Information services have been steadily raising the bar on the functionality of software tools bundled with their data, making it easier for their customers to get value from their products. As these tools get more and more robust—and as customer expectations for software performance get higher and higher— these tools also shine a bright light on any gaps and anomalies within data sets. Without urgent efforts to fill these gaps in, these data “dry holes” can torpedo even the best information product.
How does this happen? Well, it’s a corollary of the “garbage in, garbage out” rule. Extending the functionality of a data-based service means that any new data—even if it’s a simple ratio of existing datapoints—needs to be as close to 100% accurate as possible. So when adding rankings, ratings, maps, or data visualizations based on ratios of underlying datapoints to a service, that data must be 100% populated and, in an ideal world, accurate. This means that null values for either the numerator or denominator in a ratio will result in a useless ratio, and one that is likely to float bad records to the top of queries run by customers. This can negate the positive value that an investment in improving software would usually bring, so much so that it could affect the user’s perceived value for the service and, eventually, renewal rates.
On the positive side, this vulnerability is easy to fix by filling in those blanks. Simply identify the missing data points and go gather that data. Of course, there’s usually a good reason for data being absent, so any effort to fill in the blanks demands an approach beyond a simple data acquisition effort.
Here are some of the related considerations that firms typically take to address “dry holes” in their databases:
- Ensure data entry interfaces mandate that required fields must be populated before submission.
- Include field-level validation in those interfaces to flag values that are anomalous (too high, too low, unexpected value types).
- Resolve the thorniest anomalous values with robust exception handling:
-
- Enter estimates based on comparables (i.e., a divergence of more than X% from a ratio for a similar firm triggers an escalation).
- Add placeholder values, assigned by a subject matter expert, to reflect your firm’s best guess at an estimated value (like the annual revenues of a closely-held private company or the physical location of a company that uses a PO box address).
- Structure the underlying database to include metadata on the provenance of data at the field level for required fields. With this approach, the end user can drill down to examine the source and age of any single critical data point, allowing for better interpretation of the data.
Other app-related challenges crop up when information services expand to include media such as images, maps or charts requiring high-quality imagery, audio, and video source files. While this is more a display than a functionality issue, it does affect the user experience. Information services need to ensure that multimedia content is acquired for all records, is accurate (the correct subject is displayed for a given record), consistent (images are all shot from the same angle), and high-quality (clear). This often requires an additional “fill in the blanks” effort to acquire and improve imagery.
Finally, adding a “see something, say something” prompt to your app asking end users to report errors, omissions, and anomalies is another simple way to convey a firm’s commitment to stamping out missing and anomalous data. Don’t forget to do more than just display a thank you message when an end user takes the time to assist you. Drop them a line after you’ve corrected the issue and perhaps reward them with a free special report or other sign of appreciation. This may be the single cheapest and most effective way to both improve retention rates and make your product the best that it can be.