The Alpha Project: End-to-End Measurement

You may not know what a “relational database” is, but chances are good that someone else in your company does. If you keep lists, organize information or make data available over the Internet, that data is probably being fed into—and then later extracted from– a relational database. If you’re small enough, chances are you do it with desktop database software—something that doesn’t require a programming Ph.D. in order to build useful, customized applications.

As a software category and a sales market, it’s one of the broadest around. Direct marketers use it; so do small to mid-sized manufacturers, insurance agents, dental offices and almost any other enterprises.

And that makes it a tough competitive arena in which to do effective, cost-efficient search engine marketing. Since the possible uses of relational database software are so varied, how does one build a workable list of keywords and phrases and manage that list to provide the best cost per action (CPA)? How do you select your best prospects for online sales and then hook into them with offers?

That was the dilemma faced by Alpha Software, a Burlington MA-based company that sells Alpha Five, a flexible, easy-to-use and low-cost alternative to brand-name database products such as FileMaker and MS Access. Alpha sells primarily to business users and relies largely on online sales, with some sales through catalogs and value-added resellers (VARs). The company offers customers a free trial download over the Internet with a 30-day try-before-you-buy guarantee.

“It’s a very competitive space, because a lot of the keywords are fairly generic,” says Alpha co-chairman Richard Rabins. Particularly as his company’s product has evolved (Alpha Five is now on version seven, if you can keep that straight), the company has added features that make it a strong choice for a lot of different Web applications. “Finding the right search terms and at the right price is not a trivial problem. Keywords for a pillow that lets you sleep better—that’s easy. But when your product can be used by a medical practice for tracking patient information on the Web and by a site selling a large database of motocross photos over the Web and by a building company—that’s a very horizontal market.” And that makes choosing the best keywords less intuitive and more complex.

Enter Web analytics to the rescue, in the form of a B-to-B search marketing portfolio from SmartSearch. Alpha retained SmartSearch to optimize and manage its search marketing efforts in 2004.

Rabins says the company wanted help not only with building an effective keyword list but with controlling costs—specifically, the cost of the free downloads—by making sure the search ads were targeting the customers most likely to take the Alpha software for a test drive. Reducing the number of ad clickthroughs that didn’t end on a trial download would effectively cut the cost of those downloads.

SmartSearch CEO Patricia Hursh says her firm started by taking a comprehensive look at the search engine marketing Alpha was already doing. “We paused things where they weren’t getting a return, and where they were, we expanded the campaign with more categories and more keywords.” Alpha was also using a narrow range of ad copy; Hursh and company systematically tested new copy targeting specific markets and new landing pages to run behind those pay-per-click ads.

The landing page issue was important because while the aim for all these customers was the same—getting them to download the trial Alpha five software—their entry points were very different. “If someone was looking for database software for human resource applications, we wanted to take them to a page that would be relevant to HR apps and then make sure that they call to action was present on that page,” Hursh says. “It was really a matter of tightening the focus between the user’s search and the answer Alpha was providing.”

Working on its own, Alpha had been measuring visitors to its site and costs per visitor; but they were not measuring conversions and cost per conversion until SmartSearch gave them the analytic tools to do so.

Within six months of working with SmartSearch, Rabins says, Alpha had managed to triple the number of trial downloads and to cut the cost per download by 53%. He says his company was recently pitched by another SEM firm that offered to improve his results; after looking at the stats from the job done by SmartSearch, they finally admitted that Alpha was doing as well as could possibly be expected in both conversions and costs.

The Alpha project has worked well for SmartSearch, too: In November, the firm announced that it had formalized the practices it honed with Alpha and a few other B-to-B customers into a dedicated offering, SmartSearch B2B.

For Rabins, the next job is to go beyond cost per download and start measuring cost per actual sale. That’s tricky because a trial download can occur over one computer while the sale transaction after the trial may happen over another, or over the phone. Alpha and SmartSearch are working on a system that would track those sales from clickthrough to download and then to actual purchase using globally unique identifiers (GUIDs). It won’t be perfect, since it still won’t measure offline sales by phone or via VARS. But it will be good enough to give Alpha a clear picture of which terms and categories are producing its highest number of true conversions—that is, actual software sales.

“At the end of the day, the closer you can get to measuring the ultimate object and tying that back to your keywords and bidding, the more your system makes sense,” Rabins says. “That’s where things truly get exciting. Knowing the cost of your sales gives you the confidence to step on the accelerator and make the most of search engine marketing.”