Live from DMA06: The ABCs of Search Data Diving

Oct 18, 2006 6:55 PM  By

(Direct) San Francisco–Six years ago, you could do search engine marketing badly and still make out well, because so many other people were handling it even more ineptly.

But today the level of the game has been raised all around, and search marketers need to take a deep dive into their performance reports on a regular basis to make sure that they’re spending enough on profitable pay-per-click campaigns while culling the ones that don’t produce conversions.

Attendees here got a crash course in what to look for from their search engine marketing data from George Michie, vice president of client services for the Rimm-Kaufman Group, a search engine marketing firm.

The first step is gathering the performance data you’ll need, said Michie, including search terms, ad copy, landing pages, and URLs. Use keyword-level performance data (costs and sales) aggregated over a two- to three-month period; an order audit that ties order numbers and amounts to keywords and search engines; and reports of your actual costs for keyword campaigns on each search engine.

If any of this data is difficult to come by, that’s an early warning that an SEM campaign is not properly run. “If you can’t tie costs, sales, and clicks together, how are you running your bid-management campaign?” Michie asked.

Next, check how many search terms you have. Rimm-Kaufman Group recommends 5-10 keywords for every SKU you sell. Fewer than that risks lost sales and less efficiency.

Be sure to separate your branded search terms — trademarks, brand names, and domain names — from searches on more-competitive terms. Many users employ the search query box as a navigation tool, surfing to a site by entering the brand or the trademark phrase. “Those are folks who are walking through your front door,” Michie said. “You’re not generating any incremental sales there.” They will find you just as surely if you shift search budget from your branded terms to other phrases that might attract incremental shoppers.

With branded terms segmented out, take a look at your phrases in descending order of costs to see if the terms you’re spending the most on are producing the best click performance, or if terms you’re spending less on are performing well and deserve the budget attention that will push your ad higher on the results page.

Michie also recommended finding missed opportunities by looking at keywords in descending order of sales. Low-traffic terms with multiple orders probably deserve more love, too, to raise the ad rank.

Finally, take a look at your data sliced in a variety of ways: by traffic volume, by product category, by landing page, by product manufacturer, and so on. “If you spend $2,000 to generate $25,000 [in sales] on a low-traffic term, you’re being far more efficient than you need to and may be leaving opportunities on the table,” Michie said.

One substantial problem in trying to run this kind of analysis on search marketing campaigns, he added, is the question of properly attributing credit for sales. Sales that occur in a store, on the phone, or through an e-mail coupon very often began with a search, Michie said. To properly tune your search efforts, make a point to use dedicated 800-numbers in search ads or coded coupons in e-mail.

And consider instituting policies that measure the amount of channel spill from search, such as calling 100 first-time phone customers to ask how they found you or having cashiers ask the question at the point of sale.