I agree with pretty much all that you are saying, but you are only looking at one type of startup. The "we think 80% of people will pay for our product".
I work for a company that makes up a different segment., "we know that 80% of companies will buy our product" bevause it is niche based and very useful.
Thanks for the advice, we just launched our startup (shameless plug -www.LoudIsRelative)
and couldn't decide the right time to start placing ads...
All good points, but what bothers me about the "controversy" around Paul Graham's comment is that it's really putting the emphasis on the wrong thing.
What matters is making a product that someone is willing to pay for (customers, advertisers..). Figuring how they will pay for it is really a much easier problem. That's how I understand Paul Graham's position and it's not the same as saying revenue models don't matter.
I felt compelled to comment, because I couldn't agree more with this statement:
"Like a software product and it’s features, I would argue that a business model and its execution also needs to be “experimented” with and adjusted."
For MyPunchbowl, we've spent a lot of time working with potential customers, showing off the product, and adjusting our thinking. One critical point you missed (in my opinion)....whether you implement your business model in the early going is very different from whether you are actually thinking about it and strategizing. For us, it makes sense to wait to implement until we have worked out all of the product issues that would prevent us from actually experimenting with our business model. LIke UI comps, we're testing by showing-- and getting good data that will help us hone our eventual business model.
Great article (again).
- Matt http://www.mypunchbowl.com
I agree with cedric's comment, having customers willing to purchase your product is the ultimate litmus test to your long term viability.
However, the significant marketshare that can be achieved when a product is made available for free is not to be ignored. This is the main goal of shareware, and it worked for Doom and Winzip.
Of course it wasn't the shareware model that was the sole reason for the success of these products - one of them addressed the problem of limited capacity of storage media, and the other - became a legend. :)
So no matter how you promote your product it must fill a need/solve a problem that customers will be willing to pay for.
I agree with the need to have a business model at the back of your head from day 1. I am sure though that in many circumstances applying it from day 1, is not a good idea.
Many startups follow the "grow organically" approach. When you release early, especially an innovative service, you have a lot of adjustments to make. Your users are your consultants giving you feedback and enabling this process. Helping you to focus on the right things and get rid of the wrong ones. Evenmore they are your salesforce, helping to spread the word of your new service. Would it make sense to charge them at this stage? They are probably offering you a greater service than you to them!
I agree with the charge early, charge often vs. the big bang approach.
For a software company, having one key, referenceable, paying client in a vertical you are targeting is critically important. It decreases your need for funding, establishes a beachhead to target other firms in the vertical and as Dharmesh indicates, you build a better product.
The Critical mass issue is important for media companies as large advertisers and agenices need a certain amount of volume in order to make it worth their time. Establishing relationships, analyzing users, performance and billing issues cost time and money. If your site can't deliver enough volume to large advertisers than focus on smaller, niche advertisers until your volume grows.
Alternatively, enter into an agreement with an ad network like Tribal Fusion for banners or Google Adsense or Quigo for contextual CPC ads. You gain the benefit of their sales resources and technology in exchange for a revenue share.
Young media firms need to fully understand their user metrics and ad performance in order to adjust their sites for optimal revenue. Delaying this will just cost more money and time later.
Chances are you already spent the money on Quickbooks-- you might as well use it for revenue and expenses!