A slow and steady strategy may have proven successful for the tortoise when racing the hare, but it certainly isn’t the approach to apply to your knowledge management strategy. Whenever an employee in your organization searches for information, they need a rapid response to aid a business decision. Take a call center agent on the line with a customer, who needs access to information immediately to help resolve the customer issue. As TSIA’s John Ragsdale points out, when hold times stretch beyond the 3-minute timeframe, callers typically abandon the call and hang up.
To avoid these situations, your Insight Solution can’t only be effective in yielding the best results; it needs to do so as quickly as possible. In my last two posts, we discussed how to get started with – and how to manage requirements for – your Insight Solution deployment. While you may be armed with the best solution on the market, realize that’s just one link in a chain of integrated systems that must work together. The end result is a populated unified index that can be quickly and effectively searched by users.
Fortunately, indexing technologies are built for speed. They can retrieve the most relevant results among billions of documents in a fraction of a second and can acquire millions of documents per hour. Many traditional systems, however, are built for CRUD operations (Create, Retrieve, Update and Delete), at a relatively slow pace (and an unsavory acronym). When the indexing engine calls into a system for a copy of all documents, metadata and permissions, it can cause a much slower user performance.
To evaluate the performance you can expect without impacting users, it’s a best practice to test the targeted source systems early in the project. Some systems, like email servers, are able to sustain a great throughput (millions of documents per hour) without any impact on users. But other systems, like some ticketing systems, will suffer during massive indexing runs and often have to be throttled to avoid impacting users. Once you have tested all systems, you’ll then be able to determine the optimal initial indexing strategy and evaluate how long it should take. Since time is of the essence for your employees and customers, the ultimate goal is to have your indexing engine produce relevant results on the fly.
I have one last parting piece of advice in this best practices series. Above all, make sure you have experienced people on your team, as their thorough understanding of the product could increase your return on investment by preventing costly mistakes. We hope our advice aids you in the shortest and smoothest path to a successful deployment with maximized value creation. Are you running into other hiccups during your Insight Solution deployment, or have other questions you’d like answered? Let us know in the comments section.