Oracle had some PR coverage this week for their launch of Oracle Secure Enterprise Search 10g, their new enterprise search solution. Oracle can now compete not only with traditional enterprise search vendors such as Vautonomy and FAST, but also with fellow database vendors who already have enterprise search solutions, such as IBM (OmniFind) and Microsoft (Index Server).
Rather than my standard schtick on the limitations of enterprise search engines as platforms for building content applications, I thought I’d instead call out the Red Herring for buying the message that enterprise search is somehow a “new” market.
Enterprise search isn’t new. It is, however, small. Not confusing the two is important. And understanding why enterprise search is small, despite not being new, is probably most important of all.
Here’s what the Herring said:
The nascent market has been dominated by startups and smaller players despite the presence of Internet search leader Google, along with IBM, which possesses a number of patents and available source code that address enterprise search.
Enterprise search is not new:
- Verity was founded in 1988. (See here for an interesting early paper.)
- Autonomy was founded in 1996, based on work dating to 1990.
Enterprise search is, however, small:
- A “big” enterprise search vendor does $30M to $50M per quarter in revenue.
- A big BI vendor does $300M per quarter in revenue.
- A big database vendor does $3B+ per quarter.
(Yes, these are order of magnitude numbers which in Oracle’s case include non-database revenues, but the point stands.)
So why — despite similar age to BI (e.g., BOBJ founded in 1990) and despite being only about 10 years younger than RDBMS (e.g., ORCL founded 1978) — is enterprise search 1/10th the size of the BI market and 1/100th the size of the database market?
Here’s my answer:
- Content has always been the third layer on the corporate IT Maslow hierarchy. Back-office systems are food. Front-office systems are shelter. Content/documents are love and belonging. This isn’t the fault of enterprise search vendors and I do expect it to change in the future as the lower-level needs are increasingly well met.
- Enterprise search products stink. Users are not happy with them. While the technology, per se, works, it doesn’t actually help people find anything. Personal example: I can’t remember if I ever successfully found anything I needed on the Business Objects intranet using our enterprise search engine.
- Many enterprise search problems aren’t. When you think about it, enterprise search is predicated on the existence of the thing for which you are searching. In reality, I think companies should invest in custom publishing systems so people can build the document they need, instead of searching in vain for something that isn’t there. (See this post for more.)
- Enterprise search is a feature, not a market. That’s why a lot of enterprise search business is embedded. For example, at Business Objects, we needed search in our portal and we licensed it from Autonomy. (I’m not sure it helped out customers find anything either, but it was a required checkbox item.) This reality is why smart search vendors are trying to build applications that solve search-oriented problems, as opposed to try and push more and more search boxes. But that’s a packaged application approach — which makes for better sales and marketing, but which is wrong technologically.
Overall, what’s needed is an application platform designed specifically for content-oriented applications. That market is indeed nascent today, but I think in the future it will be anything but small.