Tuesday, July 15, 2003

La Morte de Software?

One relatively hot question kicked off by Larry Ellison at the beginning of the year concerns the maturity of the software industry. Has the software industry matured? First, what do we mean by matured? Typically this refers to a decline in the revenue growth of industry players leading to flat growth and eventual decline in industry revenues. Jeffray Kay in his recent post seems to think that the software industry is maturing. He cites three reasons: decreasing marginal returns on software investments by businesses, decreasing innovation (particularly in web technology and enterprise application integration), and finally consumers are just finally starting to adopt some software that enterprises have been using for years. I am not sure I completely understand this last point because it seems to only indicate that customers will continue to lag the enterprise--but this is certainly not a new trend and it is unclear how this relates to the maturation of the software industry. The first two points seem quite strong, however. I would even add that Microsoft's recent accounting changes and business model changes seem to indicate that it believes its own industry is maturing. Microsoft has recently announced dividend payments for stockholders and just last week announced that it would cease giving taditional stock options in favor of giving straight stock grants. Both of these changes indicate that MSFT does not predict the same degree of growth that it has in the past. Providing dividends essentially implies that MSFT does not believe that it will create more shareholder value by giving stockholders money now rather than investing that money. Essentially, this reflects a belief that the attractiveness of MSFT's investment opportunities has decreased. Stock options are well-suited to high-growth businesses because they allow the business to provide employees with a share in the long-term potential success of the company while at the same time "costing" the business very little in the near term. The shift to a "fair value method" of expensing stock grants ssentially amortizes the estimated economic value of a stock option over the life of the option (using an option pricing tool like Black-Scholes) and expenses the expired portion of the option value. This shift seems to indicate to me that Microsoft thinks that it should "take the hit" of stock grants now rather than deferring this economic impact. Why would MSFT want to do that? Part of it must involve a belief that stock options previously granted to employees that are "under water" are likely to remain under water for too long. Part of it might also involve a belief in improving accounting transparency. However, it is generally mature firms that care more about accounting transparency because they can afford to.

Despite the arguments cited by Jeffrey Kay and the MSFT evidence that I provided, I am not convinced that the industry is maturing. Anecdotal evidence seems to indicate that there are still bright spots in the industry--open source (driving volume but not necessarily revenue / profits), instant messaging software, business intelligence software, RFID tracking software, and some security software--intrusion prevention, authentication, etc. The appliance market is also experiencing rapid growth. I have to admit that none of this anecdotal evidence really contradicts the rationale of Kay's arguments--but it does provide some indication of ongoing growth opportunities and innovation.

Arguing against Kay's specific points:

1) Decreasing marginal returns--while it may be true that traditional enterprise software sales are slowing this is a necessary but not sufficient argument to prove that the industry is mature. A sufficient argument would require proving that there will not be new kinds of software in the future that will sell as well as traditional enterprise software, in other words that innovation is dead. This brings us to Kay's second point.
2) Decreasing innovation--again Kay seems to point to decreasing innovation in traditional enterprise software and web technology (and to a lesser extent security software) but he does not point to new types of software such as the types I mentioned above. Kay has to make a stronger argument that he believes that software in general has automated a large enough set of "human problems" (for lack of a better word) that future innovation will provide only marginal improvement on what has already been developed. Personally, I think that there are probably as many arguments that software has just scratched the surface of its capability to tackle "human problems" as there are against it.

Wednesday, July 09, 2003

Early Adoption Framework v2.0: Addendum

There is another way to identify industries and companies that are likely to be early adopters of a new technology. Let's call this method the "substitute" method. The method involves simply identifying what current product, process, or skill the technology can replace. In the case of RFID, RFID tags are very similar to bar codes and hence substitute very well for them. RFID tags have certain features that make them superior to bar codes and certain features that make them inferior. Most importantly, RFID tags can be scanned from further away than can bar codes. RFID tag techology allows for more unique IDs more easily than do bar codes--RFID technology yields about 18 thousand trillion possible values. RFID chips can be smaller than bar codes. On the inferior side, RFID technology is raising privacy concerns (see this CNET article).

By simply looking at substitutes, it is often easy to find the industries most likely to be impacted by a new technology. For example, it becomes quite obvious that RFID will substitute for UPC bar codes and scanners. Given this information, it is clear that grocery stores, retail stores, and manufacturing environments are likely early adopter candidates. However, this framework does not make it clear that embedding RFID on cars and installing readers at toll booths might be a clear opportunity see the MIT auto-id center , and other possible uses. This is a good first-order look at the impact of a new technology but not a complete framework.
Early Adoption Framework: A First Stab Using the Example of RFID

So, true to my word I have been thinking about a framework that could be used to help us identify where early adoption is likely to take place. Here is the basic framework--but remember this is v 1.0 and I hope to refine this over time. Let first make the assumption (and I realize this is quite a large assumption) that adoption occurs in those industries where the relevant technology stands to most significantly increase free cash flow (and let's assume free cash flow = profits for simplicity) at a reasonable cost. In other words, technology adoption occurs where the benefit / cost ratio of adopting that technology is greatest. So there are two sides to understanding adoption: the benefits to a company of the technology and the costs to a company of the adoption. We can think of the costs of a technology in terms of the direct implementation costs--the costs of setting up and using the technology as well as any indirect, external, or system-wide costs--e.g., the costs to the well-being of society when a new high-tech factory pollutes. These latter costs are often terribly important because although they may or may not be directly bourne by the user of the technology they are bourne by some group and this group (or the government on behalf of the group) will often try to get the user to assume part of the cost. We can simplify technology costs into three broad categories: acquisition costs, deployment costs, and maintenance costs. In general, a technology will have a higher likelihood of adoption if it minimizes these three costs. Now lets turn to the benefits side. A simple benefits framework would seek to understand how technologies could impact companies' profitability and identify those companies most likely to be significantly impacted. Here is the framework I was thinking of. Total profits = total revenue - total business costs. It is clear that technologies can impact profits by increasing revenues, decreasing business costs, or both. People often, and in my opinion, justifiably assert that technologies that are likely to succeed address "pain points." In this framework, we can simply tranlate "pain points" into language about revenue improvement or cost reduction opportunities. Creating subcategories under the revenue and cost items in the framework is probably not particularly fruitful since the categories will be highly dependent on the type of technology we are analyzing. Let me take the example of RFID, since I am quite interested in its potential.

On the techology cost side, RFID are becoming increasingly inexpensive to acquire, deploy and maintain--though currently RFID chips are quite expensive. According to a recent CNET article, Alien Technologies is likely selling chips for about 25 cents / each. The company also predicts that large quantities could be sold for as little as 5-10 cents. A recent MIT study that I looked at said that within 10 years or so, we could see RFID chips approach 1 cent. Given the prevalence of substitute products, I would guess that the demand for RFID is highly elastic. Hence, it would seem that RFID will grow extremely rapidly and in line with its price falling from 25 cents. RFID readers are also quite inexpensive, one company on the net cited that they would sell basic readers for under $150. I would not expect deployment costs to be huge although it may require building and installing relatively complex wireless networks, management software, etc.

Since we want to use this framework to understand what industries are likely to adopt RFID, we should ask will costs be lower in some industries before others. It is unlikely that RFID producers will n price discriminate across industries; however, it is common practice to offer volume discounts. So industries that have companies that could buy large amounts of RFID--i.e. companies that have a lot of products that need to be tracked--will likely be more interested in adopting quickly.

The other part of the technology cost is the "external" or "system" cost. RFID do pose some system-wide costs, the most obvious and important of which is the threat they pose to individual privacy. There is a now rather famous episode in the RFID world where Benetton, the famous clothes manufacturer, announced plans to embed RFID in a small line of clothing. Consumers protested. Benetton was rumored to have stopped the experiment, though it now appears that they are moving forward, albeit with much more caution. This particular cost will affect businesses and governments that have sensitive information and consumers who are just simply afraid of privacy violation.

Now we turn to benefits. How do RFID tags increase business revenue or decrease business costs? It is quite clear that RFID can decrease costs by managing inventory levels more effectively decreasing storage and holding costs and decreasing shrinkage at all points in the supply chain. RFID can also increase revenues by helping companies to avoid stock-outs, get the right product to the right place at the right time. RFID can increase revenue by allowing companies to get better data on their customers purchase and usage behavior and hence allow them to tailor their products and services to customer segments more effectively. Which industries will reap these benefits more than others? We need to look closely at the potential revenue and cost items in each industry and use this as a guide. Since I am running out of time, I will leave this as my next exercise--identifying the industries that should be most affected by RFID. Some of these are quite obvious since there is already broad publicity regarding the impact of RFID on retail and manufacturing. I will delve perhaps a bit deeper to try to uncover additional industries that might be impacted.

So far the earliest uses appear to be embedding RFID in retail locations for supply chain management and shrinkage prevention, clothing, and access control devices--see RFID Journal.

Tuesday, July 08, 2003

Back to RFID: Uses and Thoughts

I want to spur on the discussion of RFID. What are its primary uses? Perhaps this is too broad of a question. Where are we likely to see early adoption of RFID? What kind of a framework can we create that would allow us to guess where early adoption is likely to take place. I will be thinking about this question over the next little whil and I will get back to you.
iTunes Music Store: On the Side of Unsustainability

Several great posts are showing in blog-land on this topic. I really like the ones here and here.

First let me lay out a few categories of sustainability--for the purposes of this discussion I will focus on revenue growth as the primary metric of sustainability. Category 1: the company, organization or business continues to remain in business but loses revenue and / or market share (perhaps asymptotically approaching 0% market share or some unimpressive share of the market. Category 2: the business approximately maintains share--and revenues do not decline. Category 3: the business grows market share moderately and revenues do not decline. Category 4: the business grows market share significantly and revenues do not decline. After the initial hype wears off, I would predict that iTunes will be primarily a Category 2 business with some non-trivial chance of becoming a Category 1 business. However, in the interim, i.e., until the other online music distributors change their pricing schemes, Apple will enjoy a healthy and profitable niche. This fact can be relatively simply demonstrated by some quick calculations. Apple charges 99 cents per song, Rhapsody charges 79 cents / song and $9.95 / month, Pressplay charges $9.95 (with no burning) and $17.95 /month (to burn 10 songs / month), MusicNow charges 99 cents / song and $9.95 / month and MusicMatch charges $4.95 / month. Apple also allows unlimited burning/listening/etc., Rhapsody allows unlimited burning/listening etc., Pressplay allows no burning for $9.95 and 10 songs / month for $17.95 and MusicMatch offers no burning. Given these conditions, it is a relatively simple math exercise to show that you would have to burn over 49.75 songs / month on Rhapsody to make it a better deal for you than Apple. You would have to buy over 10.05 songs / month in Pressplays $9.95 plan and not care about burning to make it a better deal than Apple. You would have to buy over 18.13 songs / month in Pressplay's $17.95 plan to make it a better deal than Apple. You would have to buy over 5 songs / month and not care about burning to make MusicMatch a better deal than Apple. Given that a large portion of the downloading population (who own Apples) will not perform these statistics and is likely to think that 99 cents is categorically cheaper than any service that prices based on a fixed / subscription cost + a variable / per song cost, Apple should get a lot of business. Further, even if you do perform the calculations, few people are likely to download 50 songs / month regularly. However, it may be wise to sign up for a couple of months on Rhapsody and download as many songs as you can and then switch to another provider. Also, Apple users, as Thom wisely mentions below, are likely to find the interoperability of iTunes with their other applications somewhat appealing. All of this basically means that Apple has priced its way into an attractive part of the market.

However, pricing is not a sustainable strategy. I think before the recognition of game-theoretic sciences, many people thought that pricing was a sustainable strategy. Indeed, in all fairness, it can be a sustainable strategy if for some reason you can price lower than the costs of your competitors for medium-to-long periods of time. State in another way, if you can be the classic "low-cost" provider of the good. However, given that Apple has very little control over the cost of the music that it acquires and it will never be able to acquire music cheaper than some of its other competitors--pressplay is owned by Sony and Universal and will always be able to acquire music at least as cheaply as Apple, it will not be able to sustain a low-cost / low price strategy. Hence, Apple's entire strategy depends crucially on other players not lowering their prices. Will this happen? What might cause Apple's competitors to want to maintain their current pricing models. Economically speaking they should only want to maintain their current pricing if they are somehow making more money than they would under Apple's model. For all of the players except perhaps pressplay and MusicNow, it is likely that they would make much more money by adopting Apple's model. Rhapsody clearly is trying to target the heavy users with its high fixed cost but low variable cost. However, these are not simply heavy users--these are ethical heavy users because most heavy users get their music from KaZaa (or however it is spelled), etc. I would have to imagine that by giving volume discounts they could continue to attract these ethical heavy users while shifting to a low per song fee that would still attract lighter users. MusicMatch is a low cost alternative with its low monthly fee but it does not capitalize on the fact that there is something incredibly attractive about not having to enter into a subscription service. Additionally, MusicMatch does not offer the selection that many of its major rivals do. It is possible, but I would still consider it unlikely that pressplay and MusicNow might maintain their pricing schemes. The reason that it may be possible is because these two companies have strong ties to the Big 5 music labels or are owned by members of the Big 5. Since the Big 5 want to maximize their profits they might consider doing so by selling music to apple for low-willingness-to-pay customers and selling via pressplay, etc., for higher-willingness-to-pay customers.

Ultimately, prices generally tend to converge toward an optimal price to maximize revenue. Targeting distinct customer segments with price differences alone is not sustainable because, if the segment is attractive, competitors can easily copy it or tailor the value proposition and pricing of their products to conform to the segment.

I disagree that Apple's DRM is a key issue. Everyone is pushing toward the ideal DRM solution and once the cat is out of the bag the record labels will probably have to give similar deals to everybody who resells their music. It is unlikely that they will allow iTunes users to burn as many as they want for $X but not allow Rhapsody users a similar deal. In fact, a recent Forrester study showed that people who have Apples at home generally have higher incomes than those who have PCs. So, if the record labels were trying to do some nifty form of price discrimination they should price their music higher on iTunes than on other services.

I agree with Thom that the subscription free service is great. But everybody can do it.

I agree with Thom that integration is good (though I am not sure how much it really helps) and if it truly is valuable to the customer (which I somewhat doubt) it could be sustainable. However, iMovie must allow you to integrate music from Rhapsody that is downloaded to your harddrive nearly as easily as it allows you to integrate music downloaded from iTunes so I don't see this argument. Similar arguments hold for iPod, iDVD, etc.

It is possible that Apple will grow its share or even maintain a healthy share of the market, but given that selling music, and negotiating with artists are not Apple's true strengths, it will likely face a strong competitive response, and Apple users are a small (roughly 3%) segment of the installed base, iTunes will likely lose share in the long run.

This page is powered by Blogger. Isn't yours?