A Sickening Process of Admitting You Don't Know Jack

Guy Kawasaki has one more excellent post on how to build a plan. His best advice is to make it bottoms-up (what’s likely to happen vs. what you want to happen, what your best sales rep thinks you can do), to limit variable costs to keep per-unit profits high (Kozmo’s variable costs were all those bike messengers who lost the company money every time it made a sale) and to drive the model off demand-generation, a start-up’s most significant challenge.

He also talks about collaborating with investors to build the plan, which sounds like unrealistic VC advice but really isn’t — if an investor isn’t willing to work with you on a business plan for an innovative product in a large market, the investor is crap. Guy also suggests planning ahead 18 months or less. True dat, but half it: a plan really controls decisions you make over the next 6 – 9 months, best case. But you need five-year projections just to make sure you’re in the ballpark of having a business that can become large.

The best advice I can give is to apply the “Really?” test. Take all the assumptions in your plan that are kind of aggressive and pretend someone you trust said “really?” as in, “Really? $2 million per sales rep per year? Really? I mean dude, come on.” If you can defend it, keep it in there.

A few other suggestions:

Explain your assumptions: key variables in your plan may be how much traffic your site gets, whether your average selling price is $200,000, or when you ship a new release. You should call out those assumptions so your investors can use the model to understand the risks more precisely, tweaking variables in Excel to play around with what-if scenarios.

Share the plan: folks on your team should see the plan in its early stages and before it’s final. This will socialize them to the market dynamics that make you such a hard-ass as a CEO, so they take ownership over controlling costs and driving demand. And it makes it bottoms-up. Every person in the room will go to whatever line is her line and say, “I can’t do that.” Which is what you want.

Don’t get someone else to build the plan: nothing scares you about your own business like trying to make the numbers work. And if you’re not scared when starting a company, you’re doing something wrong. Out-sourcing the plan building to “someone who knows Excel” sets you up to get fried at a board meeting. P.S. if you don’t know how to use Excel, you shouldn’t be starting a high-tech company.

Think about replicating successes: for example, Redfin has started to generate real estate revenue from the Seattle market. So now we build our model off that: how much did it cost to build that market? How long did it take? How many Seattle’s are there? You can do the same thing with your first five customers. Extrapolating from something that feels real, anything at all, is a huge first step.

Think like the guy who has to make the plan work: model-driven projections can easily go off the rails, where a four-person group jumps to eight in one month, or revenues take crazy spikes.

But also, think like an investor: after you’ve built a plan that is doable, evaluate it as an investor. (First, is the plan credible and then…) Given the money raised, is there the potential for a 10x return in five years? One (somewhat dubious) rule of thumb is that a growing, profitable business without any particular superstar appeal can generate a valuation of five times revenues or 25 times earnings, and that a company needs to each $100 million per year in revenues to IPO.

Be able to answer the market-share question: at the end of your projections how much market-share do you have? If you need 70% market-share to make $100 million, you’re in trouble.

Line up reality with projections, literally: for example, you may have a line-item in your projections for marketing spending, which in your financial statements appears as three lines: promotions, tradeshows, advertising. When you miss projections, you’ll need a simple way to say what went wrong. Before finishing up a plan, it’s a good idea to ask your bookkeeper if you can report results in the same format as the plan.

It’s ok to miss revenues but never costs: there’s no excuse for being over-plan on costs, because it’s a number you control, simply by signing every check (which I hope you do).

Use this get out jail free card: if revenues don’t come in, we’ll stop spending. This prevents your plan from ever getting seriously out of whack with reality. And investors love to hear that you’ll run the business out of the cash register. Of course, the one type of spending that is hard to titrate is payroll: once you hire someone, you have to lay him off to get his payroll expense back, which you obviously want to avoid at all costs. So hiring is the expense to be most careful about. Once you’ve done a lay-off at any point in your career, it really changes how you evaluate hiring anyone.

Figure out the ratios: expenses/revenues/earnings per head, marketing cost per transaction, other cost per transaction, revenue per transaction, earnings per transaction, these are the numbers you need to know to figure out if you’ve got a good business. Thanks to Marc Singer of BEV Capital for this one…

Alright that’s it. And yeah I know there are some businesses that don’t model out — YouTube is a company that was spending $1M a month on bandwidth with no revenues — but those are the exceptions not the rules. I’ll stop blogging so much next week, it’s a weird mania isn’t it? Next up unless I chicken out is an entry on why there are so few Microsoft-inspired startups.


  • Marie

    Nice pic from Buenos Aires :)