In last week's MBA Mondays, I introduced the topic that we'll be focused on for the next month or so; projections, budgeting, and forecasting. In that post, I described projecting as a "what if" exercise that is done at a higher level of abstraction than the budgeting and forecasting exercises. I said this about projections:

These are a set of numbers, both financial and operational, that you
make about your business for various purposes, including raising
capital. They are aspirational and are often done with a "what could
be" perspective.

Since projections are not budgets and are much more "big picture" exercises, it is important to use a scenario driven approach to them. I generally like three scenarios; best case, base case, and worst case. But you could do as many scenarios as you like. It's not the results that matter so much, it's the process and the learning that comes from the projections exercise.

If you build your projections with a detailed set of assumptions and if you can assign probabilities to each assumption, you could easily do a monte carlo simulation in which literally thousands, or tens of thousands, of scenarios are run and the outcomes are charted on a bell curve. I don't recommend doing projections this way, but my point is simply that the number of scenarios is not important, it's the process by which you determine the key drivers of the business, the assumptions about them, and the probabilities associated with them.

A few weeks ago on MBA Mondays we talked about key business metrics. It is very important to identify your key business metrics before you do projections. These key business metrics will drive your projections and your assumptions about how these metrics will develop over time will determine how your scenarios play out.

Let's get specific here. I'll assume we are operating a software business and we are selling the software as a service over the internet using a freemium model. Everyone can use a lightweight version of the software for free, but to get the fully featured version the user must pay $9.95 per month. So here are some of the key business metrics you might use in projecting the business; productivity of the engineering team, feature release cycle, current outstanding known software bugs, total users, new free users per month, conversion rate from free to paid, marketing dollars invested per new free user, marketing dollars invested per new paid user; customer support incidents per day; cost to close a customer support incident. These are just examples of key business metrics you can use. Every business will have a different set.

The next step is to lay all of these metrics out in a spreadsheet and make assumptions about them. As I said, I like three assumptions, best case, base case, and worst case. Best case is not the best it could ever be but best you think it will ever be. Base case is what you genuinely expect it to be. And worst case should be the worst it could ever be. Worst case is really important. This is your nightmare scenario.

You then calculate your costs and revenues as a function of these metrics. There are some expenses that will not vary bases on the assumptions. Rent is a good example of that in the short term. But over time, rent will move up if you need to hire like crazy. I would go out at least three years in a projections exercise. Some people like to go out five years. I've even seen ten year projections. I don't think any technology driven business can project out ten years. I am not even sure about five years. I believe three years is ideal.

Getting the assumptions right and building up to a full blown projected profit and loss statement is an iterative process. You will not get it right the first time. But if you build the spreadsheet correctly the iterating process is not too painful. Do not do this exercise all by yourself. It should be done by a team of people. If you are a one person company right now, then show the results to friends, advisors, potential investors. Get feedback on your key business metrics, assumptions, and results. Think about the results. Do they make sense? Are they achievable?

In last week's comment thread we got into a conversation about "top down" vs "bottom up" analysis. Top down is when you say "the market size is $1bn, we can get 10% of it, so we'll be a $100mm business."  I think top driven analysis is not very rigorous and likely to produce bad answers .The kind of projection work I've been talking about in this post is "bottom up" and is based on what can actually be achieved. However, it is often best to take the results of a bottom up analysis and do a reality check using a top down analysis.

So when your best case scenario has your business at $100mm in revenues in year three, do yourself a favor and do a top down reality check on that. No matter how rigorous the projections process is, if the results are not believable then the whole exercise will have been wasted.

Most entrepreneurs do projections as part of a financing process. And it is a good idea to have projections for your business when you go out to raise money. But I would advise entrepreneurs to do projections for themselves too. It is a good idea to have some idea of what you are building to. Make sure it is not a waste of time for you and the team your recruit to join you.

It is true that most great tech businesses, possibly all businesses, are initially built to "scratch an itch." But once you get past the "I built this because I wanted it" and when you find yourself hiring people, raising capital, renting space, it's time to think about what you are doing as a business and having solid projections and a few scenarios is a really good way to do that.

#MBA Mondays

Comments (Archived):

  1. Julien

    What tools do you recommend? Excel immediately comes to mind, but I know you’re a fan of Google Docs. Do you think it’s good enough for this? Numbers on the Mac? Paper and pen?

    1. fredwilson

      you need a spreadsheet for thisi don’t know if google docs is strong enough yet for this workso i’ll say stick with excel for now

      1. Pascal-Emmanuel Gobry

        This is exactly the kind of thing that makes me think the right cloud integration with Office ( could crush Google Docs unless they dramatically improve the product.Microsoft’s “software plus cloud” approach is derided by some as not forward-looking enough but for some key use cases given the current technology it’s still a winner.Obviously this implies Microsoft is willing to cannibalize itself to an extent to do that.

    2. reece

      I use Numbers and I love it.

    3. Fernando Gutierrez

      I’ve done a lot of models and I’d always recommend Excel. I use a lot Google Docs also, but not for complex spreadsheets.In Excel you can manage bigger files, calculate huge amounts of data, use specific add-ons, copy and duplicate parts of the model with more flexibility, link to other Excel files, use the set of tools it has for what-if-calculations, scenario tools…The only part that Google Docs can have here is to work with your team in the assumptions. Excel is great, but it can’t beat Google Docs on collaboration.

      1. Tereza

        I totally agree. If you have to manipulation, you’ll go further faster on old-fashioned Excel. But when you hit steady-state and collaboration time, then Google docs.

        1. Fernando Gutierrez

          What I’d love to see is Excel being able to pull data from web sources easily (you can already do it but it’s quite complex), be it GDocs or any other similar service. This way you’d have the best of both wordls. I’ve read that the new MS Office coming out (this year?) is supposed to do things like that, but I don’t really believe they’ll get it right.

    4. Mark Essel

      Julien I have some great links sorted away in my financial folder for VM in gmail. They were python written I think but can really make a hackers entry to projections easier. Excel sheets feel so limited after many years of programming.I’ll dig them up if you like and link them(har to on phone).

    5. RichardF

      Julien, Excel is fine for this process as is Open Office if you don’t have access to Excel.

  2. JimHirshfield

    “If the results are not believable…”- I think that’s the crux of it – not just projections, but the whole pitch to VCs. How often do you see weak business pitches because the assumptions are unrealistic. I’m guessing quite often.

    1. fredwilson

      yes, but that doesn’t kill the deal for mewe want to back great product oriented entrepreneurs not accountants

      1. Mike

        Yes, but what if the great product oriented entrepreneur is also an accountant?

        1. Tereza

          Make sure that you or someone you rope in talks to real customers (or prospects) and can pressure-test the assumptions on that side.Sometimes a disconnect happens where the product and spreadsheet are engineered perfectly but not in relation to customers.Talking to customers can be scary because (a) it’s time consuming, and (b) they tend to spray all kinds of inconvenient realities all over you, that can complicate your thinking.So then you have to step back with the whole team and together decide What’s In, and What’s Out.That conscious decision of what you’re not going to do now — or where later down the road you’ll put it — helps make a happier team because their own success metrics are clear.

          1. Mike

            I probably should’ve clarified that though I’m personally a CPA, I am acting as the product evangelist for my own company (interacting with clients, working with our developer, etc…)I fully plan on outsourcing the accounting dept once we grow beyond what I can personally handle just like everyone else.Because NOT working in accounting is the dream of just about every accountant I know.

          2. Tereza

            LOLAnd I don’t mean to be sour on accountants. I was at one of the big firms for a while, and a lot of those guys have been big advocates of things I’ve done.A super smart, entrepreneurial businessperson w CPA is a treasured combination!

          3. JLM

            I am a huge fan of focus groups. Not little one way mirror focus groups but ones in which you dare to engage. You have to get vulnerable to get comfortable.

          4. Tereza

            …and then go to market with a hell of a lot more confidence.Also greatly helpful in prioritizing your prospective client targets. Who’s easy pickins’? Who has a long sales cycle? What is each co’s budget cycle? Who’s the budget holder?

          5. fredwilson

            Great advice. Step out from behind the model and get into the real world

        2. RichardF

          Mike – you are doing the important bit – building it and implementing it and you have the added advantage that you understand the numbers.

        3. Donna Brewington White

          It has always seemed to me that anything AND an accountant is always a great combination especially if the other “addend” involves a creative element. Just don’t tell us that you can cook too!(I should clarify that I think starting a company is a highly creative endeavor and that most entrepreneurs have a creative side.)

          1. Mark Essel

            Multi-talented chefs? That’s my fiancΓ© Michelle all right. I had to beg her to stop cooking as I was getting to fat πŸ™‚

        4. fredwilson

          Then you’ve got a very special person and you should think hard about working with them

      2. Mark Essel

        Amen to that

      3. Kevin Vogelsang

        I think Steve Blank would agree heavily with that approach

      4. ShanaC

        At some point- I have to ask, what is a product? I work with medium, and I keep asking myself, what am I working with?

        1. Tereza

          ultimately, something that is replicable/repeatable and that you will get paid for.that’s as opposed to a service.

  3. Hank Williams

    “But once you get past the “I built this because I wanted it” and when you find yourself hiring people, raising capital, renting space, it’s time to think about what you are doing as a business and having solid projections and a few scenarios is a really good way to do that.”But it seems to me that for a lot of the great investments that union square has made that this could not possibly be the case, twitter being the most obvious. I presume that the rule for this is to follow this rule except when you dont think its necessary to follow this rule… is that right?

    1. fredwilson

      that’s the public perception of twitter and tumblr and others but that’sbecause the public doesn’t get to see what is actually going on behind thescenes

      1. Fernando Gutierrez

        And you won’t tell us, will you? please, please, please! πŸ™‚

        1. Mark Essel

          Fernando we can get more info from the book Fred pointed out, and likely much more if Twitter goes public or is acquired. Some of those details are probably safer as a post mortem, or better told by the founders. I suspect Fred wants to share some of them, but he got beat up a little for the pre Chirp post, likely purposefully. Gotta love VCs who take public flak off of portfolio startups. It’s not always a competition or who you know, but how many punches you can take πŸ˜‰

          1. Fernando Gutierrez

            Yeah, we’ll have to wait… I had to try, though. And one more book to the list!

          2. fredwilson

            Well I’m not writing the book. But I’d love to read it

      2. Hank Williams

        Wow. So your saying twitter had “solid projections” when you invested? Its hard to imagine anyone having enough prescience to predict what their revenue was going to be at those very early stages. If they really did, that is incredibly impressive. It makes me feel really small πŸ™‚

        1. Mark Essel

          Odds are there projections shifted as fast as the network grew. If they had even an inkling of their growth they would have made different engineering decisions early on. The amount of “whales flying around” although tastefully done, is not usually something that is planned for a com network.

          1. fredwilson

            So true. That’s pretty much what happened

        2. fredwilson

          I’m not saying that they had ‘solid projections’. But they had projections. Some of them were outed by arrington in the theft of their proprietary memos and emails that happened a few years back

      3. Tereza

        You mean, what they’re really building inside the Trojan horse?Hold on, you don’t mean to say Twitter is a front — and behind the scenes they’re focusing their tehnical might on Project COS Alpha.(that’s code for Cone of Silence).Shhhhhhh….

        1. ShanaC

          We’re still waiting for it. I would love that….

        2. fredwilson

          They are most certainly not working on the cone of silence!

      4. Donna Brewington White

        So they are not making it up as they go, after all? ;-)I wondered if given their real-time product development model they might have a real-time business model as well.

    2. reece

      Just because Twitter didn’t announce a business model publicly doesn’t mean they didn’t have one in mind privately all along.If anything, they knew that they had to scale up large enough before they could make their model work.Still, I’m interested in the thinking of VC’s behind such a bet…

  4. Pascal-Emmanuel Gobry

    I absolutely endorse the three scenarios idea. I got that from a friend who used to work in the French intelligence services who told me that’s how the KGB worked. Whenever someone wanted to argue for doing something, he had to write three memos: best case, base case, worst case.

    1. Tereza

      LOL I’m imagining culture playing into the best/base/worst. For example is an American Best rosier than a Russian Best (I’m pretty sure it would be sunnier than a Czech Best). Whereas give a group of my peeps a bottle of Becherovka and a long night, and I promise we could craft Worst Case scenarios that would make a grown man cry. πŸ™‚

      1. JLM

        Very keen insight. Americans are much more adventurous — or they used to be. Going to the moon? Finishing the Panama Canal? Hoover Dam?I think American entrepreneurs are also unleashed by the inherent freedoms of the country and the marketplace.This is also why there are no real new jobs today.You cannot simultaneously punish and encourage the same segment of society.The beatings will continue until job creatiion increases!

        1. Tereza

          Yes and immigrants who come here are like the proverbial optimistic kid shovelling sh*t: “Just looking for the pony!”

        2. fredwilson

          There are plenty of new jobs in our portfolioThe idea that there are no new jobs right now is just not true

  5. reece

    Interesting that your list of example metrics doesn’t immediately include Cost of Acquiring a Customer (CAC) and Lifetime Value of a Customer (LTV).I feel like those traditionally get the most attention from the VC world, (with good reason, but not necessarily the most important to a given business).

    1. Mark Essel

      Those are input parameters to the model Reece. Major ones.So according to Erik these are resultants of other input parameters. I wonder if things like “social virality” are considered in these models

    2. ErikSchwartz

      Both Cost of Acquiring a Customer and Lifetime Value of a Customer are derivatives of other forces. You wouldn’t be tweaking them directly.

    3. fredwilson

      CAC is sort of in the listLTV is not but should beI was just running off a bunch that came to mind. I didn’t mean for the list to be definitive in any way

      1. reece

        I figured as much, I just thought it interesting that they weren’t the firstones up there.

  6. robertavila

    The problem with the Best/Base/Worst triptychs are two fold. First people tend to look only at the center panel treating the base as the most likely and Second the two side panels tend to be uninformatively unrealistic in that the they appear to be either “Most everything goes right” and “Most everything goes wrong” If one is going to take the important step of doing some alternative scenarios, one should do at least four and they all should be plausible alternatives. Begin by identifying the key business assumptions that actually could go significantly right or wrong and investigate what their impact on the business would be: The technology is ambitious and problems here and here could result in … ; There is only one other player if they get their act together ….; Something like our marketing approach has on one occasion been wildly successful… Then flesh out in some credible detail what the impact of each would be. For all of them these will be a mixed bag of impacts, responses and actions. the nice thing about this is that they illustrate a responsive management team, thinking about the business armed with a robust core strategy.

    1. Tereza

      Robert I like how that sets up a robust conversation between the functions about how they interact (ie between devt and marketing).The ‘muscle memory’ developed in that exercise can be useful in the future when real things start happening and there’s a lot more noise.

    2. Mike

      I like this approach, but four alternatives seems a bit arbitrary once you’re at 3-5 years out (why not 2 or 5?). If the logic behind your base case is valid, you can speak to the reasons for the best/worst case, and your model allows for some flexibility, the rest can be fleshed out and updated in real time.Plus I think each of the scenarios you point out could potentially be given a weight and assigned to one of the 3 buckets Fred mentions.

      1. Tereza

        I don’t want to speak for Robert but I think his point is less about how many (other than he finds 3 can drive an inconvenient line of thinking). The bigger point being, come up with scenarios that optimize for certain highly volatile/exogenous assumptions — some being related to product, some related to customers, competitors, etc.Thing is “Best” isn’t really a scenario. It could be tempting to just jack up/down all the levers in model to max it out. But there are multiple paths to get to Best (and Worst). It’s the story behind Best/ Worst that needs to be picked through, that’s of great benefit to the whole team. Through this they understand each other’s challenges. A scenario narrative is more like: we get to market on time, product/ mkt fit is perfect, customers love it, we have backlog; but wait they’re pissed there’s backlog, need to manage that….Obviously you don’t want to over-engineer this; but being open to making sure the stories behind the scenarios are logical and internally consistent is wise and possibly quite beneficial.

        1. Mike

          Nicholas and Tereza, great comments. I think we’re basically in agreement in that that the real key is to know your business model cold. Inputs, assumptions, scenarios, etc…There are an infinite number of scenarios that an investor might consider, so without a thorough knowledge of how the key drivers of your business interact you’re SOL regardless of whether there’s 4, 5, or 10 cases on your spreadsheet.

        2. robertavila

          Tereza, you can speak for me any time you want!Another problem with worse case scenarios is that often that is the “Management Team does a rotten job” scenario. By spelling out specific down side risks and indicating what would the management team would do, the down side risk can be come a plus for the proposal.

          1. Tereza

            Ah, the Elephant in the Room Scenario ?!But Robert, that is Wholly Implausible and therefore not worth modelling! “My team” would never let that happen!LOL

      2. Nicholas Davis

        I’m a fan of Robert’s approach. Yes, choosing four is arbitrary, but both two and three scenarios suffer from similar problems – a good/bad or good/neutral/bad intuition from the audience. Five+ can work, but once you get past four you might be overloading the deck.Mike, I think categorizing is useful if you take the approach of first constructing four (or so!) plausible but challenging scenarios (based primarily on external uncertainties) and *then* thinking about what they imply for the bottom line/funding, categorizing them as good/base/bad based on impact on the business. But the trick is not to *start* by trying to construct scenarios to fit these categories, or you’re prone to miss a range of risks and opportunities that scenario analysis could offer. And don’t necessarily think you have to abide by purely quantitative indicators of success – considering a few extra qualitative uncertainties to the mix can be extremely valuable.In my mind the most important aspect is that you consider a variety of scenarios based on (clearly identified) uncertainties that could seriously impact the plan. That shows a) a recognition that “no battle plan survives contact with the enemy” and b) that you believe the business can be designed to be robust to “known unknowns”. And as Tereza says, it also might help with your speed of response if it hits the fan.

      3. Fernando Gutierrez

        That’s why I prefer a few thousands and some analysis. The problem is that I think that our minds don’t feel comfortable with a distribution and prefer values. It’s easier to compare and tell others… And you also risk loosing too much time/energy on crunching the numbers.

        1. fredwilson

          That’s true for many people but I love looking at distributions

    3. fredwilson

      That’s a good argument for a few more scenarios. I’m with you on best case. But I think worst case is really a valuable scenario

  7. LIAD

    I’m curious as to what kind of spreads you see in top line revenue between best, base and worst case. Do they get as extreme as – $100m, $10m, $1m?

    1. Mike

      I’ve generally seen something more akin to a sensitivity analysis, where you take the base case and add or subtract a percentage based on the projected (or historical) volatility of the relevant inputs.

    2. Tereza

      Good question.

    3. fredwilson

      Its all over the map. But a good set of projections for a 10mm revenue business might have 20pcnt to 30pcnt spreads

  8. Mark Essel

    My heart skipped a beat at the mention of thousands of Monte Carlo runs, this is what I do for a day job. I’m like a projections “guru”. What if scenarios, changing assumptions, turn the crank and generate another set of slides. Real data changes all the assumptions and you refine your models and understanding. Modeling and simulations is etched into my DNA, as much as I work to branch out.But let’s not restrict our projections to normal distributions, the shape of that curve is like stepping back and looking at the process and model of your business. You can find out which parameters your growth is most sensitive to and push like crazy on those sensitivity points. Am I the only one who get’s excited about hacking the worlds biggest industries?

  9. taylorwc

    Great insight that works from both sides of the table; as soon as the investor is trying to justify your valuation, he/she will be doing the same thing on the back of a napkin. Worst, best, and base case for the investment. It’s amazing how much of this advice for entrepreneurs works as advice to investors.

  10. zackmansfield

    One of the most important reasons to do a legit, honest bottoms up financial projection is so you can get a real idea of what the potential financial scenarios are for your business as you look into financing options. There is absolutely nothing wrong with starting a business that is not going to be VC-backed (b/c it will never deliver the scale or growth possibilities needed for VCs to invest). Unfortunately, too many use a different order – build a model that a VC will invest in, whether or not it’s realistic or not.The goal is to build a great business, not to raise VC money. That business may be a $2mm/year lifestyle biz, and that could be a beautiful thing. Or it could be a high growth VC-backed business. Both can be great, but make sense to be intellectually honest up front to avoid wasting time/energy/personal funds on a dream which is mismatched with the appropriate financing strategy.

    1. fredwilson

      Many years ago the VC who taught me the venture business met with an entrepreneur that he took an instant liking to. The entrepreneur told him that with an investment of 1mm, he could build a business that would be 10mm in revs in five years. The VC passed because the business was too small for VC investment. The entrepreneur fired up lotus 123 (it was a long time ago) and ran some numbers and came back to the VC with new numbers. The business would now be 30mm in five years. The VC invested, the businesss worked and after five years it was a nicely profitable 10mm revenue business

      1. Tereza


  11. Dan Ramsden

    The thing that I struggle with the most is the ‘key business metrics.” In sectors like energy or real estate, these are straight forward… but in digital media, not so much. Are Facebook’s business metrics the same today as, say two years ago, or even last year? At what point would iTunes have been incorporated into Apple’s key business metrics on a 5-year model? The year before it launched? Two years before? Three?What I’m trying to say, not sure if I’m succeeding, is that the very business profile of an enterprise in this sector is prone to change in material ways from one year to the next, or faster even, and this makes a multi-year forecast very murky… more so in this sector than in others.

    1. Tereza

      True. Big pivots.

      1. Dan Ramsden

        And if the 5-year model is used for valuation purposes, even more treacherous… considering that most of the present value comes from the terminal (year-5) valuation estimate, which could be showing an entirely different business than will eventually be the case.

        1. Tereza

          LOL. Totally.But seriously — this early stage, isn’t the valuation truly whatever the market can bear?Isn’t the model — important as it is — just a reference point providing conversational grist for the negotiations. At end of day, if you’re negotiating with many, valuation goes up! If you’re having a convo w no one except your belly button, valuation is just whatever cash you put in the business. πŸ™

          1. Dan Ramsden

            Yes on all counts.I guess the moral of the story is that at best the 5-year model should be updated/revised/reformulated constantly, and at worst may be meaningless.

          2. JLM

            Even when the forecast is suspect, its preparation has the benefit of informing the mind of the preparer. When you know how speculative the outcomes truly are, you begin to understand the real risks.

          3. Dan Ramsden

            And there is a point at which the outcome is so speculative, that the value is really option value.

          4. JLM

            I think I understand what you are saying and agree. Elaborate, please. Thanks.

          5. Dan Ramsden

            When interim cash flows are negligible if not negative, and when 100% of the expected return on an investment will be realized from its sale, then all the value is in the terminal value. When there is no way to really know how to assess that future value because there is no way to know what the underlying business will be, then you are investing in an option. You purchased a call on a future asset that remains to be defined. If things work out, your call is in the money when you exit.

          6. JLM

            Got it. Perfectly explained and understandable. Thank you, Dan.

          7. Dan Ramsden

            My pleasure.

          8. fredwilson

            Remember that the process is more important than the result.

          9. Fernando Gutierrez

            Pure Adam Smith!

        2. JLM

          When the NPV of a deal comes from the most highly predictable element of its cash flows, then it becomes more attractive.When the NPV of a deal comes from the most temporaly distant and uncertain cash flows (disposition), then the deal is more risky and tenuous.In some ways, the discount rate has to become risk adjusted and this brings a bit of clarity.

      2. ShanaC

        The web seems to be a large pivot. It’s a <3

    2. Fernando Gutierrez

      I’m not so sure about that. A few years ago someone in real estate would be watching population growth, price indexes, mortgages prices… Today, if he still on business, he’ll be worried about unemployment rate, foreclosures, government money…I know these are exceptional times, but most times are exceptional in one way or another.

      1. Dan Ramsden

        Yes, but those variables you list would impact the underlying assumptions about growth, pricing, etc… without changing the core business model. In your scenario the real estate CEO is still in the business of building and leasing square footage. What I’m referring to, on the other hand, in keeping with your analogy, is a scenario in which that same real estate business becomes a vendor of advertising solutions. How do you model for that possibility?

        1. Fernando Gutierrez

          IMO, you can’t. Just hope to be lean and flexible enough to survive.By the way, I know a few real estate guys doing completely different things to survive. Related to their field, but with different metrics. One started to collect debts from their own debtors and finished doing that for other people. Today that is the company main business.

          1. JLM

            Real estate is very, very cyclical. Institutional investors wanted developers to create product (office, industrial, multi-family) and then they wanted them to stop and more effectively manage the existing inventory. This is quite typical.When the dentists are getting in — get out.When the insitutions are getting out — get in.The real challenge with buying low and selling high is to have the courage to buy low. And, yet, that sentiment and entry point has always been obvious and apparent. What is lacking is courage.

          2. Fernando Gutierrez

            That’s so true. I think Warren Buffett said something similar about stocks and the need to have your own criteria to get in when everyone was running scared.

          3. fredwilson

            That dentists/institutions line is priceless. I’m sure its common wisdom in real estate but I really love it

      2. ShanaC

        At the end of the day though- the one thing you need to worry about in real estate is how many empty units you have. There is no equivalent in digital media (you can do quite well in a niche community if that is your goal)

      3. JLM

        The big challenge is to find the right metrics.Real estate — debt as a % of income?

        1. Fernando Gutierrez

          That one is quite reasonable… although a few guys I know in real estate would cry like babies when looking at it. I know in the US you’ve got a huge housing crisis, but I live in Spain and here it’s been apocalipsys. The previous years were crazy and the crash was huge. Not so bad in the long term, though. The country needed to refocus.

    3. fredwilson

      Yes. That is very much true about tech investing. I don’t know if I said this in my key business metrics post (I hope I did) but you need to constantly reassessing your metrics and throwing out old ones and adding new ones

  12. amin

    I have a similar situation from the free/premium scenario you describe on your 5th paragraph.I currently allow free users and premium users but since am starting out, I have been upgrading everyone free of charge.In actuality, on my end, there is really not much cost difference between one user and the other but looking at competitors, the average paid premium service is about 200 to 300 bucks a year for the kind of service am dealing with.With that in mind, because we are in such early stage, we really can’t accurately or confidently project most of the stuff you describe here because we are fixing our own bugs on our own time, we are doing our own marketing initiatives with no money other than some basic facebook ads but actually 98% of the users we have recruited so far were contacted directly.How can someone in such early stage can come up with reasonable projections?I wonder how the approach on one’s assumptions would be best when you are in such informal phase.

    1. fredwilson

      The process is more important than the result. I say just do it, make up numbers if you have to at the start. You can replace them with real numbers later

  13. ErikSchwartz

    I’ve done this a bunch of times for a bunch of start ups (and a few times for biz units of larger companies).For me I find the process more useful than the end results. It makes you understand the driving forces behind what you’re doing. For me the assumptions document is more interesting than any of the projections. It’s much harder to bullshit yourself writing out the supporting assumptions than to BS yourself in Excel.Then give the whole thing to the most skeptical person you know and get their feedback.

    1. RichardF

      I agree Erik, in a pure start up environment, particularly tech, it’s very difficult to get revenue forecasts anywhere near correct, except maybe year 1 where most people do forecast in the range 0 to very little (or if it’s biotech very little for a long time)However the assumptions process is key.

    2. Mark Essel

      “Then give the whole thing to the most skeptical person you know and get their feedback. “Excellent advice thanks Erik.

      1. ErikSchwartz

        Sometimes that person is Fred πŸ™‚

        1. Matt A. Myers

          I LOL’d – thanks

    3. Tereza


    4. JLM

      The key to quality painting is the preparation before the paint brush ever dips into the can.The key to a great business is the founder knowing how the numbers were derived.Making the numbers is just practice for how we intend to play the game.We all play the way we practice.

    5. ShanaC

      Thinking about the comments- Scenarios hold a great deal of insight about a person’s psychology….

      1. ErikSchwartz

        It all kicks back to that discussion of the importance of narrative we had last week.All start ups say they will be doing $50M in revenue after 5 years. The narrative and assumptions lets us know how credible that is.

    6. Tariq

      I agree – the process really helps you see how all the different parts of your business translate into your financials. And giving that data to a skeptic is an excellent way to check your assumptions against whether or not they are too bullish

    7. fredwilson

      Great advice erik

  14. Dan Cornish

    Fred,Can you share some metrics you like to see for engineering productivity? Every time I try to measure new development, I have to create a new set of metrics. We use Fogbugz and their evidence based estimates to estimate time to build, but how do other people measure it?

    1. fredwilson

      I am not the best person to address this issue to be honest. Maybe someone else in this thread can help

  15. jkaljundi

    I’ve always felt startup entrepreneurs can learn a lot from real options analysis when going through their scenarios.

  16. JLM

    The ability to predict a set of future cash flows from a set of assumptions is the essence of business forecasting and modeling. I have done this literally thousands of times and though I have the calmness of having seen the same movie a thousand times, I am still always in suspense awaiting the denouement. I always run out of popcorn before the credits run.The only useful things I have learned are:Document all the assumptions and make the first tough conversations not about the spreadsheet that results from applying the assumptions but the assumptions themselves thereby foregoing our general nature of rooting for a particular outcome. I tend to root for John Wayne and hey, sometimes but not too frequently, he is NOT going to get the girl.Take a minute to write out the logic of the assumptions and the ranges you have picked. Put allthe assumptions on the first sheet of the spreadsheet so you cannot see the spreadsheet as one peruses the assumptions.I don’t really care whether there are 3-5 different sets of assumptions and how they are mixed and matched. I just want to know why someone picks a particular set of assumptions but always BEFORE I see the finished spreadsheet. Then I become influenced by the outcome rather than the assumptions. Like seeing the end of the movie before the first scene. Nah! Not good.Make your scenarios flow from the range of reasonable assumptions not from the finished product. The scenarios should simply dance to the assumptions and not be “bashed to fit”. There is nothing worse than “adjusting” a spreadsheet because you don’t like the outcome. Guilty!Try to separate all fixed costs (rent, utilities, even people to a certain extent) from the variable revenues/costs thereby understanding that there is a “nut” which must be cracked regardless of what happens with the variables. How much is fixed and how much is variable? How much of the variable is directly controllable and how much is not? Try to get a handle on what percentage of the business is left to the whims of the gods. What the hell can you really manage?Graph as many of the outcomes and variables over to time to see if the graphical image seems reasonable. I have kept myself from wandering into the swamp a few times when confronted by the slope of the curve and saying — wow, that doesn’t look right.Get a lot of feedback from others. A lot of feedback but always start w/ the assumptions. It is great to learn from experience but it is much less expensive to learn from somebody else’s experience.Most things take twice as long and cost twice as much as we initially think they will. Understand and anticipate this.Continually update your forecasts as you sink your teeth into the business. I would update any forecast every 90 days even if the update was just to change the damn date. As the real numbers begin to emerge your confidence interval will close because the assumptions will now become realities.Point of order — when making forecasts to raise money or get a deal done, don’t be afraid to apply just a smidge of vaseline because the money guys are going to arrive w/ the clipping shears in hand. As they said in the Godfather — this is the business we have chosen!Bad fact — some deals don’t work. Don’t beat a dead horse.

    1. Tereza

      Much to like in this comment but my two favorites are (1) thinking thru the assumptions/scenarios first, before running them thru the model, and (2) defining that “nut”. That nut drives a lot of clarity for the team.

    2. paramendra

      “Pay” for one blog post, get one free.

    3. ShanaC

      Ok- signs the deal isn’t working?

      1. JLM

        When the deal does not achieve your hurdle rate expectations across the entire continuum of reasonable outcomes. Never, ever draw to an inside straight.When the range of outcomes is very, very broad indicating a huge amount of sensitivy unless you are operating within your own comfort zone. Your personal strength.When the numbers look right but your gut tells you —- Whoa, Nelly! Believe your gut when it tells you to charge Hell w/ a half thimble of water but also when it says — No way, Jose!When the rewards are way out of whack in comparison to the risks.When a black cat walks across your path on the way to the pitch! When you used to date the guy you are pitching to or his sister and they were both lousy kissers. OK, that one’s not as likely to happen, but it could! It could?

        1. ShanaC

          Don’t believe in black cats- I might start a business with a guy I dated, but do you know how hard it is to get a date in this town (at least for me???)

      2. Tereza

        Same as dating. He’s squirmy, defensive, non-committal, doesn’t call, he’s defensive….All signs he’s just not that into you.Better to give him space. Go have fun on a whole lot of other dates. Then when he bumps into you while you’re out on a date with another guy, he’ll suddenly be aroused.

        1. ShanaC

          I hate when you are right about Men ;)- business I don’t mind at all. Though what is with people giving me dating advice on this blog (not the first time)

    4. Mark Essel

      JLM = $$

    5. Donna Brewington White

      I am going to save this comment! The advice is transferable to so many different scenarios (no pun intended).

    6. fredwilson

      That’s a blog post jlm and the readers of this blog’s comments are better off because you wrote it

  17. paramendra

    Best, base, worst. The good, the bad, the ugly.

  18. mikenolan99

    Another great post – you continue to post the things I want to talk about – not only before me, but better than I ever could.With your permission, I posted an excerpt at the Minnesota Small Business Development non-profit that I run. http://www.myminnesotabusin

    1. fredwilson

      My posts are free to reblog/cross post anywhere appropriate and anytime as long as I get attribution and a link back

  19. GlennKelman

    I don’t have to much to add to the (excellent) comments here except that in a brand-new business that has not raised a large round of financing, projecting costs is also important. In the early days, boards give executives plenty of leeway when it comes to variance on revenue, so long as the assumptions behind different revenue projections are clear; it’s harder to understand when costs vary from plan.We’ve found it useful to set up revenue gates for incremental spending, so that we don’t have to re-factor the whole plan as revenues move up or down. A long time ago, we also measured all of our costs compared to plan, and listed the dollar-amounts we spent per employee on health care, equipment, rent and the like. This might be useful for others:…One reason reason we wrote this down was for our own future reference. I sometimes forget to read the old plan when I write the new plan.

    1. fredwilson

      I like the revenue gates approach, particulalry as you are transitioning from revenue focus to profit focus

  20. markslater

    So the 1 billion dollar question:what happens when you are building a business with no ‘revenue model’ but you are replacing revenue with users, or some other growth metric? kinda pointless to do projections except on the burn side.

    1. fredwilson

      Burn is critical. ‘Cash is more important than your mother’

  21. markslater

    So the 1 billion dollar question:what happens when you are building a business with no ‘revenue model’ but you are replacing revenue with users, or some other growth metric? kinda pointless to do projections except on the burn side.

  22. HowieG

    My last corporate job was sales for a company that made things that did not exist yet (highly engineered products) for Aerospace, Military, Automotive, and Medical clients. This is called a FMEA or a DFMEA. The first is taking a product (or service) and figuring out everything that can go wrong so that the end user knows 1] what to do themselves/or will happen with failure 2] know their vendor can help. The second is actually designing something specifically to reduce failure potential. Automotive uses the first. This ensures a part fails in the same controlled way everytime (I was working on Hydrogen Fuel Cell Fuel and Storage Systems for Autos. All tanks have to rupture in a crash the same way). For Aerospace they do the second. A part needs to work once and it MUST work or a space ship or rocket or satellite or mission worth possibly a billion dollars might blow up. Lockheed installed a device of mine backwards and had a NASA probe crash to the ground at 200mph. I was interrogated (as my companies representative because I was the west coast rep and it was at JPL) by a US Attorney and a NASA Inspector General who wanted to see if they could go after my customer for Negligence. That was not fun. So the start up I was with before my current company was started by two kids from France (22 and 23yr olds). They freaked when I said I need a FMEA for your product. You want me to solicit orders worth $300k for an Advertising Service that has street teams in 250 locations that has never been proven from national brands and media clients. I need to know what could go wrong and prove you have a plan to ensure our service goes off without a hitch.VC’s want from their investments and Business Owners as sound operating procedures need to have these contingency plans and contingencies (for all parts of your business) identified as part of strategic planning.

  23. David

    Great info, I’m now going back and reading your past posts on the series (and all those great links to related posts) and I’ll be catching up on those to date. Great timing for me; I’m neck deep in projections for a series B right now!!! Thanks again.