Sunday, April 27, 2008

Four Circles of Product Management Hell

I typically work on software for hardware platforms, where the design cycle is considerably longer than that of a desktop or web application. 18 months for a new platform is not uncommon, and it is important to nail down the main feature set early enough in the process for the hardware capabilities to be designed in.

Nailing down these requirements is the job of the product manager, and at this point in my career I've worked closely with about a dozen of them. They generally approach the requirements definitions using a similar set of techniques:

  • visit current customers to find out what they like and dislike about the product, and what they need the product to do
  • visit prospective customers to get first impressions and feedback
  • work with the sales teams to break down barriers to a big sale (which can mean one-off feature development if the deal is big enough)
  • keep track of what the competition is doing
  • generally stay abreast of where the market is going

Over time my opinion of what constitutes "good product management" have evolved considerably. I'll give a few examples that portray what the younger, naive me thought of a particular style of product management versus what the older, cantankerous me thinks.

1. All requirements generated directly from customer requests

What I used to think:

"Wow, this is great. These are all features which obviously appeal to customers. We won't waste time on stuff that nobody will ever use, everything we develop will help sell the product."

What I now realize:

This is "Building Yesterdays Technology, Tomorrow!" Customers will ask for improvements and extensions to the existing functionality of your product, and that is fine. When a customer asks for something completely new, it is generally because they can already buy it from a competitor. In fact if the input is via an RFQ they are probably trying to steer the deal to that particular vendor by putting in a set of criteria which only that one vendor can meet. The competitor likely helped the customer put the RFQ together, as any good salescritter can supply a list of specifications which only their product can meet.

Certainly, there will always be a healthy amount of catch-up requirements in any ongoing product development. Competitors are not sitting idle, and they're probably not incompetent, so they will come up with things which your product will need to match. Any product manager will end up listing requirements which simply match what competitors already have.

However there must also be a significant amount of forward-looking development as well, building the product to meet anticipated features which the customer does not yet realize they need. Constantly playing catchup only works for companies in truly commanding (i.e. monopolistic) market positions.

2. Requirements list both externally visible behavior and internal implementation details

What I used to think:

"Umm, ok. Its a little weird that the product requirements are so detailed."

What I now realize:

The product manager believes themself to be a systems architect trapped in a product managers body. They almost certainly used to be an engineering individual contributor (many product managers come from a hardware or software engineering background). They likely feel they have such a firm grasp of the tradeoffs in developing the product that they can easily guide the engineers in how to proceed.

Also, and I am as guilty of this as anyone: engineers often have little respect for the product manager... and the reverse is true as well. The product manager may be trying to provide a top-level design for the product because they consider the engineering team to be incapable of doing it correctly.

3. Anticipates market needs with uncanny and unfailing accuracy, down to the smallest detail

What I used to think:

"Wow, this Product Manager really knows their stuff. This is great!"

What I now realize:

The company is doomed, because we're in a market which is too easy. We'll have five hundred competitors, who will all build exactly the same product and end up racing to the bottom of the price curve so nobody will make any money. Polish the resume.

The Gentle Reader might ask, "What about the product manager who can analyze a difficult and complex market to generate the perfect product requirements with 100% accuracy?" The answer is quite simple really: such a person does not exist. We developers very much want the paragon of product management to exist, someone who can predict the unpredictable and know the unknowable, in order to satisfy some deep-seated need in our world view. However there is no such thing as omniscience, the best product manager is one who can generate a market analysis which is good enough to get some initial traction and be refined in subsequent updates.

4. Product manager is often not in the office, and only revs the requirements document sporadically

What I used to think:


What I now realize:

Actually, these are some of the signs of a really good product manager. When kicking off a new product they will spend a lot of time visiting customers and potential customers to gather inputs.

They apply a certain amount of skepticism, filtering and summarization to the raw inputs coming from outside the development environment, but without introducing too much delay. A product manager who passes on new inputs every time they meet with a customer or return from a trade show is not a good product manager, they're just a note taker. A product manager who suffers from analysis paralysis, not wanting to make a call until the correct course of action is absolutely clear, is also not a good product manager. There is a finite window of market opportunity, miss it and you'll find out from your competitors what the right thing to do was.

The best product managers know not to change the requirements too often, or the development team will never finish anything. So they will not be constantly releasing new versions of the requirements spec, and any changes which are made will have gotten enough validation to be credible without taking so long in validation as to be stale.

Unfortunately it is also possible that the reason they are not in the office and don't produce much is because they really are a slacker. Caveat emptor.


In fairness, I should acknowledge that the product manager job entails more than just defining the product. As they fit in a role equidistant between engineering, marketing and sales, the product manager is frequently pulled in to handle crises in any of the three areas.

  • as an expert user of the product, they do a lot of sales demos
  • they frequently end up training resellers (and customers) on the use of the product
  • they invariably get sent to all of the boring trade shows
  • they assist in preparing white papers, sales guides for particular vertical markets, etc

Finally, for a perspective from the other side of the fence I highly recommend The Cranky Product Manager. Entertaining, with a dash of truthiness.