Thursday, May 24, 2012

Tests for the business model

Startup teams should think hard about which assumptions in the business model are the most critical and the most risky. What most needs to be true for this to work? Test those assumptions first. I can't tell you exactly how to experiment on and test each and every aspect and not all business models contain each part of the business model. So you have to use some creative thinking on your own. Below are some guidelines and suggestions to get you started thinking. In general, you should try to be as quantitative and objective as you can in deciding what worked and what didn't. You're not trying to run tests for the sake of running tests or to generate glory metrics (we had 100,000 users - ok, well is that 100,000 daily users, email signups, hits to the website?). Rather the point is to actually think critically and learn something about how best to organize your business - before your competitors learn it faster!

It is up to you to prioritize your testing process. One part of your business model may be highly uncertain. You might want to test this first, particularly if it's a very important aspect of the model and if you can find a way to test it relatively cheaply and quickly. You want to turn your assumptions into data points and you want to generate the most valuable pieces of data first, before you spend a lot of time and money. However, you may have debates within your team about which tests and data points you should focus on generating first. There's no way that I can tell you ahead of time for all the different types of businesses out there where to start precisely.


As I mentioned on the videos, marketing starts with a thorough understanding of your customers and market segment as well as the type of market (new, growing, mature). First brainstorm several potential ways to reach your target customers. Ideally you can involve potential customers in your brainstorming. Then prioritize these marketing strategies into the cheapest and easiest ones to test out first.

You want to carefully track these marketing campaigns so that you can later calculate your cost of user acquisition. How much did you have to spend on each marketing campaign to get a single customer to buy? Or if they can't buy yet, how much did you have to spend to get them to leave an email address? Then you can start to track which type of marketing strategies are the most cost effective? You might try a campaign on facebook, hold a raffle, buy some google adwords, etc.


If you have a consumer web product or if you have a physical product or are selling to other businesses your sales process may look very different. Nonetheless, there will be some type of sales funnel or process. If you're selling to a business, who are the key decision makers? What does their process look like for approving a sale? How long does it take? Can you brainstorm several possible sales strategies and order them from cheapest and easiest to the most expensive. Start with the cheap or free ones (contacting your network). Can you try to make a few initial sales? Which strategies or tactics seem to work best? Where to potential customers get stuck and why?


Does your business model require partners? If you need a supplier or a distributor, can you reach out and talk to a few of them? What information do they need? Are they potentially interested in partnering with you? If so, under what terms and conditions? What types of firms seem to be most interested in partnering with you? Do you have bi-directional relevance that Alex Rampell speaks about?


How will you distribute your product? Are there distribution partners you need to contact? Is your distribution through the web? How much will setting up distribution channels cost you? Should you sell the product directly or through a third party or retailer? What is the best way to reach your customers? Who can you talk with to learn the best way to set up distribution for your company?


What are your costs and how can you get them to be as low as possible? See if you can estimate the cost of goods if you're producing a physical product. If you have a service or a website, then how much will personnel cost you in the first year or two? How much will server space and hosting cost you? Do you need customer support? I don't want you to spend a ton of time on financial models that will never come true. However, to check whether your business model is viable, you need to get sense for whether the dollars coming in through revenue will be greater than the dollars flowing out in costs.

Revenue Model

How will customers pay you and how much are they willing to pay? Is a subscription model better than a one time payment? Is it better to rent to them or sell? Does a freemium model make sense and if so, what features will people pay to upgrade for? Run some tests with customers to see what the best revenue model would be. Are you planning to make money through advertising? If so, contact a few potential advertisers and see if they are interested and how much they would pay you. How many users do you need for them to be interested in advertising on your site. Try to think outside of the box and not just rely on advertising-based business models. It's great if you can have a few different potential revenue streams to test out. Are there other ways you might charge for the product or service? Which might yield the most revenue? How do you know?

There may be additional aspects to your business model. If so, generate a few alternatives and get out into the real world and start testing some of them out to see what works best. You can't figure these things out just sitting around a table and brainstorming or theorizing. You have to get out there and talk to people out in the world who will often tell you things you never would have thought of or expected. Use this feedback to then refine your model. We want to see what you thought initially, what you learned from the real world interaction/testing and then how you revised your business model based on that feedback. Hypothesize, test, revise or in other words, build, test, iterate.

It's discovery-driven planning or customer development.