

Your move.

Comment below, or, tweet me @cjpberry

.

Assumptions, Facts, and Models

Without assumptions, analysts wouldn't be able to say very much about the world. Even facts have assumptions.

Consider the following statement:

"There were 19,000 pageviews in October, 2010."

Okay. I'm willing to accept that fact as true. Assuming that all the tags were in place, on every page. Assuming that server errors identified and not counted as pageviews. Assuming that all the tags fired without fault. Assuming the web analytics software is calibrated to the WAA definition of the term 'pageview'.

People form associations and communities, in part, to standardize assumptions so that progress can be made. Whoever is behind the Metric System. The IEEE. The WAA.

Consider the following statement:

"Traffic to the website causes conversions."

Oh boy.

So let's unpack.

What needs to be true for that statement to be true?

Assume that traffic means visits.
Assume that 'cause' means 'cause', that is to say, Conversion occurs as a consequence of Traffic.
Assume that 'cause' does not mean 'only cause'.
Assume that conversion means 'a monetary exchange that is measurable on the site'.

Under those four assumptions, I'm willing to accept that statement as truthful.

But there are more specific statements.

"Qualified traffic to the website causes conversions."
"Qualified traffic to a trustable, W3C compliant, with content that inspires desire to purchase a product, and a strong checkout funnel, causes conversions."

And less specific ones.

"Marketing causes conversions."

All of which can be unpacked to expose a very specific model of how marketing works. And by model, I mean a range of variables which interact in way where causality is causality and correlation may be observed statistically.

We're willing to accept a high degree of abstraction because that's the price of brevity. And different communities pack an awful lot of meaning, even models, into single terms. Moreover, entire communities identify themselves along lines of assumptions that may not be completely acceptable to other communities.

Brand managers have their own universe of assumptions, facts, and models. As do data miners, public relations folks, CRMers, and marketing scientists. To spend at least a day in each of their worlds, making the effort to understand, is valuable.

We'd all be better off if, instead of gagging at the sight of something and rejecting it, to probe into those assumptions. Check to see if the logic holds, and then, discuss. There may be entire regions of knowledge that could be incredibly enriching and make us better.