He still has pride-of-place among my Legaltech swag
20 January 2020 (Paris, France) – When Hewlett-Packard (it changed its name to “HP” in 2015 and I’ll use that from this point on) bought Autonomy in October 2011 for $11bn, the e-discovery community was agog. Scores of e-discovery pundits called it a “game-changing” acquisition that would revolutionise the e-discovery world. I was at the Enterprise Search Summit in London a few weeks later and analyst after analyst called it a “structural panic buy”.
That point is important. As I wrote in a long post at the time, HP wanted to make itself more like IBM, and very quickly, and it thought that Autonomy’s number-crunching technology could take on I.B.M. in data analysis. But HP executed in all the wrong ways, with the lack of preparedness, lack of cooperation and lack of technology understanding quickly becoming evident. The web has scores of “behind the scenes” stories. At IBM, Gerstner refused to act precipitously after taking over in 1993, not dismembering Big Blue as he was urged to do by many IBM insiders and industry pundits. He chose to go a different way, had an alternative vision … with great success … but he made damn sure it would work in practice by playing thousands of scenarios, learning about every member of his key teams before making any final decisions .
Shortly after the HP buy I had a chat with Nick Patience, one of the founders of 451 Research (recently acquired by S&P Global) who said his team had watched Autonomy very closely through the years and the numbers just did not add up. He said there had to be “a willing suspension of disbelief”. As an example, he pointed to something he had written in July 2010 on a recent Autonomy quarter (click here). He was spot on.
Alas, as we learned, from scores of technology analysts who knew the industry, it was all “vaporware wrapped up in fancy Cambridge talk” with the kind of accounting tricks managers have engaged in since the dawn of publicly traded stock. The former finance chief of Autonomy would be sentenced to five years in prison for accounting fraud linked to the sale. Mike Lynch, former CEO of Autonomy, is still subject to ongoing litigation.
HP’ s claims were investigated by the UK Serious Fraud Office, but the SFO announced in January 2015 that it was ending its investigation with no action.
Lost in all the analysis at the time was “can Autonomy products do the job?” For many software analysts Autonomy’s core enterprise search and data analysis technology was viable but it was very complicated and expensive, and other vendors could do it. Several analysts wrote about the October 2012 HP demonstration of the Autonomy product used in an application to help figure out what to pay for Web ads. Initially it failed to work, and then delivered results that appeared similar to those of several other search products tasked to do the same thing.
I saw how some of the software was used when I was involved with the SocGen/Kerviel trial. I learned how SocGen had used Autonomy software in that massive bank fraud case to work out what had been going on with Kerviel and his trades, as well as his communications — the same technology used for early case assessment, document review software, and social media analysis. It was interesting software and it afforded me my first opportunity to dig deep into the enterprise software market and to see how Autonomy sought to commercialize statistical techniques (developed by Lynch at Cambridge University) based on Bayesian inference, a mathematical technique that can estimate the probability of potential outcomes based on previous evidence. Hat tip to the 451 Research team who guided me through that.
But for the e-discovery community, which at times seems to be so gullible to hype and divorced from reality, it was all a bitter, bitter disappointment. Well, other than for the litigation lawyers. As Craig Ball opined “it can be safely concluded that the acquisition will indeed pump millions into the e-discovery economy … as HP sues everyone in sight”.
Ah, but there are still lessons to be learned.
I was scouting around for material to support a brief I am co-writing for an e-discovery practice fraud case (ta da!) and I found a presentation prepared by Qatalyst Partners prior to Hewlett Packard’s purchase of Autonomy in 2011. The slide deck covers:
• Historical trading performance and related financial data, and how the search market works
• Shareholder ownership
• Comparative financial data; for example, Google, Oracle, HP, and other firms.
There is actually a “first” Qatalyst Autonomy presentation which you can find on the Vdocuments web site (click here). You will be able to locate other Autonomy documents if you do some scouting around the site, a site I have noted in previous posts. You can find oodles of stuff on the search and text analytics industry, and other fascinating technology industry items that people love to share for our greater education.
The venture capital partnership in which I have been a limited partner for over 17 years uses that site and many others, as does a master limited partnership in which I am a shareholder. There are gold mines of information out there. Both of these funds started with blogging platforms and some content curators and then gradually moved into E-Commerce, FinTech, and mobile payment providers. Over the last 8 years we expanded into more maturing companies, joining late stage funding efforts, adding several cyber security companies plus several e-discovery vendors. I have seen some interesting financials over these 17+ years.
The Autonomy documents are almost a decade old, but they provide useful information for anyone considering an investment in (or purchase of) an organization engaged in enterprise search and/or text analysis software. Documents like these provide some of the factual foundation serious analysts use for reports and deep-dive analysis. It is far easier to talk about the revenue potential of search and text processing. It is far more difficult to generate sustainable revenue and growing profits. Why? Well, lots of reasons but just a few points:
• Ignoring the highly particularized nature of search and text analysis. That is, “one size fits all” doesn’t work, so expensive, one-off tailoring is required.
• Making a “try our innovative search or text analysis” sale is time consuming. The reasons range from “we have been burned before” to “this got the previous information people fired.” Go to Legaltech a few years in a row and see who has disappeared.
• Keeping the search and text analysis system up and running is expensive.
• Staying competitive is very expensive. Innovation is easy to talk about but difficult to deliver.
• Growth requires acquisitions, and these just add to the cost of dealing with the technical debt the acquirer has to generate money to pay.
Documents like the Autonomy set are useful and often difficult to obtain. Many are on the Net. But you need to search.