The announcement that Recommind had sold out to OpenText does not seem to have raised much in the way of comment. Both companies go back some way into the time line of information discovery. OpenText started life at the University of Waterloo as the outcome of a project to digitize the Oxford English Dictionary. I worked on this project in 1984 when at Reed Publishing, as readers of the 1st edition of Enterprise Search will be aware. The company now has revenues of around $2billion and the presentations to the May 12 Investor Day make interesting reading, if only because the  word ‘search’ does not appear at all!  Along the way OpenText has acquired 53 companies, many of which at the time were positioned as the next best thing to happen to the market. RedDot and Vignette come to mind. In the distant past OpenText also acquired Basis through its purchase of Information Dimensions, which was a seriously-good search application developed by a team at Battelle in the early 1970s.

Recommind is also the outcome of a research project, in this case the development of Probabilistic Latent Semantic Analysis by Thomas Hofmann, though the original paper in 1999 referred to it as Probabilitistic Latent Semantic Indexing. It is is one of a number of probabilistic topic approaches to information discovery.  Although the technology is very different to that used by Autonomy there is a common interest in finding patterns in text which will go beyond what are often described as ‘simple keyword approaches’, even if ‘simple’ is substantial misnomer. In the new world of open source search Recommind is just about as proprietary as you can get, and that makes for some problems in trying to optimise performance.Where Recommind has made a particular mark on e-Discovery is in the area of predictive coding for the analysis of texts submitted in legal cases. This has been widely adopted in the USA and is now recognised in the UK, which could have been a catalyst for the acquisition given the importance of the UK market to OpenText. The e-Discovery market is highly competitive, with kCura, FTI Technology, Nuix (of Panama papers fame) Zylab and HP along side Recommind in the Leader quadrant of the 2015 Garner e-Discovery Magic Quadrant.

OpenText has acquired Recommind for $163 million, which at (as a guess) 20-times earnings puts the company at $8m earnings on $80 million revenues. For a $2B company this is not a big buy. For comparison OpenText acquired Vignette for $310 million in 2009. What happens next is anyone’s guess, and that probably goes for the sales teams at both OpenText and Recommind given the OpenText track record. Because it is a small unit ($80 million/$2 billion) I can’t see it being retained as a stand-alone unit post the closure of the transaction in 2017. Just how Recommind is going to fit into OpenText is not yet easy to work out, as Recommind has a range of information governance applications as well as the Decisiv search application. The key executives will stay around because they will have earn-out agreements but other staff may well be brushing up their cvs.  This could make like difficult for on-going support for Recommind clients as there is very little external expertise available from search system integrators. It will also be interesting to see what happens to the recent partnership between Recommind and BAInsight.

In an ideal world OpenText would be wise to capitalise on the innovations that pervade the Recommind technology and make wider use of it in other ECM applications. Somehow, based on the history of the 53 other acquisitions, I’m not going to hold my breath. Based on the Investor Day presentations ‘search’ is not a core element of OpenText strategy.

Martin White