A number of positive developments in enterprise search took place in 2018: Elastic had a very encouraging IPO, Google rediscovered enterprise search and Microsoft showed signs of acknowledging that its potpourri search strategy was not fit for purpose. At long last the information retrieval community (well, at least my side of the Atlantic) turned its attention to gaining a better understanding of how to optimize enterprise search, and thanks to the combined (as of Dec. 21 2018, literally) commitment of Charlie Hull at Flax and Doug Turnbull of OpenSource Connections to relevance engineering, two new conferences emerged and are set to become permanent fixtures on the search calendar.
Will this momentum continue in 2019?
Related Article: The Next Generation of Enterprise Search Is in Sight
Google Cloud Search – At What Price?
Six months after the launch of Cloud Search and its pricing model is still not visible. My sense is Google will take a "document/item" approach, much as it did for its Google Search Appliance product. I just can’t see that working. Of course the investment is not just in the software license but also in the integration support. Google lists almost 50 implementation partners, but as I look through these partner's websites I see very little information about the extent of their enterprise search expertise.
When you reach the scale of an enterprise search implementation, the challenges are always multi-lingual and multi-country. Finding a single implementation partner with the optimum balance of skills and expertise is going to be difficult. Search implementation skills also need to be aligned to a good knowledge of the business sector and its processes.
Microsoft – An Integrated Search Platform?
Microsoft bought FAST Search back in 2008. At the time, FAST was arguably the best search application on the block. Microsoft then proceeded to emasculate it and eventually bury it as a stand-alone application.
Ten years later, Microsoft is now making promises for a new, integrated search platform. Microsoft clearly has the technical capability to offer this, but do its customers care? None of them bought SharePoint as an enterprise search solution, and very few have subsequently invested in a search team with the skills to optimize search performance. Microsoft integration partners also appear to have limited skills (see: Google's partners) and often do not easily operate across multiple countries. And I won't even touch the challenge of multiple languages. Google at least does not have legacy applications to work around, but in the case of Microsoft there is a huge range in implementation maturity.
Optimizing Search Performance
In this area, I am much more optimistic. As I noted in "Achieving Enterprise Search Satisfaction," there's been an upsurge in academic research into enterprise information seeking and search, and I am certain this will continue. This research is not just about optimizing search technology but about understanding the role search plays in the enterprise. In this respect the Salton Award Keynote: Information Interaction in Context given at SIGIR 2018 by Kalervo Järvelin is an essential read. The same issue of SIGIR Forum December 2018 includes a report on the SIGIR 2018 Workshop on Professional Search and a very valuable discussion paper on what should be taught to students in information retrieval courses. While the lack of enterprise skills is concerning, one of the paper's authors, Marteen de Rijke, assured me he would welcome the views of the wider search community. I would encourage you to look at the paper in some detail, and perhaps consider if your own information retrieval skills are as broad as they need to be!
Open Source Search
In my opinion, the Elastic IPO will not have a significant impact on open source enterprise search other than slightly raising the visibility. Elastic, like Google and Microsoft, seems to me to view enterprise search as something of minor value in the greater scheme of revenue and profit success. It's encouraging to find vendors, such as those in the intranet-in-a-box business, starting to use Elastic as the search component, though the recent report from Clearbox Consulting indicates the search implementation in these products is still not as good as it needs to be.
One positive step has been the merging of Flax and OpenSource Solutions. Their Haystack events on relevance engineering good practice will undoubtedly become established and very valuable events in the years ahead, but these are still very small companies punching well above their nominal weight. I wish them every success.
Related Article: Enterprise Search Has an Open Source Secret
While the alliteration may be pleasing, the reality is not. A recent applications audit in a small (2000 employee) but globally-located client revealed over a dozen applications that broadly speaking support collaborative working, terabytes of shared drive content and four video-conference applications. Each application is loved by its users but finding out who can give permission to join a work group is a challenge in its own right.
When it comes to delivering a decent search experience, this could be my most challenging assignment to date. Federated search is more an art than a science. Add to the mix multiple silos, five languages and an acquisition-hungry company and I think it’s going to be a busy but fascinating year ahead. I hope it is for you as well.