At the heart of every paid search management platform is an algorithm—a set of mathematical equations that uses a variety of inputs to compute the optimal cost per click (CPC) to achieve your goals for revenue and return on ad spend (ROAS). Most search platform vendors take a one-size-fits-all approach to developing their algorithms. This would work fine if all types of online sellers, all products, and all customers were similar. Obviously, this is not the case.
Adlucent realized long ago that retail is different from other verticals. The kinds of products e-tailers sell are very different than, say, what travel agents, financial firms, or auto manufacturers sell online. The differences in the products they sell lead to major differences in how buyers of these products shop, which in turn dictates how their algorithms should be developed. For instance, customers spend far less time shopping online for a DVD than they do for a car. These faster purchase cycles mean that most retailers’ algorithms should use a shorter look-back period.
An effective algorithm for retailers must do four things:
- First, it must take into account all the retail-specific factors that effect paid search campaigns. Over the last several blog posts, I’ve discussed the importance of pausing campaigns that advertise out-of-stock products, establishing the appropriate look-back period, adjusting campaigns for seasonality, and factoring in customers’ acquisition cost and lifetime value. Adlucent integrates all of these factors—and many more—into its algorithms.
- Second, it must be specific to the unique needs of every retailer—reflecting their product categories and sub-categories, their specific customers’ buying behavior and, most importantly, their strategic business objectives. With Adlucent, every retailer client gets its own custom algorithm, and in most cases multiple different algorithms. If buyers of a particular category or sub-category behave in a specific way, a different algorithm is needed because we’ll need to react to their shopping behavior differently.
- Third, the algorithm must be continuously refined to reflect newly-learned information. Adlucent doesn’t just set up an algorithm and let it run indefinitely. We continuously update it as we collect more data, every day. Our goal is continuous revenue growth. If we didn’t evolve your algorithm, your growth would plateau (in fact, that’s precisely the situation most of our new clients come to us with). And while we love automation, we’ve found there’s no substitute for having an experienced retail expert look at the data sets we collect and apply a human intelligence filter to it.
- Finally, the algorithm must be actively managed to coordinate with your changing offline and online marketing efforts. Conversion rates (CVR) change when retailers run promotions or offer discounts. Most search platforms aren’t “told” about these short-term events. As a result, their algorithms observe sudden spikes in CVR and falsely assume they will continue forever. And when the promotions end and CVRs fall, these platforms will over-penalize the involved keywords for dropping in effectiveness. This is another example of why over-reliance on campaign automation can be a bad thing.
We’ve said it before, and it bears repeating: retail is different, and every retailer is unique. We believe that our clients deserve a custom algorithm that is optimized for their goals, backed by a team of talented Account Managers who intelligently tweak it to continuously drive your growth. Do you deserve anything less?