Big Outfits Buy Search Vendors: Does Chaos Commence?
May 25, 2012
I don’t want to mention any specifics in this write up. I have a for-fee Overflight on the subject. I do want to highlight some of the preliminary thoughts the goslings and I collected before creating our client-focused analysis. This write up was sparked by the recent news that the founder of Autonomy, which HP acquired for $10 billion, is seeking new opportunities after eight months immersed in the HP way. See “Hewlett-Packard Can’t Say It Wasn’t Warned about Autonomy.” This write up contained a remarkable statement, even when measured against the work of other “real” journalists:
Some will say this is a classic case of an entrepreneurial business being bought by a hulking, bureaucratic institution which failed to integrate it and failed to understand its culture. Others will say HP, desperate to do a deal, simply overpaid for a company that was going to struggle to maintain its sales and earnings momentum and was deluded about its abilities. Certainly warnings about the latter were there for HP to see before it handed over all that cash. Here’s what Marc Geall, a Deutsche Bank analyst who used to work at Autonomy, said in October 2010 about the business model: “…investment in the business has lagged revenues… [which] could affect customer satisfaction towards the product and the value it delivers.” He went on to warn that Autonomy’s service business was “too lean” and that it “risks falling short of standards demanded by customers”. All of which prompted Geall to question whether the company needed to change its business model – “traditionally, software companies have needed to change their business models at around $1bn in revenues”.
Yep, now the issues are easy to identify: the brutal cost of customer support, the yawning maw of research and development, the time and cost of customizing a system. The problem is that these issues have been identified. However, senior managers looking for the next big thing are extremely confident of their business and technical acumen. Search is a slam dunk. Heck, I can find what I want in Google. How tough can it be to find that purchase order? That confidence may work in business school, but it has not worked in the wild-and-crazy world of enterprise search and content processing.
Think back to the notable search acquisitions over the last few years. Here are some to jump start your memory:
- IBM in 2005 and 2006 purchases iPhrase (a MarkLogic precursor with semantic components) and Language Analysis Systems (a next generation content processing vendor)
- Microsoft which acquired Powerset and Fast Search & Transfer in the 2008 to 2009 period. Both vendors had next-generation systems with semantic, natural language processing, and other near-magical capabilities
- Oracle acquired TripleHop in 2005, focused on its less-and-less visible Secure Enterprise Search line up (SES10g and SES11g), then went on a buying spree to snap up InQuira (actually the company formed when two weaker players, Answerfriend Inc. and Electric Knowledge Inc., merged in 2002 or 2003, RightNow (which uses the Q-Go natural language processing system purchased in 2010 or 2011), and Endeca, an established search vendor with technology dating from the late 1990s)
- SAP snagged some search functions with its NetWeaver buy in 2004 which coexisted in a truce of sorts with the SAP TREX system. SAP bought Business Objects in 2007, the company inherited the Inxight Software, a text analytics vendor with assorted wizardry explained in buzzwords by marketing mavens.
So what have we learned from these buy outs by big companies? Here are the observations:
First, search and content processing does not behave the way other types of software learns to sit, come, and roll over. The MBAs, lawyers, and accountants issue commands like good organizational team players. The enterprise search and content processing crowd listens to the management edicts with bemusement. Everyone thinks search is a slam dunk. How tough can a utility function be? Well, let me remind you, gentle reader, search is pretty darned difficult. Unlike a cloud service for managing contacts, search is not one thing. Furthermore, those who have to use search are generally annoyed because systems have since 1970 failed to generate answers. Search outputs create more work. Usually the outputs are mostly wide of the mark. Big companies want to sell a software product or service that solves a problem like what is the back log for the Midwestern region or when did I last call Mr. Jones? The big companies don’t get this type of system when they buy, often for a premium, companies which purport to make content findable, smart, and accessible. So we have a situation in which a sales presentation whets the appetite of the big company executive who perceives himself or herself as an expert in search. Then when anticipation is at its peak, the sales person closes the deal. In the aftermath, the executives realize that search just does not follow the groove of an accounting system, a videoconferencing system, or a security system. Panic sets in, and you get crazy actions. IBM pretty much jettisoned its search systems and fell in love with open source Lucene / Solr. Good enough was a lot better than trying to figure out the mysteries of proprietary search and how to pay for the brutal research and development costs search requires.
Second, search is a moving target. I find that as recently as my meetings with sleek MBAs from six major financial firms, search was assumed to be a no brainer. Google has figured out search. Move on. When I asked the group how many considered themselves experts in search, everyone replied, “Yes.” I submit that none of these well-paid movers-and-shakers are very good at search and retrieval. Few of them have the time or patience for old fashioned research. Most get information from colleagues, via phone calls which include “I have a hard stop in five minutes”, and emails sent to people whom they have met at social functions or at conferences. Search is not looking up a phone number. Search is not slamming the name of a company into Google. Search is not wandering around midtown Manhattan with an iPhone displaying the location of a pizza joint. Search is whatever the user wishes to find, access, know, or learn at any point in time and in any context. Google is okay at some search functions. Other vendors are okay at others. The problem is that virtually all search and retrieval solutions are okay. People have been trying for about 50 years to deliver responses to queries that are what the user requires. Most systems dissatisfy more than half their users and have for 50 years. A big company buying a next generation search system wants these problems solved. The big company wants to close deals, get client access licenses, or cloud transactions for queries. But the big companies don’t get these things, so the MBAs, lawyers, and accountants are really confused. Confused people make crazy decisions. You get the idea.
Third, search does not mean search. Search technology includes figuring out which words to index in a document. Search does a miserable job of indexing videos unless the video audio track is converted to ASCII and then that ASCII is indexed. Even with this type of content processing system, search does not deliver a usable output. What a user gets is garbled snippets and maybe the opportunity to look at a video to figure out if the information is relevant. Search includes figuring out what a user wants before the user asks the question or even knows what the question is. One company is collecting millions in venture money to achieve this goal. Good luck on that. Search includes providing outputs that answer an employee’s specific question. Most systems provide a horseshoe type of result; that is, the search vendor wants points for getting close to the answer. Employees who have to click, scan, close, and repeat the process are not amused. The employee wants the Smith invoice from April, not increased risk of carpal tunnel problems. The poobahs who acquire search companies want none of these excuses. The poobahs want sales. What search acquisitions generate are increased costs, long sales cycles, and much friction. Marketers overstate and search systems routinely under deliver.
Who cares?
Another enterprise search train wreck. The engineer was either an MBA, an accountant, or a lawyer. No big deal. Just get another search train. How tough can it be to run a search system? Thanks to http://www.eccchistory.org/CCRailroads.htm
Well, the executives selling big companies a search and content processing just want the money. After years of backbreaking effort to generate revenues, the founders usually figure out that there are easier ways to earn a living. If the founders don’t bail out, they get a new job or become a guru at a venture capital firm.
The acquiring company’s chief financial officer cares. This is the man or woman who wrote a check for the acquisition or maybe cut the deal with the bank to borrow the money to get the deal. This is the person who has to figure out how to pay for spikes in customer support, requests for additional computing services, field demands for headcount increases. What the heck is going on? Well, sales people will report:
- Prospects are not converting to customers quickly. Instead of a two month cycle, a search sale can take six months or a year to close. Bad news for sales people who want their commission checks. It doesn’t take long for a sales person to quit selling search and start selling products and services that close quickly. Sorry, sales people have to pay bills.
- Existing customers are generally angry and only get more angry over time. When a search customer gets fed up with a system which runs slowly or not at all, the customers quit paying their bills. Bad news all around. One big search vendor is undergoing investigation for alleged criminal behavior when customers quit paying their bills.
- Engineers quit. A person with knowledge of a search system can start an integration and services firm or reposition himself or herself into a text analytics or business intelligence person. When the brains walk out the door, a non search engineer is going to have a tough time getting up to speed and then solve a problem.
- Competitors move ahead. Suddenly the acquisition looks like a week old chicken leg left in the fridge. In order to revivify the chicken leg, one has to work against nature, bacteria, and decay. Last time I checked, the only way to salvage a legacy search system was to freeze it and buy or license a new system. The cycle then repeats itself.
What’s the net net?
Search is tough. Big companies do not learn from their mistakes or the errors of others. The current crop of senior managers are too busy to do research. Why not plug the query into a mobile phone and make a decision. The outcome certainly cannot be worse than the track record of search missteps in the last seven years. Honk!
Stephen E Arnold, May 25, 2012
Sponsored by Polyspot
Comments
2 Responses to “Big Outfits Buy Search Vendors: Does Chaos Commence?”
-
- Subscribe to Beyond Search
Feature archive
News archive
-
I absolutely love this article. Great work! I would have more to say, but I agree 100% with your analysis, and it would be either redundant or dilutive.
This article is great. You have really captured the “realities” of search well.
Indeed, there are many “decision makers” who think that search is just a box where you type in what you want and get the correct result listed at the top of the page.
And I liked your description of the search solution procurement.
By the way – I purchased your brilliant report “The New Landscape of Enterprise Search”, and it certainly is valuable.
Thanks