As we celebrate our 10th anniversary since the founding of Researchscape, I have a newfound appreciation for our dedication to continuous innovation.
At this point, our team has logged 2,128 suggestions for process improvements and implemented 825 of them in ResearchStory Enterprise, our proprietary platform that enables us to quickly and consistently produce high quality reports. While many other research firms outsource labor to low-wage countries, our team has feet on the ground in the US and Canada. To compete with offshore labor, we need to have incredibly high productivity.
To accomplish this, we’ve automated most tasks involved with survey research, from questionnaire design, to survey fielding, to data prep and analysis.
- We use our own markup language for questionnaire design, which enables us to rapidly convert questionnaires into working surveys. For instance, we almost always randomize the list order of all-that-apply questions, to minimize order bias; our tool handles that for us, providing consistency without requiring us to remember to do three mouse clicks per question to randomize options.
- In the surveys we field for our clients, we ask respondents to rate the questionnaire and welcome any optional feedback. We integrate this into future surveys; for instance, we’ve increased the font size we use, and we now ask some types of personal questions less often. We’re deeply thankful to our respondents—211,000 last year alone—who enable us to do the work we do.
- Once the data has come in, we run a range of diagnostics on respondent behavior and answers as part of our data review, to ensure our results are as representative as possible. This, along with weighting, is vital to ensuring quality results.
- Finally, survey reports are generated for clients using our groundbreaking AI platform, ResearchStory Enterprise. This is where most of our process improvements have been made. Almost all the steps that we used to do manually are now automated. For instance, we have different chart preferences for color scheme and chart type depending on question type; the software handles all these automatically. It generates an executive summary, uses named sections for the report, and curates top crosstabs.
At this point, reports often pass the Turing test (a test of a machine’s ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human). Clients assume we’ve spent a day or so on these reports, but in reality we’ve instead automated as many things as we would have done in a day. And that’s where our ten years of experience in this business enable us to serve our clients a little differently. Every time we do a task, we note those parts that could be automated and create a suggestion for that, or—if such a suggestion already exists—upvote it.
When we present results, we note those points that confused the audience and document them. This has led us to embrace Bayesian averages, proportional rounding, different visualization techniques, and hundreds of minor improvements to wording and presentation that have added up over time.
So each project we complete reflects a decade of learning and continuous improvement, and each project teaches us something to make future projects better—that continuous innovation cycle I noted at the onset.
Our dedication to automation also enables our senior team to treat clients a bit differently. For many boutique firms, the principals are involved in the sales process and then rarely seen again. Our automation allows us to be there for the client throughout the process.
Thanks for the chance to work with you these past ten years, and we look forward to what we will learn in the next decade. Have any ideas for us to prioritize? We’ll put them ahead of the 1,303 ideas we want to implement!
Author Notes:
Jeffrey Henning
