Dirty Data Equal Bad Bids

Garbage in, garbage out. That saying, from the early days of data processing, illustrates the fact that even the most brilliant computer program can’t deliver when poor data are fed into the system. Well, the problem hasn’t gone away; in fact it’s a serious problem today because many search marketing campaigns are based on garbage data.

A huge segment of marketers base their entire bid strategy on bad data. Sometimes, reasonably good data is incorrectly interpreted in respect to bid intelligence. Marketers and agencies seem to focus heavily on the bidding process when it comes to managing paid placement search campaigns. Often, data used to make these bidding decisions is assumed to be the right kind when in fact it’s only surface data. It doesn’t accurately reflect business and marketing objectives.

Situations where the wrong or inadequate data are collected include:

  1. Binary data is often the only thing collected when numerical data would be far more valuable. This is the case where the fact an order or lead has been generated is recorded in a vacuum with no additional information. Sure, you may be managing around a cost-per-lead/acquisition (CPA) or cost-per-order (CPO), but chances are your business is driven by more than averages.

  2. Focusing only on a single conversion event without recognizing the prospect, consumer or buyer goes through a more complex buying cycle.

  3. Missing opportunities for data richness or data pass-back. Better campaign management and Web analytics systems/solutions can benefit from additional data relating to marketing objectives. For example, passing back Zip Code, customer score, new/returning customer, or gender data can be important, particularly now that MSN allows for bid boost/targeting by demographics, and Google’s Site Targeting network allows site selection based on visitors’ demographic profiles.

  4. Failure to test search-specific buy funnel behavior. If your site is like most others, most search visitors, even those arriving on product or brand keywords, don’t immediately buy during that session. Someone typed in a search term that flagged him as interested in what you have to offer, but didn’t visibly consummate a sale or otherwise engage in one of your preferred positive (success) behaviors. If your site had a positive impact on that visitor they may return via:
    1. Organic search on a new keyword or the same keyword
    2. Paid search on a new keyword or the same keyword
    3. Direct navigation or bookmark
    4. Other media

  5. Neglecting the organic placement/paid placement interaction effects. Savvy marketers know how to maximize profit when they have top organic rank. Usually, this involves using paid search to provide alternative messaging to searchers while increasing screen real estate. Any cannibalization or interaction effects between the paid and organic listings can be tested due to the control you have over the paid listings.

  6. Neglecting to include offline conversion. Study after study shows that most retail conversions to a sale occur offline, after search engines were used to make a product or vendor selection.

Another problem marketers face is when the data collection methodology is somehow flawed, or the way data are managed creates issues that must be taken into account when running campaigns. Flaws in data collection methodology arise from a combination of human error, site-side technology issues, and third-party analytics or campaign management technology bugs. Common issues resulting in flawed, inaccurate data include:

  1. Missing tracking codes/bugs/pixels. Ensure you have a QC process that verifies pixels are deployed on all the right pages and pass the right information into third-party technology applications.
  2. Setting your objectives within campaign management technology without taking into account the ongoing problem of cookie deletion due to spyware and manual paranoia.
  3. Failing to tag keywords individually within the engines, resulting in a loss of understanding of the campaign’s linguistic drivers.
  4. Failing to analyze referrers on inbound search traffic, resulting in an inability to separate the phrases that arise from a broad match listing.
  5. Not tagging contextual search traffic as separately identifiable, resulting in an inability to know the relative ROI of search and contextual parts of a search engine’s network.
  6. Not capturing time/date information in a click. Without time and date information, dayparting cannot be successfully implemented.

Any data relevancy or data quality issues covered in this column can completely derail a campaign, regardless of how much energy is put into bid management. Bid and campaign management systems rely on the data they collect (or are fed).

Garbage in, garbage out. Take the time to ensure your campaign is empowered by accurate, relevant data, not crippled by wrong or inaccurate information.

Related reading