- Ensure that campaigns are being optimized to the most granular data possible. In search, keywords provide highly granular insights. In display, similarly granular insights and optimization signals can be attained by targeting using unstructured data techniques such as keyword level search retargeting, keyword contextual targeting, and page level site retargeting. Using unstructured data will help a small business extract the most optimization value from its limited data set.
- Work with a local focused network or newspaper publisher who can apply learnings across different, non-competing geographies. Such companies provide online advertising services to non-competing businesses across multiple geographies. To the extent that these companies have the rights and ability to apply lessons learned on specific verticals (such as plumbing) across geographies, they can pool data and gain insights that benefit all in each vertical.
- Leverage look-alike targeting. In search marketing, advertisers can see exactly what keywords are working and add similar terms to expand reach. In display using unstructured data, advertisers can extend reach by analyzing the precise search patterns and site visitation patterns of converters, and target others with similar patterns. Especially for companies with few signals, it is paramount to execute look-alikes off of more accurate and more granular unstructured data instead of broad-based, opaque audience segments.
The promise of Big Data for display advertisers is compelling. The more insights you have, whether they are from your own first-party data or from third-party data, the more likely you’ll be able to optimize to meet the performance metrics of a campaign. And when it comes to driving both performance and scale in a real-time bidding environment: the bigger the data, the better. But like a shipwreck survivor suffering from dehydration while surrounded by ocean water, some advertisers are surrounded by data, just not the type of data that they need. And just like drinking the wrong type of water, using the wrong type of data can have adverse effects. But, what about advertisers who feel that they don’t have enough of the right type of data to work with? This challenge of scarce data is a reality for many businesses. Some companies, especially small businesses, may not generate much data in a campaign while others, like luxury retailers, may have a wealth of data, but only a small amount that actually provides a signal related to driving conversions on key products. The result: Companies with scarce data can have difficulties in getting their campaigns optimized to their CTR, CPC, or CPA/L goals. Either their budgets aren’t large enough to get their campaigns to their desired performance, or they end up casting a wider net, incurring waste in their budget because the quality of their look-alike models may be compromised from overly-broad data or from mixing in conversion signals from diverse products. Neither is ideal. Smart Data, Smart Decisions The good news is that it is possible to optimize to small data sets. Programmatic optimization strategies aren’t just for advertisers with huge budgets and lots of conversions. The key is finding ways to accurately amplify scarce data signals, enabling companies to optimize the valuable data that they do have. By expanding on the data signals in their audience profiles, even companies with scarce data can achieve strong results. Consider the following examples of two very different businesses types with data scarcity. Optimization Solutions for the Small Business For our first example, consider a small local company such as a plumbing business. The company operates and advertises locally, often through a local-focused ad network or newspaper publisher. When it comes to optimizing online campaigns, however, this business faces a few problems. Given a small budget and limited site traffic, it takes a long time to gather enough data to understand which signals are statistically significant. In fact, if the company does make assumptions based on a small data sample that is too small, they are in danger of drawing erroneous conclusions. Which brings us some solutions for small businesses that are struggling with scarce data: