Now that HP has announced its intent to buy Autonomy, the deal has come under a lot of scrutiny. One area though that few have yet to look at in detail -- and of particular interest to us as buyers' advocates -- is the whole topic of the IDOL search OEM business. Much has been made of the scale and pervasiveness of Autonomy's IDOL search engine platform, and that it is OEM'ed so widely that it has become a defacto standard.
I wonder if this is really the case.
When a vendor OEMs a product, this typically means that they are embedding a subsystem or specialized technology from a 3rd-party supplier. The key thing here is that embedding someone else's technology can be better, cheaper, and easier than building that same functionality yourself. So for example many ECM vendors embed Oracle's (formerly Stellent) INSO document viewer in their offerings. Another ECM example would be the OEM of Crystal Reports, to provide reporting capabilities drawing from the base audit trail in an the ECM platform.
The questions raised in Autonomy's case are twofold: first, how do you embed IDOL at all, and secondly how many people are actually doing so?
Leslie Owens over at Forrester recently wrote a piece on this issue, and described IDOL in terms of a vacuum cleaner. It's an excellent analogy. You see, IDOL does not embed itself in other vendor's products as much as it sits in the corner and sucks data out for processing. It is essentially a separate black box, and as such is one that could be easily replaced by any other search engine.
So, far from other vendors becoming dependent on IDOL, it is in fact loosely coupled and easily swapped out for alternatives. For example, Oracle and Sybase have already replaced Autonomy search technologies with alternatives. Many others have followed suit, particularly since there are now viable and considerably cheaper open source options in the various forms of Lucene/Solr. And finally, it's important to note that where Autonomy is present in 3rd-party software, it is more typically the old (and very basic) Verity engine, not IDOL.
The argument goes that IDOL may well be expensive and complicated to use, but at the high end of the market it is the best, or as other analyst firms like to say it a "market leader." I can't help but wonder whether this is a victory of marketing over substance. Sure there may be some organizations (the much vaunted law enforcement / security / intelligence buyers) who fully leverage IDOL, but what about the much greater mass of regular enterprises and public organizations? In our experience most of them are struggling with basic search, with both Microsoft options and the Google Search Appliance prevalent throughout pockets of their organizations.
For buyers, the HP acquisition of Autonomy could prove enlightening. It will doubtless prompt a shakeout and insight into the true nature of a very secretive company, but more importantly it is reawakening. It reminds us that there are many enterprise search options available, that there is no defacto standard, and that search in general still typically under-delivers.
If nothing else the move by HP should be taken as a time for current IDOL buyers to consider their options, and to recognize that they in fact do have many options available ranging from Oracle and IBM to specialists such as Recommind, Exalead, Coveo, and ISYS, some of which are much cheaper and easier to use than IDOL. Hopefully, the marketplace will create still more options in the coming years, as most customers would testify that they're not satisfied with their incumbent systems...