Understanding how analysts’ categories shift

The recent storm after Oracle and other vendors were dropped from a Gartner Magic Quadrant (MQ) is one example of the pressures on the market created by market categories. It’s certainly not an easy conversation when major vendors get dropped from categories. The process is political, both inside the analyst firms and externally when vendors lobby for new categories.

Elizabeth Pontikes, at the University of Chicago, has proposed a naturalistic picture of how market categories develop. She made a massive analysis of every electronically-produced MQ, from 1999 onwards: the highly disruptive period of the dot.com boom. Her feeling is that Gartner borrows terminologies from the vendors around them and use these to produce categories, but she can’t quite see the reality that analysts can also create and redefine new categories, as Richard Stiennon explains in his book.

However, we know that behind Gartner’s published MQs is a detailed cataloging of clients’ interests. Sometimes the framing, discrimination and drawing of these sub-markets seem like a cataloging process which can summon new segments into being. That’s not merely a question of cataloging, however, but a process of making sure that research reflects the clients’ interests. As topics rise or fall, then categories get broken down or rolled up. There are elements of functionality that define each MQ, but the divisions between the sub-segments inside each MQ can be hard to anticipate.  Pontikes says an excellent example of this is the iPad: depending on whether the iPad is or is not a computer, then Apple is or is not a leading computer company.

However, clients are not the only drivers of these divisions. Expertise has a role, as is hinted by the response to the retirement of French Caldwell from Gartner. As he left, the MQ he looked after fragmented. That perhaps suggests that his personal authority allowed his MQ to be persistent: Pontikes found that most MQs last less than a year (something which I find hard to imagine and, admittedly, the Gartner website doesn’t make it easy to see all the MQs historically). Indeed, it’s complicated by merger and acquisition. New vendors coming into an existing firm can shift an incumbent up in an existing MQ: in practice that is a new market entry of a new solution, but it’s concealed by the continuing existence of a brand. For example, BT is a small player in the UK mobile market: when it integrates EE, one of the biggest UK networks, it’s position will be transformed. That could allow BT to appear to have organically developed but, in fact, it’s a whole new business.

What happens is that as a new market develops the analysts’ categorisation gets more detailed. It grows into four, then sixteen and often 48 sub-segments at Gartner: that suggests that there’s a rule of thumb there: when markets are created as formal categories, there’s a machinery that needs to identify the sub-segments and that, in way, becomes the fiefdom than each analyst defends and extends.

However, vendors also play a significant role. In the course of my work, and more recently in my Ph.D. research, I’ve been tracking how vendors pitch to analysts and how analysts respond to that. In the next step, I am asking analysts how they how many use heuristics around their categorisation of markets and sub-markets. There are several aspects of that.

• How successful can analysts be if they impose elements of the structure of adjoining markets on to new markets, perhaps as a working hypothesis?
• How far do they assume that new products or features will be added into already-existing solutions for adjoining markets.
• How many sub-segments do they track, and how many is enough (or too many) for a market to take?
• How far are sub-segments themselves future research areas in development, in which individual expertise needs to be developed?
• How far are the segments and sub-segments props (or stalking horses) that get used for measuring (in a structured way) and assessing (in more unstructured ways) how far there might me enough potential in a new space to justify the investment needed in a new research focus area, such as a new MQ?
• Is there, as with business school rankings, a commercial imperative to create new segments and new charts to bring clients back to read more research by presenting it in different ways that (hopefully) better meet clients interests?

These patterns seem to have an impact all across the ICT and TMT space and the way in which it is segmented: the techniques of clarification; the implications of segment classification for analysts’ personal careers and team prospects; the influence of vendors and how that is mobilised; and the changing impact of customer inquiries.
That customer impact is especially impressive. Research directors that are tracking usage and customer inquiries can pressure analysts’ wishes to control their turf. Analysts can enlist vendors or users as allies against the impulses of their managers to segment research so that it better reflects the differing volumes of client inquiries. Researchers might prefer a broad conspectus since that reflects the way that vendors design their solutions.

Over the next weeks, my colleague Viktoria and I are scheduling some short interviews where Neil Pollock and I are speaking to analysts at a wide range of firms. We’ll draw the consensus in these calls, and any signs of conflict or paradox, into a draft to share more widely.