The Attacker’s Dilemma

Screen shot 2012-09-07 at 5.30.21 PM

Some of you might remember how hard it was to search the internet in the mid-90s. Because search sites were mostly hand-curated, it was often diabolically hard to find even basic information.

I still remember trying to find the official site for the 1996 Olympics in Atlanta. It took about an hour, and I used at least three different search sites (yahoo, lycos & alta vista). None of them retunred a direct link to the official site, and it took forever to find a site writing about the Olympics that included a link.

It’s a wonder that we got anything done online back then.

So when Google arrived with it’s search algorithm, it was able to overtake the most popular site on the internet at the time (yahoo) almost instantly. That’s because if you searched for a phrase like “Olympics official site”, it tended to return the official Olympic site in the top spot.

That was much more than a 10X performance improvement, and that’s why Google was able to take on the giants of the internet, in their areas of strength, and still win.

So what will it take to knock off Google?

That is something that Microsoft has been trying to do with Bing. A couple of weeks ago they launched the Bing It On challenge, which enabled you to enter search terms and compare the results from Google & Bing side by side. Then you picked which results you thought were better.

Here are the results from mine (sorry for the very egocentric searches! It seemed safest to stick with topics that I know well):

You can see that I preferred Google’s results to Bing’s in 3 out of the 5 searches. But the big question is: even if I had preferred Bing in 5 out of 5, would I switch?

The answer is: almost certainly not.

You probably wouldn’t either.

There are all kinds of reasons for this, including:

  • The differences are tiny. On most of these searches, 7 or 8 of the top 10 ten sites returned were the same across both. So when I picked one over the other, it tended to be because I liked the 9th result on one search better than on the other. This difference is trivial. Even if I preferred Bing on all 5 searches, a 10% performance improvement usually isn’t enough to justify a change, because:
  • Changing search engines changes your workflow. Google is strongly integrated into my workflow. I use Google Chrome as a browser, I read my RSS feeds on Google Reader, get email through gmail, etc. If I change search engines, I end up needing to change all of these at well. Even if I don’t, they won’t perform as well. So if I’m going to disrupt my entire workflow, I need a lot more than a 10-20% improvement in basic search results.
  • The search problem has been solved. This is the big one. I had a big search problem in 1996. Now I don’t. The search problem has been solved – at least the basic one has. I can see ways in which we can still improve search – but a slightly improved algorithm doesn’t address these ideas.

All of this adds up to the Attacker’s Dilemma. And that is: unless you bring a major performance improvement, there is no point in directly attacking a strong incumbent in their area of strength.

You need to find a niche that isn’t being served. You need to find area where you can build a learning advantage. The actions you take when you are entering a market are quite different from those that you take when you are building one.

The Bing it One challenge would have been great in 1998. This would have been a great promotion back when the algorithmic search market was still being built.

Now? Not so much.

Student and teacher of innovation - University of Queensland Business School - links to academic papers, twitter, and so on can be found here.

Please note: I reserve the right to delete comments that are offensive or off-topic.

6 thoughts on “The Attacker’s Dilemma

  1. Sound advice, Tim, and to take it a step further, the marginal improvement on algorithm is all but squashed when you consider the emotional element of change. Even if we were entirely rational, how we feel about Microsoft vs. Google weights the data one way or the other.

    Trying to build a new model isn’t easy, either. Even when you’re dedicated to serving the customer more than anyone else, there’s still so much marketing driving them to the old way of doing things, they gladly linger at the stale buffet of mediocrity. Sigh.

    • MS v. Google is a good point Brian. So is the stale buffet of mediocrity, bu that’s an entirely separate issue! Lock-in is a real problem in many of these domains.

  2. Great post Tim. I think there are similarities here with the Pepsi Challenge. With the blind fold on, I and most every other person who took the Challenge prefer Pepsi. But take if you take the blind fold off, I and most others, prefer Coke every time. Now if we look at disruption in the soft drinks business, Red Bull springs to mind as a classic example. It focused on delivering new measures of performance in areas that Coke would never think of. Additionally, it tasted awful!! So 10x greater performance is a great target to take on an incumbent or alternatively identify completely new measures of performance, ones don’t exist in that category yet.

    • Very good points Iain – thanks!

      I thought of the Pepsi Challenge as well when I was writing the post. I too preferred Pepsi, but both with and without the blindfold. But you’re right – similar issues.

      And you’re absolutely right about identifying completely new measures of performance. In some ways, Jolt Cola was trying the same thing as Red Bull, but they probably made a mistake by making a cola – so then it was just an awful tasting cola instead of an awful tasting something completely different…

Comments are closed.