Wednesday, 20 August 2008

The economic efficiency of sport

Should we spend so much money on sport?

Hilariously, now that the Olympics have started and nobody can have the conversation about how badly Britain will do, a new topic arises: are we too good at cycling?

This question is sometimes framed like this: are we spending too much money on something as trivial as sport? And sometimes it is more like: should we be doing better in 'real' sports such as athletics instead of in sports that "young African men can't afford to play"? Either way, those who choose to spend money on training elite sportspeople are rightly asked to justify it.

Perhaps the standard answers are valid:
  • Olympic success encourages other people to take up sports - improving quality of life and reducing healthcare expenditure long-term.
  • The Olympics in London will encourage tourism and that will generate more money for the economy than they will cost.


But whether consciously acknowledged by the funders or not, there's another reason why this success will bring economic benefits. I examine here what conditions must be satisfied to result in a positive return on the £265 million spent since 2004.

Coordination is important in economics. Adam Smith's pin factory, Coase's theory of the firm, Schelling's study of micromotives and macrobehaviour - all of these are powerful demonstrations of why people acting alone are usually beaten by those working together. Well-coordinated teams simply produce more output and more value than the same number of individuals working apart (accepting the Mythical Man-Month's caveat for certain types of work).

The prisoner's dilemma shows very convincingly that allowing coordination to supersede your natural instinct for individualism can bring you better returns as an individual.

Now, let me make two assertions:
  1. That there is a specific economic gain from coordination, whose value can be estimated
  2. That success in the Olympics encourages people to identify with the UK as a community, and improves their commitment to coordination over selfishness


Why is the second statement true? Because we seek to flatter ourselves. Seeing Chris Hoy wearing three gold medals, we subconsciously think "Part of that success is not just because he was born with big thighs and trained hard - it comes from his stoic grit, his stiff upper lip...his politeness in a queue. In other words, his underlying British character. I too have an underlying British character, therefore his medals prove that I'm great too."

In fact Chris Hoy is Scottish, so I'm even more great - as of course everyone knows our underlying Scottish character is more dominant than the British one.

Rationally, this may not bear much resemblance to reality - but when you compare the results of the UK team to the US, France or (especially) Russia - don't you want to feel that there's some basic, human reason that the UK has done well?

My postulate is that we translate this subconscious inference about British character into a greater sense of similarity and thus community with the people around us. That, I suggest, provides a small boost to cooperation and a small suppression of the individual instinct which can be so counterproductive to economic success.

(Bear with me here - I'm still a capitalist, this is about coordination not corporatism)

So if you accept that argument, how do we quantify it? The gains to coordination are truly vast - primarily from permitting specialisation which increases the leverage on our mental and physical resources; and from the sharing of knowledge which increases our capacity to produce. Specialisation on its own provided a hundredfold gain in productivity even in Adam Smith's time. Now, across an integrated economy and with modern communication technology, we probably get ten times more benefit again.

The gains from knowledge are hard to measure but what David Warsh calls new growth theory has tried. It would be fair to say that the larger share of economic growth in the last sixty years has been due to the dissemination of knowledge.

Putting these together, one could plausibly argue that over 95% of our current prosperity is due to coordination in one form or another. Of course, much of this coordination happens anyway - it is governed by firm existing structures such as contracts and the physical arrangement of the economy across the country. But a good quarter or more of it can be ascribed to soft structures - personal relationships, psychological commitment to a well-running society, people's subconscious models of their identity and that of the entities around them. Imagine these soft links become just 2% stronger, for the 50% of the population who is following the Olympics, or at least the headlines; let the effect last just a month; and make the reasonable assumption for simplicity that the economic results are a linear function at this level of change. What are the results?

Under these simple assumptions this works out to just under a 0.24% boost to the economy for a month. The size of the UK economy is about £1.4 trillion and so this boost is worth £280 million. The spend on elite sport these four years is £265 million. So, in pure, numerical terms of plain economic growth - ignoring any multiplier effect of the actual spend, the entertainment value of the results, and any improvement in national health stimulated by the spectacle - the investment has made a positive return in 2008 alone! And the nature of GDP growth is a ratchet - so these boosts are likely to be retained to some extent in future years, providing an ongoing return.

The £280 and £265 million look very close. Coincidence? No, of course not. Who made up the figures of 2%, 50% and one month, after all? How did I choose them? According to what I felt might be "about right". Actually I originally estimated just a 1% boost to coordination. But that way the return was only 50% in 2008 and I thought a 100% return would look better.

More seriously, this direction could offer a framework for how to measure the return. It shows one possible set of conditions which can be satisfied for the investment to be economically worthwhile - a 2% coordination boost among 50% of the population, lasting a month. If we believe - or experiment shows - these conditions to be met, there's a good argument to spend the money again for the 2012 Olympics - or for the 2010 World Cup - or on (ahem) the Millennium Dome.

Next time I had better show that Lottery spending on the arts provides an even bigger return - otherwise my friends who campaigned against the diversion of arts funding to the Olympics will probably excommunicate me.

Tuesday, 19 August 2008

Software maintenance: client value models

Dennis Howlett has a great post on his ZDNet blog: A fresh model for software maintenance.

He targets a number of things that vendors should do to reduce the cost and burden of software maintenance - treating customers individually, training their resellers and partners better, and helping the customers to build their own centres of excellence so they can support themselves better. The ideas are based in part on
C.K. Prahalad and M.S. Krishnan's book, The New Age of Innovation: Driving Cocreated Value Through Global Networks. Looks like something I should be reading soon.

This is an issue close to my heart, since it is a very sore point for many clients. Software buyers are much more likely to query the maintenance charges than the initial build costs - and yet maintenance seems to be an area where most small software companies don't make any money. Oracle seems to be an exception (Dennis quotes their 30% margins and says that much of it is from maintenance revenues), while SAP has just eliminated its low-cost maintenance models and is moving all its customers onto the top-priced Enterprise support package. Clearly a sensitive area.

Howlett's model seems to provide a good way to control costs by targetting vendors' skills where they bring most value - but we at Inon are starting to pitch our customers on something perhaps even more radical.

We want to charge our clients for support and maintenance according to the measurable value that we create for them. Our clients want three main things from our software:
  • increased revenue
  • increased productivity
  • and more accurate figures and reporting


As the relationship with each client evolves, we're in a better position to know how to achieve those goals and to suggest new things to try. So, why don't we charge them for delivering those exact results?

And after all, that's exactly what the maintenance contract is supposed to provide: maximum revenue by minimising downtime, keeping productivity high by avoiding workarounds and bugs, and making sure the outputs of the system are accurate by preventing errors.

And in this context, why distinguish between maintenance, support and new features? If I have a genuine partnership with my client, and I understand their business, I want to be able to weigh up a new feature idea (which will boost productivity by 8% for 20 users) against a bug fix (which will save 100 users 0.5 hours per week) and decide which one is in the client's interest for to implement first. At the moment I probably have to make the bug fix because it's in the maintenance agreement, and never get to add the new feature because it needs approval of new capital expenditure - or because the client doesn't agree with my estimate of the additional value that it will provide.

I would much rather my client pays me £20,000 for increasing their productivity by 8% than guarantees me £2,900 a quarter for support based on a percentage of their licence fee. This way I am motivated to deliver the best possible service and I will put my efforts in the places where they will bring the best return. What's more, I can make more money this way and it's in my client's interest for me to do so.

As Dennis points out, any model of this kind "requires that vendors have a much clearer understanding of their customers, their levels of competency [and] the value they can deliver". What's more, it requires both parties to agree on a good-faith model for measuring the returns - increased productivity is going to be hard to measure, and if it's not achieved, the vendor can try to blame the incompetence or change resistance of the client's employees. But if both parties trust each other, this simply puts the vendor in the position of having to work constructively with a trainer to get the best outcome for the client. A model that encourages effective working relationships like that is going to produce better results than one that doesn't.

But whatever the additional effort and intellectual difficulty of implementing this, the potential returns can be so spectacularly beyond those of the normal model that it must be worth it. At least for any vendor who can attract staff intelligent and committed enough to make it work. We believe that our revenue, and the money our clients make out of our services, can both more than double by implementing this model.

Saturday, 16 August 2008

A new generation of software

All the recent news in software has been about putting established categories of application software online. Online CRM software - salesforce.com. Online project management software - Basecamp. Online accounting software - Xero, Freshbooks, FreeAgent. Online ERP software - SAP Business One and Netsuite. Office productivity software - Google Apps. Email software - GMail. Even Photoshop has online competitors - Picnik (among others).

This will continue, and I think most observers agree that desktop and client-server software in these categories will gradually be overtaken by the online/SaaS alternatives - though Microsoft Office will probably hold out longer than most.

But when was the last time a genuinely new category of software became prominent? Think of the main categories of software:
- CRM
- ERP
- Office suites (which used to be several separate categories: word processors, spreadsheets, presentation software)
- Project management
- Drawing and graphics
- Email clients
- Web browsers
- DTP (remember them?)
- Games
- Antivirus software
- Databases
- CAD

Most of these categories have identifiable periods when they emerged as important, widely adopted applications. Usually they are associated with a particular company who introduced the category or became an early leader - Wordperfect, Netscape, Photoshop, Pagemaker, Eudora, Pong - often to be overtaken by the current market leaders, in several cases Microsoft (though not as many as you might assume).

But how often in the last ten years has a genuinely new and important category of software emerged? Web browsers were big in 1995 and the term CRM emerged around the same time. Some types of Internet software - blogging, peer-to-peer sharing, instant messaging or content management - probably have a claim to be new categories but all have been around for at least five years. Wikipedia has a useful list of categories at http://en.wikipedia.org/wiki/Software_genre - and again, none seem to be newly emergent in recent years. Perhaps social networking has a reasonable claim to be a new category, although it is probably associated more with new websites than new software applications. I would like to think CVM (client value management or customer value management) will make it as a new category, but that's yet to be seen.

Naturally it takes some time for new categories to be recognised as such - many people will tell you they have invented a new type of software, in line with the recommendations of numerous marketing gurus - create your own category and be the market leader. But surely five years is enough time for new types of application to become recognised?

Maybe everyone has become too good at marketing since the mid-90s, and everyone has learned that they are supposed to invent a new category name for the applications they produce. And a category won't really be accepted and referenced until there are three or four vendors all selling applications under the same name. Maybe we will never see a new category again. Maybe all the billions to be gained from creating new categories and getting consumers and companies to buy into them will be foregone by game-theory-optimising companies which are not willing to invest their marketing money into validating a market segment that someone else owns. Maybe existing vendors will never spot a newly successful entrant who has invented a category for their product and try to outspend them to grab some of their glory and profits. Maybe.

Yeah, right. At least we can rely on Microsoft for something.

Thursday, 14 August 2008

Structured Pricing

Another area of theory I have been working on in the last few weeks is pricing. It's well-known that suppliers can use price discrimination to distinguish between those customers who can or will pay more, and those who can't. Here are some (edited) notes from a forthcoming report we are publishing about the concept of structured pricing.

Economic theory relies on agents trading products and services in a way that makes all of them better off. Any two people entering into a transaction should both gain from it; whatever is traded must be worth more to the buyer than the money they pay; and the money must be worth more to the seller than the object they’re selling. Any trade therefore provides higher value to both parties than they currently have – or else it will not happen. In classical economics, the price of something is regarded as a signal to tell the buyer and seller how much other people (at the margin where the supply and demand curve intersect) value it, and therefore whether they should buy or sell it, or manufacture more or less of it. However this signal is quite simplistic, and especially difficult to use in cases where two services or products are hard to compare and are not tradable as commodities.

There are more sophisticated economic models which involve the psychology of the buyer and seller, their mental model of the world, and their “utility function” which is a way for them to evaluate different states of the world and determine how happy they are with each. These models allow for more complex interactions than just buying and selling a commodity object at a fixed price. In these models it is possible for people to consider, according to their own subjective preferences, the effects of a complex contract or transaction, and decide whether to enter into it.
Structured pricing, while consistent with the basic laws of classical economics, takes advantage of this new cognitive economics too. Its goal is to set prices in line with the subjective value defined by the client, and pass back value through the supply chain so that participants at each stage are paid by the value of what they produce. This influences decisions by employees or suppliers about how to design the solutions they provide – orienting them around client value instead of around either maximising (in the case of hourly pricing) or minimising (in the case of fixed price projects) supplier cost. Cognitive economics also allows for a richer analysis of what is valuable to people and companies – money is not the only measure of return, particularly for owner-managed businesses where the outcomes of the business are more directly aligned with the benefits for the individuals involved.

The standard microeconomic model for the sale of services is the same as for goods: the supplier offers a fixed service, publishing the service and its price in the marketplace; the client evaluates whether it will give them more value than the money it will cost to buy it; if so, a transaction takes place. This model assumes (among other things) that the service is predictable, and that the client can gauge what value they’ll get in advance.

Neither of these are really true of knowledge-based services. Almost by definition, you don’t know what you’ll get in advance – otherwise you probably wouldn’t need the service. If a lawyer is drawing up a contract for you, or an accountant preparing your tax return, or a software developer creating a bespoke database, knowing what the end result will be is equivalent to having the end result already. What’s more, the buyer rarely knows exactly what it will be worth to them until they get it.

So there is a range of different values that can be created by a project. Certain types of project are notorious for this – bespoke software, litigation and some building or civil engineering projects. Usually not only the value, but also the cost of producing the service is unpredictable. A whole industry of estimation has grown up to mitigate this problem; but the estimator’s role is based on making the project as predictable as possible so that they can come up with a fixed number.

Thus, in order for a transaction to take place, the buyer needs to either be willing to absorb a lot of risk, or the agreed price has to be at the lower end of the range of values. If the project may generate anywhere from £1 million to £3 million in returns, the client will probably want to pay less than £1 million for it. If it works out well for them, they end up with a £2 million bonus which the supplier can’t share in. If it doesn’t, they still get their £1m of value.

In this context, any project which costs £1.5m to deliver, and returns a range of possible returns from £1m to £5m will never be made. Even though it is likely, on balance, to generate value overall. So companies which trade on a fixed-price basis are already losing opportunities for economic value.

(Sometimes this problem can be mitigated by actions such as prototyping, which may help to get a better estimate of the expected end value. But this can only work in limited circumstances, and often tends to slow down the economic process - delaying the value and therefore reducing the total discounted value that is created.)

More importantly, the supplier has lost its incentive to deliver a great result. Because knowledge-based services are unpredictable, the specific choices and performance of the supplier are likely to have a big influence on the outcome and value generated by a project. If they are paid a fixed price, they are incentivised to deliver as little as possible which still meets the basic threshold of the project spec. If they are paid hourly, they are incentivised to deliver as many hours as possible with no regard to their effectiveness. Only if they are paid by the value delivered, are they incentivised to actually deliver what the client wants.

Rational, game-theory-optimising clients may prefer the benefit of a tight negotiation and a reduced price over the cost of writing and negotiating a detailed spec, and losing the supplier’s creativity – as long as they only intend the relationship to live for a single transaction. But if the parties intend to work together beyond the conclusion of a transaction, the incentives of the supplier become more important. The reason is that their behaviour beyond the end of the project is, by definition, not constrained by the specification; the moment when the supplier becomes free is the moment at which their incentives become important again.

At this point, if the supplier is incentivised to deliver value, their knowledge about the client’s business and their understanding of their own field can be combined to deliver new ideas about how to create value for the client. In traditional economic activities, perhaps this did not matter – because everyone (supposedly) knew how to create value, and the constraints on it were only related to available capital or resource, and the skills of the available labour. In intellectual businesses, the constraint is on the knowledge of how to add value, and the supplier is as likely as the client to see how to do this.

Our economic models demonstrate that (under plausible assumptions which are detailed in the appendix) in a typical supplier-client relationship, over the course of a year, structured pricing can deliver substantially more value to the client and to the supplier than conventional hourly rates or fixed project-based quotes. We will publish soon our estimates of the scale of the extra value created.

One of the goals of Inon's Client Value Management software is to help you to implement structured pricing. Although CVM can be used successfully with hourly or fixed price quotes, there is a whole new dimension of profit to be made by entering into a structured pricing relationship with your clients.

Saturday, 9 August 2008

Is CVM the new CRM?

A fair bit of the economic research I do is speculative and it's gratifying when it ends up in a useful commercial output. This is one good example.

The work I've done on how people evaluate prospective utility and make decisions has led to the concept of structured value modelling. This in turn allows us to consider how people influence each other's models of value. And the outcome of that is that we have created a new category of software: Client Value Management or CVM software.

Traditional CRM systems are good for high-volume marketing - especially to consumers. They are not very popular amongst business-to-business services providers - for instance professional services firms. The reason being that CRM is a reductionist tool - its concept is to allow the simultaneous management of large numbers of people by making simplifying assumptions. If we assume that people fall into one of four demographic groups then we can send them messages at a cost of 4p each and hit a million people with a £40,000 campaign.

However the consequence is that the messages are extremely untargetted and therefore much less effective - because they are not adjusted to the specific value model of each recipient. If you sell consumer goods costing a few pounds each, you may have to live with this - because you simply can't afford to spend more on sales than the price of the product.

But if you sell business-to-business; if your market size is small; if your service costs enough; and especially if the service is tailored uniquely to each client; then CRM is not going to sell it for you. You need to customise your pitch to each client and make sure that what you offer is a match for what that individual client finds valuable.

The way people usually do this is through networking and individual conversations; sales meetings and customised proposals; and indeed by iteratively refining the service they offer, even after being commissioned to provide it. All effective processes, but they take a lot of work.

By creating tools to model the psychology of value, and to understand and measure the relationship between what person A does, and what person B receives, we can support this process, make it more effective and amplify it by orders of magnitude. These tools can help sales teams and marketers to understand why their clients buy from them, and thus make it happen more often. So we've spent time building this software and it's now one of our key products for 2008 and beyond.

Client Value Management is, in short, the new CRM - for professional services firms and other business-to-business service providers.