I have had several individuals ask us what the steps are to getting value out of Eureka over the past year. Although there is not a secret sauce, other than surrounding yourself by smart, hardworking, inquisitive individuals, I have decide to post how we think about and approach a problem/initiative in a series of steps. I do not impose a time-frame on these steps, but I do insist that they are followed before jumping in and writing code. This guide has allowed us to have a clear agreed on direction that keeps us focused on the goal, in a way, keeping us grounded and not spend time in vain.
Think about the programming approach you want to take to achieve the desired outcome. Is your purpose for compliance, measurement, behavioral change, improved speech clarity, procedural/sequence of events, etc.? Answer the what, why, how, when and who. What filters will be included/excluded, is there a specific time in the interaction that this happens, any specific department or time of week/month/year, what metadata can you use, are you going to use Semantic Building Blocks, will you create a score or report, what resources can you use (get other’s ideas), how much effort will this take, what is the expected ROI, etc. See the I wish I had a category for... post for more details on programming approaches.
This is all about brainstorming/thinking. In my opinion, this is the most valuable time. Before you place your fingers on the keyboard to code in what is initially on your thoughts, spend time (sometimes a week or more) thinking about the situation. What words/phrases do you expect? What other ways can it be expressed? Are there others who can provide insight? When does it happen? How would a consumer vs. an agent say this? Can the word/phrase have another meaning? What are the aliases?
Code your syntax and use Discovery methods to find out ways things are said that you did not come up with in the brainstorming stage. What other words are around the ones you came up with? Does the meaning change in different context? How do you eliminate false positives? Exclude the words/phrases you coded to discover what else is there? Are there phrases that you did not capture? Have you coded in the present/past/future tense of the words? Did you use the proper Distance operators? Proximity operators? Location operators?
Save as a Category. I know this may be the uncomfortable step. You think, “what if”, but who is going to see the results at this point. This is something we learned well past when I wish we did. Initially we saved everything as a Search to fine tune it to perfection, which is nearly impossible. Save the code as a Category and let it tag the contacts, you can always tune it as you go. Ultimately, if you did the previous steps well, you are going to capture a vast majority of how things are said. People are fairly predictable. Plus, once you have some volume of contacts pass through your Category, the left pane will “tell a story” and any obvious misses will show their ugly heads and you can then use the Category as a negative filter to find the other “needles in that haystack”. Let the Category bake for enough time to get a true sense of what you have tagged and then make any adjustments as needed, but you may surprise yourself and deserve a pat on your back for nailing it.
Deliver the end product, be it a report or score and request feedback. Make sure you communicate the benefits and limitations to the end user. Follow up and measure if your product fulfilled your strategic vision that was envisioned in step 1. Make adjustments as needed as you receive feedback or the environment changes. Are their new in policies/procedures, regulatory, department, service, product changes that may require modifications? Are there changes to filters used in this Category that may affect the outcome? Feedback and Maintenance are the keys to this step.
I could probably add a lot more content to this post as this is not an exhaustive list, but I think this provides a general direction on how we think about projects. Of course there is so much information on Engagement Optimization, that you can probable create your own “how to guide” by extracting those pieces and putting them into a digestible format. I would start with Yeliana Liman post Start up Guide for some great directions.
Happy to hear other’s thoughts on how you think about problems/projects.