The Yahoo! Post-Bartz post and the perils of Microsoft Excel

I started thinking about this post when I was reading Bob Cringely’s excellent analysis of Bartz tenure at Yahoo!. I am not going to add my full analysis here but instead pull on a strand that highlights problems that exist at a number of internet companies and certainly existed at Yahoo! when I was there.
Yahoo! star
Part of the thought process that got me on the trail of this post was that it reminded me of the introduction to Cringely’s Accidental Empires book written in the 1991:

… PCs killed the office typewriter, made most secretaries obsolete, and made it possible for a 27-year-old M.B.A. with a PC, a spreadsheet program, and three pieces of questionable data to talk his bosses into looting the company pension plan and doing a leveraged buy-out.

Spreadsheets and the business models inside them can be extremely powerful business tools and also weapons of mass destruction.

Powerful Business Tools

Firstly about the power of spreadsheets and their models in an internet business. Whilst at Yahoo!, my former colleague Salim used to be able to take the first few months traffic figures for the search business and provide a pretty accurate forecast of what the rest of the year looked like. That could be further extrapolated into reasonable revenue projections based on average conversion rates and cost-per-click values. Pretty handy for a business that relied on the fickle general public.

Weapons of Mass Destruction

Efficiency and innovation

Accounting models are often used to make cuts in terms of manpower. What they fail to do however is ensure that the cuts are sufficiently surgical. This is less of an issue in a conventional manufacturing setting where there is likely to be a degree of redundancy in skills due to process design. Business management theory and analytical tools came out of this industrial age. Software and web services follow much more of an artisan model – great coders like mathematicians can find elegant solutions to problems through intuitive leaps forward.

Although there is a large amount of outsourcing to cheaper countries, many successful breakout products or features are developed by small teams or even individuals for example:

  • Andy Hertzfield – the MacOS QuickDraw 2D graphics library that has been used for over 25 years and is only now being phased out in the latest versions of OSX
  • Cal Henderson and Stewart Butterfield – Flickr and Glitch
  • Joshua Schachter – Delicious
  • Linus Torvalds – Linux kernel

However spreadsheet models often don’t recognise who these rock-stars are.

What this means is that in a time of cuts the very people who could drive the innovation that would fuel future growth are let go or choose to leave because their area has been hacked to pieces. A classic example of this under Carol Bartz was the Flickr team: George Oates was let go and others like Paul Hammond, Seth Fitzsimmons, and Matthew Rothenberg departed.

Carol Bartz quite rightly once said that ‘you can’t cut yourself to growth‘, but you can’t outsource it in the longer term, you also need the tinkerers and the thinkers in the organisation creating the innovation seed corn to drive that future growth. There doesn’t seem to be a spreadsheet model that takes adequate account of this.

Niches versus the mainstream

Back when I worked at Yahoo! there was an inordinate amount of attention paid to the number of unique users that properties got. This is important for a service like search that is universal in its appeal, but a general purpose metric like unique users falls down flat for many other properties that have a specific context around it.

Let’s look at three examples:

  • In the West, Yahoo! Answers has a substantial user base of unique users, but a quick look at Google Adplanner shows that this user base is skewed to lower socioeconomic groups who are time-rich, but cash poor. This means that it could be hard to sell advertising inventory to many brands and the corresponding cost of inventory is likely to be cheaper
  • Yahoo! image service Flickr, has far less pictures than Facebook; but it is a highly engaged community of people passionate about photography and the creative classes. Facebook is like the digital equivalent of Prontaprint – who used to publish their film envelopes in local newspapers and develop the general public’s holiday snaps. This means you could charge more for the service, which is why Flickr has a freemium offering and come up with creative marketing packages for advertisers
  • Social bookmarking pioneer Delicious was a slow growing property beloved of geeks and the creative classes. Attractive both because  of its audience’s demographics but also the level of insight available from the data that these people provide voluntarily. A creative marketing vehicle similar in nature to Twitter’s promoted tweets has the potential to be a premium-priced product for advertisers

However using spreadsheet models with metrics that lack distinction Yahoo! Answers looks like a great product whilst Delicious and Flickr look marginal at best. It is no coincidence that Flickr had an outflow of talent under Carol Bartz and Delicious was sold after a protracted period of uncertainty about the service’s future.

Ultimately tools that can create a flawed understanding can be more damaging than no tools at all. Carol Bartz was brought into cut a business that she didn’t understand that well (it wasn’t anything like her previous roles) and had analytical tools at her disposal that weren’t sufficiently finessed for a modern information economy-based company. You add this to Bartz dogged personality and you can see at least part of the reason by she was not able to turn the company around.

More links
How not to run Yahoo! – I, Cringely
Yahoo! Announces Leadership Reorganization – Yahoo!’s official statement