Things Not To Do: Declare Your Category Dead

I was reading this interesting post, Emerging Enterprise Content Management Trends, on the Gilbane Group blog this morning when I stumbled into this rather amazing soundbite.

Jeff Fried, VP Product Management for Microsoft’s FAST search engine actually proclaimed that “keyword search is dead!”

Now, last I checked, Fast was doing around $50M — oh sorry, I mean post-correction of accounting irregularities, $35M a quarter in enterprise search revenue and that Microsoft paid $1B for the company in order to do a “best defense is a good offense” strategy vs. Google.

So regardless of what Brother Fried or his PR mavens think, I can assure you that Microsoft doesn’t think keyword search is dead. Oh, and did I forget to mention, as they say in Brooklyn, Bada-Bing!

One of my pet peeves is people or companies who think it’s cool or controversial to declare themselves dead. Why?

The first time I heard something was “dead” was in 1987 at the original Ingres. The thing in question was the relational database. At one company meeting, our executives patiently explained to us unwashed employees that because of the ANSI SQL standard relational databases were “commoditizing” and ergo that we would be de-investing in the core RDBMS engine and instead investing in application development tools.

I’m not sure there’s a font big enough to write this, but OOPS.

In 1987, the RDBMS market in total was maybe $200M. Today it’s a approximately $10B oligopoly shared by Oracle, IBM, and Microsoft. Application development tools are somewhere between 1/10th and 1/100th the size and a high fragmented market by comparison.

What went wrong?

  • Lack of understanding of product differentiation. Yes, the products were arguably becoming more similar due to the SQL standard (e.g., Ingres still primarily spoke Quel at the time), but more similar != identical != commoditized. The possibilities of high-speed, low-cost, parallel-optimized, query-optimized, platform-optimized, non-stop or a dozen other possible bases of differentiation seemed to elude Ingres management. My take: if people can differentiate white rice (e.g., regular, parboiled, in a bag, basmati, jasmine, texmati) then you can sure as hell differentiate technology.
  • Non-observance of industry structure in RDBMS. Product differences is just one piece of “degree of rivalry” in Michael Porter’s five forces analysis. Substitute threats were low (i.e., switching costs were high), buyer power was low, barriers to entry were high, and supplier power was low. By seeing only product and not seeing industry structure, Ingres missed that a huge, oligopolistic market was in formation. (Only Oracle seemed to really get that a landgrab was in progress, that switching costs were high, and that the goal should be to get as much market share as possible in the short term — even at the cost of making a mess — which you could then sort out later.)
  • Missing industry structure in application development tools. The flip-side of the attractive industry structure in RDBMS was a rather appalling industry structure in application development tools. Barriers to entry were low, competitors were numerous, the large number of competitors was putting downward pressure on prices, and the dawn of free runtimes was already under way. Simply put, it’s very hard to make money in tools and that’s one reason why “tool” really is a four-letter word on Sand Hill Road.
  • Confusing being “up the stack” with “value”. One argument is that RDBMS is just plumbing and that tools are higher in the stack and ergo deliver more value and more potential for profit. This is wrong. Why? Because while tools are indeed higher up the stack, profit potential comes from industry structure, not stack altitude. Microsoft makes plenty of profit and they are at the bottom of the stack. The Oracle DBMS business in one level up and is a key driver of Oracle’s 40%ish operating margins. There are many misconceptions about the applications business in this regard, but I won’t go there now. See the Profit Zone for more on this general topic.
  • Too much emphasis on vision. If your vision for the future goes out so far that we’re all dead, then perhaps you should dial it back a bit to make it useful. Yes, we’re all dead in 100 years and one day RDBMS services may be as commoditized as electricity. But, some 20 years later, RDBMSs are nowhere near a commodity and a lot of people have made a lot of money in the meantime. It’s not predicting the eventual end that’s the hard part. It’s figuring out what happens along the way and how to make money during that evolution.

So before you go ahead and declare your business dead, ask yourself some questions:

  • Am I doing this for PR soundbite? If so, is it really the kind of message you want to communicate? Is this the best you can do to sound visionary?
  • If you really believe it, then should you turn in your badge and let someone run the business who doesn’t?

6 responses to “Things Not To Do: Declare Your Category Dead

  1. Daniel Tunkelang

    Dave, I'm hardly the person who should step up to defend Fried or Microsoft / FAST, but I think you're reading a lot into a glib sound bite. I'd read it more as an assertion that the market is moving beyond the "10 blue links" search interface.But that assertion itself is worth challenging. I think it's increasingly true in the enterprise (much to the benefit of companies like Endeca, my horse in this race), but hardly evident on the web, Bing's "decision engine" marketing campaign notwithstanding.I wish we were all there, but we've still got a ways to go.

  2. Dave – Surely there's a line from Monty Python in here somewhere, eh?Great point on Microsoft/Bing – interesting that they're branding Bing as a "decision engine" and built it because "everyone hates Google." Can be a typical ploy to kill a category to create another one, although I agree that seems unlikely in this case.The marketing machine of Microsoft usually fires a bit more on all cylinders, but sometimes the machinery gets a bit of sludge in it.What's interesting about the "category is dead" statement is also the flip-side of the coin, which is "that category isn't important" – specifcally, because WE DON'T DO IT (yet). Happens over and over again – "only feature x matters" until 12 months later when a company HAS that feature, and now it's incredibly important, critical even! (sigh)It's completely understandable to talk up strengths and minimize/ignore weaknesses, although it doesn't take very much critical thinking to poke holes in marketing stands like that.Good catching up with John and JT today – we should peel off some time in near term to catch up directly.Best,Dan

  3. Daniel,I'm probably one of the few marketing-background people who generally doesn't believe in glib soundbites. That is, if a spokeperson says something, the company should mean in — no soundbite defense, no Twinkie defense. That said, I admit that I went off on a bit of a rant in response to the quote, so the quote ending up sparking what was hopefully a good post.

  4. Dan,Great points. Yes, it always amazes me when vendors say [1] "thing X doesn't matter" when they don't have it, then [2] "thing X is critical" the second they have it, and then [3] "the new implementation of thing X is really critical and the old ones, as everyone knows, didn't work" and then [4] "thing X doesn't matter" once everyone else has it. Consider, the RDBMS vendors and XML. I think we're in stage 3/4.

  5. Dave:I attended the Gilbane this year and sure didn't remember Jeff making the claim. I called him, and he even checked his slides, and nowhere do they make the claim keyword search is dead. I think Larry may have misunderstood something…Miles

  6. Miles,Thanks for your comment. Frankly, since I wasn't at Gilbane, I can't attest to what the Fast guy said or didn't say. Happily, for the post, most of it was a rant related to the general topic of self-declaration of death of category which happens far too often in tech.Best,Dave

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.