Market concentration – or fragmentation – is an important big picture view of your market. Insights into the nature of competition for your customers will help you make decisions about your product. But only if you correctly define your market.
HHI – Measuring Market Concentration
I wrote about the mechanics of measuring market concentration, using the HHI – the Herfindal-Hirschman Index – last year.
The US government has declared that markets be classified based on the following ranges of HHI values:
- HHI below 1000 – a competitive market. There are no dominant competitors in this market.
- HHI between 1000 and 1800 – a moderately concentrated market
- HHI above 1800 – a concentrated market. There are one or more dominant competitors in this market. Higher HHI values translate into fewer, more dominant competitors.
The key ideas from that article can be summed up:
- Larger values of HHI represent markets with fewer, more dominant, competitors.
- Small differences in HHI values are meaningless – only large differences matter.
There’s also a fantastic discussion thread – thanks to Roger Cauvin (@rcauvin) and David Locke (@davidwlocke)! – that started by asking how knowing the HHI of your market could affect positioning, and quickly changed into a discussion about what positioning is.
HHI – What Are You Really Measuring?
Here’s the tricky part – what are you really measuring?
Lies, Damn Lies, and Statistics – (not originated by Twain or Disraeli, to my surprise) – refers to selectively promoting a measurement that supports your argument. Product managers would never do that. At least, not to themselves. The trick is to measure market concentration in a way that is useful to you – measure something that helps you understand your market and inform product decisions.
Consider the Android mobile operating system and the global smartphone market. The following diagram shows the global market share in Q2 of 2010 for each major smartphone operating system. This market is concentrated, with an HHI of 2712.
The next diagram shows the (presumably global) share, within the Android slice of the market, for different Android devices. This data comes from TweetDeck’s recent analysis of platforms on which their mobile application was installed. This analysis shows a competitive HHI value of 1253
This second data set is definitely biased, in that it represents a subset of the Android-purchasing customers – specifically, the ones who are both early-adopters and who are active users of social media.
If you look at a similar analysis of Android device share data from AdMob, you get a very similar looking graph – but with different handsets in the top 10. The HHI analysis still shows a (slightly less) competitive market with a 1508 rating.
In fact, there’s no (mathematically) discernible correlation in the two Android-device-share data sets.
[Tangent: Since the AdMob data represents "all users", this becomes an interesting data point that undermines the conventional wisdom that the early-adopters (especially the social-media-savvy, and presumably highly vocal ones) are the thought-leaders that influence mass-market purchasers. Time to brush up on my Geoffrey Moore and Malcolm Gladwell! Do they predict this directly, or is that the conclusion people reach from reading their works?]
You now have two different HHI numbers for trying to understand Android device competitiveness in the global smart-phone market – representing two different assessments of the market.
Is this market fragmented or concentrated?
Asking What Matters for Your Product
Which analysis better informs the decisions about your product? Are the customers in your target market users of any smartphone operating system? Or are you focused explicitly on a segment of smartphone users who have already self-selected as members of the Church of Android?
If you’re building an Android application (that can be downloaded onto Android devices), you’re probably focused on the handset-share data (within the universe of Android handsets). If you’re developing a product that integrates smartphones with car electronics / stereo systems, you probably care about the overall smartphone OS share data.
If you’re developing mobile advertising, operating-system share (and by extension, advertising “real estate” by ad-network share) is probably most important to you – although variations in form-factors within handsets can also be useful data.
When you’re developing an understanding of how your market operates, including the concentration of competitors in your market, make sure you’ve got the right definition of your market. Otherwise, you’ll just be basing your decisions on a stack of lies.