Do you find that website traffic numbers don't always match up? Our media strategies editor explains how to handle these discrepancies.

How many visitors come to a given website? Is traffic to "Site X" on the way up or down? Do the sites an agency recommends for a media buy have high reach or low reach against the target audience?

Answering these questions is standard operating procedure for online media planning and buying. Having the analytics sites and the sites themselves be at odds over these answers is just as standard.

Joshua Schachter, founder of Delicious, has said that comScore's numbers for the site are incorrect. Betsy Schiffman, a writer for Wired, commenting on the surge in Google's per-share price on news that their paid clicks rose 20 percent in April as reported by comScore data, writes: "What sort of significance does comScore data really have? Not much, as far as we can tell."

So, why is it that the numbers we all use as the basis for making decisions -- sometimes worth millions of dollars -- are at the same time so universally vilified as being meaningless and inaccurate?

Like a Zen koan "It is what it is. It is what is isn't."
The inaccuracy of panel data, survey data and even server data is an inherent fact of this business and of all businesses.

The old saying "nature abhors a vacuum" applies when it comes to business intelligence of any kind. In the absence of "true data," business will accept anything that might resemble true data.

Agencies need to have numbers so that they can justify their recommendations to clients... and often those numbers do correlate to truths about a media plan and its potential impact on the marketplace. 

Clients want numbers because they feel that the true decision-makers within their organizations will want more than just the marketing guy's interpretive knowledge or experience to justify spending money. 

To quote an old Buddhist saying, "A good horse runs even at the shadow of the whip."

Quantification is a kind of "spirit from beyond" that converts business instinct, experience, practical knowledge and the "gut" into some objective truth, a reality upon which business leaders and decision- makers feel they can depend.

Many, many companies are still reliant upon old-school marketing methodology, and that means quantification of a communication delivery goal for the sake of predictively modeling one's business prospects.  

Why? Because they more or less tend to work. Are there better, more precise ways of going about the task of influencing one's business and predicting the outcome of that influence? Sure. More precise research, larger panels, a greater collection of behavioral segments and their click streams cross-tabbed with survey data cross-tabbed with purchase behavior cross-tabbed with media usage can yield a picture that is more exact in its representation of reality. But so long as the out-of-pocket costs of the picture that is pretty close to reality are greater than a representation that is better than pretty close to reality, not much will change. 

And don't forget that the intangible costs of getting "better than pretty good" methods approved and executed (political will, changes in points of view, the restructuring of mental infrastructure, training with new technologies, etc.) remain high. As a result, "pretty good" is what we will continue to use.

Agree to disagree?
When the numbers don't match, either between two independent sources of data or between a 3rd party resource and the site being represented, what do you do?

Look at it from all angles. I like to take the eMarketer approach, which is to look at all the different facts and the sources for those facts, and assemble a picture of reality from them. This may not be the final TRUTH, but you can feel confident that the representation ultimately given to the client is probably better than if you relied on a single source.

But don't get too caught up in the numbers. As much as agencies continue to reposition themselves as analytics companies or research service providers, in addition to offering the old tried and true media planning and buying services, agencies are still primarily perception and relationship businesses. Sometimes it isn't the data you have, but how believable it is to the party to whom it is presented. Copernicus was right about the Earth moving around the Sun, but it simply didn't seem plausible.

Decisions to do business with one site or another are never, or should never, be determined entirely by a discrepancy in the count of a site's visitors. If a planner is making the decision to include a site on a buy solely based on its size, the planner isn't doing very good media planning. An agency or brand doesn't need to spend the money to hire college graduates to determine relative sizes of numbers. Even my cat knows the difference between a small bowl of food and a large one, and he'll work for a small bag of Greenies.

Media Strategies Editor Jim Meskauskas is vice president and director of online media for ICON International, Inc., an Omnicom Company.