SayAhh’s Revenue Projections

Since last checking in with the SayAhh team, they have spent a few months consumed with building an early version of the product and speaking to potential customers, all the while watching their cash balance steadily diminish. They realize the clock is ticking and have decided that it is time to create a robust set of financial projections in order to provide themselves with a better sense of when they will need to raise more money.

The co-founders decided to divide-and-conquer, with Dick tackling expense projections and Jane tackling revenue projections. Their plan was to combine the two in order to predict their cash burn over the next two years (with a focus on the next twelve months). Jane asked Josh, who provided SayAhh with solid advice on setting up their accounting systems, for help in creating the revenue side of their financial forecast. Josh told Jane to take a first shot and he would comment. Here’s a snapshot of what Jane produced:

Josh worded his feedback carefully:

“Jane, this is a good start. I am glad to see that you are forecasting revenues based on business drivers. In this case, the # of users and the average monthly revenue per user. That’s what you should be doing. However, do you really understand the key underlying drivers of your business? Based on the drivers you chose, I am not so sure you do.

Certainly the # of users matters. But take it a step further. What drives the # of users? Presumably you will have new users, return users, and lost users each month. Decomposing users into these three component parts is important because it will allow you to better understand what is going on with your business, which will in turn allow you develop actionable strategies for improving your business.

For example, assume the # users remains flat for four months in a row. If you only track monthly users, you might assume that you are not attracting any new users and you need to change your marketing approach. However, what if it turns out that your marketing approach is just fine and you are bringing in lots of new users every month, but at the same time you are losing an equal number of existing users? In that case, the problem is that users don’t like your product. You need to fix that problem, not adjust your marketing approach. Take another shot at this and come back to me with a model that you think drills down to the key drivers of your business.”

Jane did some research, had a nice glass of wine, and really thought through their business model. Then she came back to Josh with a revenue model built on a set of key business drivers:

Josh told Jane that her second effort was much better and Jane in turn felt that she now had a much better understanding of SayAhh’s business model. A skeptic might assert that it is a waste of time for a pre-customer startup to forecast revenues since they are guaranteed to be incorrect. When I look at a startup’s revenue projections, I don’t pay much attention to the actual numbers for just that reason. However, I do look at the structure of the model to see if they really understand their business and are actively tracking their key business drivers.

In the next post, we will see how Dick fares with the expense forecast.

Note that Jane’s second attempt is a step in the right direction, but by no means perfect. Tying the number of new users to advertising spend seems particularly questionable, for example. What other problems do you see? What has been your experience/advice in developing revenue projections?

  • My experience with revenue projections is that they are not nearly as valuable as operating results.  A good understanding of the structure of the future revenue model is the same as having a very detailed understanding of a fantasy.  And I agree, it is definitely beneficial to ensure you are going to *track* the right business drivers.  But once you start tracking results, you’ll have a much better framework for understanding the business.

    In terms of the example, the viral coefficient users seems dangerously unrealistic.  Zero to 10k in one month is not how viral loops work.  So then that becomes the first place to scrutinize with proven results (and if my pessimistic view turns out to be wrong, party time!)

    • Absolutely correct on the viral loop seeming unrealistic!

      • From what I can tell, the viral loop seems to be assuming 100% active users and 100% conversion, inflating the #s. Shouldn’t the 20% active users be factored into the coefficient? 
        Newbie by the way – hola! 

        • Kevin O’Leary

          – I definitely think you’re right that the team at Sayahh is assuming 100%
          active users in the viral coefficient and that is unrealistically high, given
          the assumption that only 20% of users are active users.  I often find
          there are two common issues when thinking about the viral coefficient: one
          being that entrepreneurs overestimate its impact as Sayahh seems to be doing,
          and the other being that entrepreneurs fail to consider it at all.

          the end of the day, I think that the primary value in these projections at this
          point is to help the Sayahh team think high level about their business and the
          drivers behind the business.  Inevitably, I think that a lot of the
          assumptions made in the revenue forecast will be wildly inaccurate (I really
          like Brad’s post: Your Revenue Forecast Is Wrong).  I think that a big part of the
          value for the Sayahh team in these projections thus lies in helping them to
          think about what drivers are important in their forecast.  They understand
          that there will be a viral aspect to their business, which I think is a really
          good first step, because they’re at least considering it at this stage, and
          that is important.  While their
          assumption that each month, every new user will drive another new user might
          not be accurate, they’re at least thinking about it.  Later on, when the
          team gets more data on usage, I think that this will help them to think about
          how to alter their future projections.

          while their numbers are unrealistic, they’re at least on the right track of
          thinking about what drives their business and using that to drive their
          projections.  Hopefully, with a bit more experience, they will understand
          the mechanics of how viral coefficients work and can drive to a more accurate

          Would love to hear your thoughts on that.

    • Lateef – I agree with you in part, but I think this kind of breakdown at least gives you a sense of what you need to *do* to hit your numbers – ie acquire X users through marketing (so we need to do marketing), Y users through viral mechanisms (so we need to make it easy for the product to be shared virally), etc. 

      I write a blog about, among other things, how to gain business insight from operational data – please take a look if this is something you’re interested in:

  • Jorge

    Thanks for sharing the information.  I run an online company but I have a real estate finance background and I usually try to make online accounting and finance fit into an apartment building model (highly modified).  It works on many levels but fails on others.

  • I agree, putting a revenue forecast in place is hugely valuable.  Jane’s weak first attempt gave a starting place for improvement.  Taking a page from Eric Ries’s “Lean Startup,” the gauge of a revenue model is how it helps founders 1) test, 2) learn, and 3) improve. 

    Each of the assumptions in the 2nd model looks testable.  I suspect more details/granularity will be in order before Jane will really understand her customers.  For example, will ALL users really refer one new user or just those who become active?  This model queues Jane up to quickly learn the details she needs to add to understand her customer conversion process.  In tandem, she can improve her metrics, model, and product.

    Check out this blog post I put up recently.  Gives a process for developing your revenue model and a downloadable Excel version of the model.

  • Yesuifen20

    the gauge of a revenue model is how it helps founders 1) test, 2) learn, and 3) improve. 

  • Pingback: Reblog – Finance Fridays (Sayahh’s Revenue Projections) | ithacaVC()

  • Scenario modeling is a great idea in concept, but it can give a false sense of having actually considered all scenarios. The biggest problem tends to be that they have all variables move in lockstep from one scenario to the next, which creates extreme best and worst case scenarios. In reality no business drivers are perfectly correlated, and in many cases drivers are inversely correlated or not correlated at all.

    On the other hand it’s not practical to have X^3 scenarios to model all possible combinations of independent best/worst/middle variables (where X is the number of business driver variables). In my experience – which involves way more financial modeling than I’d like to admit – the only solution is to play with the model and see how changes in each variable affect the results. Having multiple scenarios side by side – or being able to toggle b/w them by changing only a single cell – is nice, but it is no substitute for reality testing both the inputs *and* the equations. If someone has a better solution I would love to hear it!

    She also seems to forget to use % active users in her model.

    • Aaron Lucas

      This is why I do best case worst case.  Then I do a monte carlo simulation where I set expectations on all the variables.  Still the usefulness of the simulation/model all depends on your ability do understand all the major drivers in the business.  No math trickery will fix a lack of understanding.

  • Thanks for posting. I’m going to use this for our startup. I noticed something, Brad: the last sheet isn’t subtracting user loss – it’s adding it instead. fyi.

    • Good catch. I’ll try to update at some point.

  • Pingback: SayAhh’s Revenue Projections()

  • Daniel Wiggins

    Cash vs GAAP.  Both work if the model is correct (though I think cash is better in this situation). Whatever provides the most direct answer to cashflow (and least risk associated with knowing what the company’s cash balance will be at the end of month #1, 2, 3 etc).  Its about how much cash does the company REALLY need (or when is it going to run out).  More cash > less cash, cash now > cash later. Cash cycle also important (e.g. A/R collection, A/P payments). Generally financial projections are the weakest part of any business plan – unrealistic assumptions, especially around timing of revenue ramp, development time and CF breakeven.  Compounding this is the ommission of expense and balance sheet (e.g. prepaids and capx). Be honest about the forecast. 

  • Pingback: Online Cash Advance()