Don't settle for any old ROI

Young-Bean Song
Web Analytics Posted on February 10, 2009
Something kinda stinks, doesn't it?
You may now be asking yourself, why does only one ad get credit? Aren't consumers exposed to many ads, across multiple sites, over extended periods of time? Why are consumers clicking on my ads in the first place? Let's compare what your intuition is telling you right now with some in-market research from hundreds of online display and search campaigns (all of the research quoted below can be found at the Atlas Institute):

  • Intuition: "There's probably some overlap and combined effect between my search and display campaigns."

    Research:  Ad impressions to users who clicked on sponsored search ads increased their conversion rates by an average of 22 percent. 

  • Intuition: "A lot of the folks who click on my sponsored search ads are simply trying to find my site."

    Research: Of all sponsored search clicks, 72 percent are navigational in nature. Navigational clicks are branded terms (exact-match brand names and URLs), or repeat clicks on search ads that had been previously clicked in prior sessions.

  • Intuition: "My customers are seeing many of my ads, and all of them probably have some incremental amount of impact."

    Research:  The last click or view typically only accounts for 6 percent of the touchpoints (impressions, clicks, and interactions) between a given advertiser and converters. Said another way, the average converter has a frequency of over 15 to 20 touchpoints.

  • Intuition: "Not only am I reaching the same people multiple times, but I'm doing so across many different sites."

    Research:  Nine out of 10 converters were exposed to ads from the same advertiser across two or more sites.

So if there's all this great research, why aren't more marketers doing something about it?  There's always a trade-off between simplicity and accuracy. Clicks are an easy target for the digital intelligentsia, fraught with fraud, repeat clicks, demographic bias, and unclear motives (i.e., the classic woman-in-a-bikini-to-get-a-high-CTR experiment). Yet clicks remain a mainstay of our reporting, and are the basis of Google's $22 billion business. 

We shouldn't be surprised. Clicks clearly demonstrate that an ad had at least some utility. They are also easy to count, explain, and understand. And although everyone acknowledges we can and should raise the bar of our metrics and standards, all of the branding studies, data warehouses, expert consultancies, and media mix models have yet to deliver a ubiquitous and fulfilling notion of ROI. Being a research geek myself, it pains me to admit that many of these solutions are simply too complicated, too expensive, and too difficult to explain, let alone scale.

<< Previous page | Next page >>

More articles from Web Analytics

 

Comments

Gretchen Hyman
Gretchen Hyman February 10, 2009 at 12:38 PM

We apologize to The Atlas Institute for any misunderstanding over the headline of the article. It has been changed to more properly reflect the contents of the piece and intention of the author.

Young-Bean Song
Young-Bean Song February 10, 2009 at 11:16 AM

For the record, I need to point out the title of this article was not the original, and was placed without my review or consent. The original title was "Save the World with Better ROI”. The rest of the article is sound. I want to stress that my criticism regarding the Last Ad model is not directed in toward Google. Last Ad reporting is an industry wide standard, and impacts every digital channel. I also would like to point out the Microsoft owns a search and display network that is evaluated and paid based on the Last Ad model, so the company that I work for is not exempt from the same issues. All that said, there is much progress being made. One of the biggest challenges is fear of replacing one flawed model with another. I would argue that the bar is so low (said another way, the current Last Ad standard is so extreme), and the marketing touchpoints are so many, that almost any reasonable attempt to associate some credit beyond the last ad has some merit. And if anything, the exercise will expose the richness of our marketing that currently is obscured by the bluntness of our reporting. A lot more information and research can be found regarding these new methods, and how variables and weights are selected at our website: www.atlassolutions.com/emap

Daniel DeYoung
Daniel DeYoung February 10, 2009 at 10:43 AM

Young-Bean,
Excellent article!

I have been preaching to my clients for years to look at all the exposures (touch points) that lead to the conversion.

I have written a small tracking script that ties to set a cookie on the first visitor and I call this lead source the original source. Then on each future visit I append each new lead source to the list always keeping track all of the date and type of media the prospect responded to.

When the prospect is converted to a customer the cookie is written to the database and you can go back and see just what level of activity it took to convert an that prospect.

I know some clients who use a hybrid program of recording the First Source & Last Source for each prospect, while this is an improvement over the Last Source only method is still leaves a missing picture of the activity in between the bookends.

The only real issue I have with "Multi Sourcing” a lead is how do you do the weighting?

You still need to divide that budget up next month so…

How do you apply your "credit for the conversion” for the conversion that take place across multiple sources (touch points)?

Do you give equal weight to all sources?

Do you give more weight on the First & Last source?

Once again great posting,

Daniel A. DeYoung

A Direct Response Mercenary

http://www.linkedin.com/in/directresponsemarketing

Stephan Pretorius
Stephan Pretorius February 10, 2009 at 8:14 AM

This is a good general discussion of why we need to move beyond the last-click-wins model and I agree with that imperative. But I have 2 criticisms:
1) Last-click-wins is not "Google's ROI equation”, it is our industry's ROI equation. Does it benefit search in favor of display? Sure, but to imply that it is Google's model is disingenuous and implies that they have imposed it on our industry, something which is blatantly false.
2) It is not good enough to gloss over the methodology for improved conversion attribution and to rely on a system to do the math for you. We cannot go from one flawed model to another and not take the time properly to understand the underlying methodology. There are massive unanswered questions in this area (how to weight exposure to account for time-lag to conversion, how to do attribution on small conversion data sets etc) and we need to do a lot more work before we can rely on a new model for our media buying decision making. Without this rigor the whole exercise will simply look like a thinly-veiled attempt to divert money away from search to display (or should I say, from Google to Microsoft?).