handful of glitter
We all would love a brand new, functioning search platform for little to no effort. The reality, unfortunately, is not so easy. PHOTO: Anthony DELANOIX

The search implementation community has been trying to inject some much-needed reality into the search business to counteract the promises and pixie dust spread by analysts and vendors.  

The catalyst for this reality check was the release of both the Gartner Magic Quadrant and the Forrester Wave reports on enterprise search (and insight and cognitive search) vendors. Charlie Hull, managing director of Flax, started the discussion, which I quickly followed with a three part post and then Miles Kehoe weighed in. 

This isn't the first and I suspect won't be the last time the search implementation community tries to set realistic expectations in the market.

Kehoe recently dove into the complexity of enterprise search, highlighting the importance of investing in a search team, a topic I have covered in some detail in my Making Search Work report. Real Story Group founder Tony Byrne has published a thoughtful analysis of the value of industry analyst reports, a topic explored with special reference to enterprise search by systems analyst Russell Reinsch.

In a related post, search relevance consultant Doug Turnbull shows the level of detail required to make even an initial comparison of Algolia, Lucene and Elastic.  

The Parallel Universe of Search Vendors

Meanwhile, nothing changes in the search vendor business. 

Vendors continue to promote a view that search is a turn-key implementation. They demonstrate the quality of their software by its positioning on a 2-D chart created by analysts who have never been directly involved in making search work well. 

Many of us in the UK search community still recall a presentation given by Coveo at the 2013 Search Solutions conference entitled “Complex enterprise search delivered in a day. No, I'm not insane.” The delegates concluded insanity was a major factor as the speaker produced more and more caveats to justify the statement.

And then we have the offer from hosted search vendor Swiftype: “With Swiftype Enterprise Search your org can be up and running in six minutes.” 

I’m assuming the six minutes includes indexing the millions of documents my clients usually have in their repositories. The company also claims, “thousands of businesses trust Swiftype to deliver millions of highly relevant results every day.” With all these claims, why isn't Swiftype on the Gartner and Forrester charts — it seems to have an installed base second only to SharePoint? 

Software engineer David Walsh wrote up a good independent case study for Swiftype — why not tout that instead of resorting to hype?

Education Is Your Strongest Selling Point

Search vendors need to sell, that's a given. But I strongly believe they also need to educate the market. Read any search vendor's white papers and I guarantee they will have a technology focus. Everything can be fixed quickly and without issue. 

Nowhere is the hype more apparent currently than in the use of search technology to identify experts. Expert search is much more complex than any vendor would lead you to believe and has as much to do with knowledge management as it does information retrieval. (See my recent report on people and expertise search if you are interested in this topic.)

What you will not find in the white papers or on the vendor websites is guidance on what prospective customers need to consider ahead of their preliminary discussion with the vendor. 

Why is this important? Very few companies have prior experience selecting and implementing enterprise search applications. They may be using the provided search application inside Microsoft SharePoint, but SharePoint was not selected on the basis of its search capabilities.

Case studies again are high on hype and low on information. The prospective customer wants to understand what they are committing to when adopting an enterprise search application. Case studies rarely provide insight into implementation issues or how they were addressed, and above all fail to even gently suggest how important it is to have a search team in place, supporting the implementation. 

6 Minute Search? Try 12 Month Process 

I advise clients to allow for at least three four-month phases when implementing a new enterprise search application. 

Phase 1 is requirements gathering and developing a business case. The process of writing an RFP and getting to a contract is Phase 2 and can also take four months. Increasingly I find that selecting a ‘preferred vendor’ on the basis of their response to a well-constructed RFI, followed by a joint creation of the RFP by vendor and potential customer, works well. 

Phase 3 — from contract to user acceptance testing testing — can easily take four months. So expect at least a 12-month process from vision to installation. 

Don’t be fooled by anyone suggesting a legacy search application can be replaced in a much shorter time scale. The requirements for the legacy application may well date back five years. Far better to start with looking forward five years. Defining requirements will certainly take you four months, as will finding the all-important people who will make up your support team.