Swinging For The Fences

A number of our portfolio companies use scoring systems to prioritize projects on their roadmaps. I like this one called RICE (reach, impact, confidence, effort). There are similar versions of this out there. The important thing about these systems is that you need to include probability of success in your analysis. Just looking at effort and impact isn’t enough.

But one thing I’ve seen about these scoring systems is that they can lead your team to do lots of low and medium impact things that have a very high probability of working.

I generally would like to see our portfolio companies taking at least one or two big swings a year. These are high impact, often high effort, and often low confidence. They don’t score that well as a result.

But if you aren’t going for it with at least some of your resources, you can get mired in a rut of small wins and that can be a problem for your growth trajectory, morale, and overall business mojo.

So while I like it when our portfolio companies use scoring systems, I generally suggest they add a requirement to take at least one big swing a year. Over the course of several years, they will get at least one of these right and it can make a huge impact on the business.


Comments (Archived):

  1. jason wright

    sWINging for the fences

    1. William Mougayar

      good one.

  2. William Mougayar

    Probability of success is a primordial factor. We used it in sales forecasting and in long term planning in the area of risk.It’s also important to understand the factors that could change the % probability so you can assess if they are in your control (internal) or not (external).

    1. DJL

      I am seeing some companies use coin offerings as a “swing for the fences” approach to saving their business.

      1. William Mougayar

        haha…touché!!! and I am seeing them too. :)Tokenize or Die !

    2. awaldstein

      When you use a crm system does the system tell you the probability of success to target an account or visa versa?Analytics in sales, especially enterprise sales are set by the human factor and skill.You seem to imply the later.

      1. William Mougayar

        “When you use a crm system does the system tell you the probability of success to target an account or visa versa?”- not in the targeting, but in the “sales forecasting” as I said. that’s standard in sales forecasting: Dollar amount x % close = Forecast. maybe you misunderstood.

        1. awaldstein

          yup seems like it.sorry my friend.sales forecasting and sales are distinct processes obviously.

    3. Richard

      This field is developing fast. There are tools (AMPL) for optimization where you can assign a probability distribution vs a discrete distribution.

      1. sigmaalgebra

        AMPL is old and, really, just a way to make it easier to enter problems to the integer linear programming software such as C-PLEX or GuRoBi where the real challenge is.A lot of solid applied math is known for how to do the best in some really complicated, uncertain situations, but commonly need more data, e.g., on probabilities, and computer resources than have available.

  3. Tom Labus

    You always need a couple of Aaron Judges in the lineup. If you don’t want home runs then those who can hit them will go elsewhere!

    1. creative group

      Tom Labus:You need the team to get accustomed to winning. The Aaron Judges still don’t have the experience but having a winning DNA that can’t be taught…Your teams future need to understand home runs are not the only way to win in business. Sometimes doubles will have you receive just as much adulation and wins.

      1. JLM

        .The most elite military units expect to win — every time and under any conditions.Their leaders condition them to expect to be able to find the path to victory even when seemingly outnumbered and overwhelmed.The ability to punch above one’s weight class is a real thing.When leaders in business inculcate their organizations with an expectation of victory/success, the power of that bias absolutely impacts performance.The first duty of effective leaders is to set the tone and to set expectations. Even when expectations are not met, the gap from high expectations v lower expectations elevates performance.A leader has to look, act, inspire, motivate — like a winner.JLMwww.themusingsofthebigredca…

        1. Tom Labus

          People know this instantly and instinctively.

      2. Tom Labus

        Yes, it helps if there’s a few guys on when that ball goes in the left field seats.

  4. JamesHRH

    Bet the company no longer in vogue?

  5. Moritz

    We use the V2MOM system for individual and company scoring here at Neufund. Anybody else got experience with that?It’s a nice exercise, but I personally haven’t derived much value from it yet…

  6. LIAD

    no one ever got fired for buying IBM.seems not swinging for the fences is innovators dilemma with another name

    1. PhilipSugar

      You know they are now. I can give you some examples privately if you wish.

  7. JimHirshfield

    Generally the initial product is swinging for the fences…otherwise you wouldn’t be in it.And then…Years later…Twitter announces bookmarks.

  8. Sandro

    I like the RICE framework a lot although I add a D at the end for cost to Delay as sometimes things are important but there’s not much difference between launching it next month or the month after.As for the big swings, I always like to get those top-down from the exec/leadership teams and/or board. They’re usually longer polling initiatives which require deeper thinking and planning.

  9. TeddyBeingTeddy

    If they truly ran probability trees, they would shut the startup down and get a corporate job. But you’ll never hit a homerun unless you swing the bat. (Though getting singles and doubles wins a lot of games too…)

  10. Mike Longo

    What does a “big swing” look like in practice? Would be great to see a post outlining a case of a successful vs. an unsuccessful “big swing.”

    1. Tom Labus

      These are your best people. If they swing for the fences and fly out you don’t go after them and blame them for other stuff. Those people need to know this without it being said

    2. sigmaalgebra

      What does a “big swing” look like in practice? Okay, since you asked, there’s a really easy answer:http://iliketowastemytime.c…For more, GPS. More, WMAP. For more, LHC. For more, the F-22: It can fly higher and faster than the other plane can, and neither its radar nor its pilot can see it:https://upload.wikimedia.or…Failed projects? Apparently EEStor (high energy and power capacitor based on barium titanate). Apparently Joule Unlimited (gasoline directly from sunlight, water, CO2, and some special microorganisms).

  11. DJL

    What you suggest is essentially a “growth” based portfolio management of projects. Put X percent of your investment in more traditional, less risky time investments with a predictable return, and then put some fraction (20%) X in high-risk/high-return projects. And then you re-balance based on your business situation each year. Nice.

    1. TeddyBeingTeddy

      Why not do like the best wealth managers do and only put 5-10% into high risk equity (projects)?

      1. DJL

        Certainly. Depending on how risky your company needs to be.

  12. JaredMermey

    How does this change over the lifestyle blog the company? As the company becomes more established, customer expectations shift (generally higher as well as sticky, emotional tie to we existing products), it’s product becomes heavier and it has more resources.

  13. Brian Manning

    Good post. Important to do this but also super important to know when to quit…I’ve seen execs chase the next big thing for too long and drain lots of resources because they take too long to admit it was a bad idea.https://www.amazon.com/Dip-

    1. PhilipSugar

      REALLY good point. Pet projects or worrying about sunk costs, or the worst: scared of admitting failure and moving on are a huge drain.

      1. LE

        Same with both investing and/or selling assets. Till the bitter end. Good money after bad.But even with success ‘when to sell’ is somewhat similar.An example of this is when someone is considering purchasing an asset and will then compute the holding cost of the asset by using, say, the rate they would have to pay to borrow the money or the opportunity cost of the money if they are paying cash. (Perhaps simply assigning the borrowing costs if a cash purchase). However a person selling who has an asset that has increased greatly in value (say real estate) over time will almost never compute the same numbers. They will just think ‘wow I have made money this has gone up in price!’. So if you own an apartment in Manhattan that you paid $100,000 for that is now worth $2,000,000 you don’t think of the money you are losing by not investing the gain (net of taxes of course). You just continue to hold. (Of course some sell…)Now let’s take the hypothetical example of bitcoin purchase. Someone has purchased 100 bitcoin at $50 for a total cost of $5000. Well today bitcoin is close to $5k again ($4800 as of right now). So that investment is now worth in theory, if they sold, close to $500,000. What else could they do with that money? Would the same person borrow or use existing $500,000 to buy it at that price? Most likely not. Because they would consider carrying and/or opportunity costs. So they hold onto it because in their mind there is no cost to holding unlike buying. Plus they can think ‘I am an investor’. Holding money in a CD doesn’t have the same psychic benefit.Separately it is somewhat well know in negotiating that the more time you waste with someone will sometimes lead to getting a deal done simply because seller is less likely to quit and give up when there are sunk costs and admitting failure and so on. (Of course there is also ‘time kills all deals’ to content with as well). [1][1] The example I use is car buying. Better chance of getting the price you want if you waste the salesman’s time than if you just walk in cold and offer the same price which is easy and quick to reject because there are no sunk time costs.

      2. sigmaalgebra

        Simple: Those are just people doing projects for results that are likely both new and significant but just flatly don’t know how.”Know when to quit”? Of course need to know when to quit, but for a well run project will know when to quit because will have some very clear evidence. If the project just drags on and on and on with doubt about when to quit, then it was badly managed to begin with.Also for the risky, speculative, highly uncertain parts of a project, those should all be handled in the early parts of the project where the costs are so low no one much cares.All this should work for the technical parts, that is, information technology. But if a project is to be a swing for the fences, then it should have at its crucial core some rock solid technology and that so powerful that it removes nearly all risk from the rest of the project. The great example would be one pill, safe, effective, and cheap, taken just once to cure any cancer. Do that well, and the rest of the project, shape of the pills, color of the label, PR effort, etc. will all be low risk.Only a tiny fraction of the population, only a similar fraction of the people in the computer industry, know at all well how to do projects that are both new and significant. And for good project success, one of the keys is good project selection at the beginning, and not many people know how the heck to do this.It’s possible to fly at 80,000 feet at Mach 3 for 2000 miles without refueling safely, routinely, but it’s not going out on “a wing and a prayer”. Instead such successful shots over the fences are based on good project management with good, new technical ideas with good applied math, physical science, engineering, and execution. Sorry ’bout that!

    2. Rob Larson

      Agreed. The challenge is that it can be hard to figure out when to quit. Some big opportunities can take much longer (and more $) than expected, but if you don’t quit, the payoff when you finally crack the problem is huge. So how to know when you’re chasing an opportunity of that kind vs. quixotically chasing a bad idea? It’s hard to know. (though I think Clay Christensen’s “Jobs to be Done” framework can help a bit)That said, the deeper your available resources, the more rope you can afford to keep giving executives to pursue these. Jeff Bezos answers that question this way: for as long as he has at least one high-caliber executive who believes in the quest and is willing to pursue it with full enthusiasm & effort, Jeff will continue to fund the quest. When the last “true believer” gives up on the idea, the funding is cut. Of course, Amazon can afford to absorb a lot more mistakes than typical startups can, so anyone reading this should probably be willing to cut bait sooner than Jeff would.

      1. Brian Manning

        Love the Amazon example. For a startup that approach could be deadly.

  14. PhilipSugar

    I really like it.And employees like it too.If everyone just works on the safe project it gets really boring.If 20% of the time, and that doesn’t mean one day a week it means you get assigned a project you work on 100% of the time for 20% of your year, or once every quarter or whatever, you get to do some blue sky project that is interesting.

  15. Salt Shaker

    Company lifestage and risk assessment aren’t mutually exclusive. If you’re resource constrained and on an early growth path, then “swinging for the fences” may not be so prudent. There’s a much higher prob of success having a Mike Trout or Jose Altuve in your lineup early stage than a free swinger. There’s an if/when aspect to when you can and should take those deep swings w/ a 40 oz. bat. There’s an opportunity cost w/ any endeavor, while managing risk is hardly an exact science.

    1. creative group

      Salt Shaker:there was a time before your team acknowledges you may have a Mike Trout or Jose Altuve.When Aaron Judge went into his hitting slump people questioned all the adulation until his form returned.The talent scouting is done before the hire. Trust the process and results will follow. Adjust according.The experienced Operational Managers use that experience to grow positively. If you need to question your process it is the wrong process.There is no room for emotions. The business needs to be nimble and adjust with the customer needs.

  16. Rob Underwood

    In 2007 when I was still w/ Deloitte a couple Deloitte partners and I pitched an internal futures market to Google that I had come up with for selecting and prioritizing roadmap projects. Google told us it was too radical, too out there. I still chuckle about Google telling Deloitte that, especially in 2007.

  17. JLM

    .”you can get mired in a rut of small wins”Sounds Hellish.There are a lot of grand things built on a series of small wins.JLMwww.themusingsofthebigredca…

    1. PhilipSugar

      That is why 80% of your people better be focused on that goal as a technology company. And you cannot just have one group that gets to have “the fun” that needs to rotate.

    2. Lawrence Brass

      I guess that is how Nature and evolution works. We are and live within a chain of tiny and successful experiments.

      1. JLM

        .And a fair number of unsuccessful ones?JLMwww.themusingsofthebigredca…

        1. Lawrence Brass

          Yes. The unsuccessful ones were part of the process but they aren’t part of the virtuous chain. It is cruel.

        2. Lawrence Brass

          Also, failure leaves a lot of useful data behind.

        3. jason wright

          we are the 1% Club

        1. Lawrence Brass

          We are all very lucky and successful for being here today, cousin jason.. :)https://uploads.disquscdn.c…

    3. cavepainting

      Any scoring system will be gamed. It is important to use them only for the type of projects where it is appropriate and at the right stage.There are three different kinds of roadmap projects: incremental, transformative, and new. The incremental improves current customer experience, the transformative dramatically changes the experience relative to status quo and competition, and new refers to solving problems for markets different than the core through new products.Incremental can be planned well because RICE is very well understood. Transformative is hard because there are many unknowns. Needs compelling vision and clarity from the founding team. Hard to quantify effort and confidence in the early days.New projects are adventures beyond the current market. Challenge these through first principles and treat them as startups in their own right.The transformative and new projects need to be seen as a portfolio of bets that are constantly monitored and assessed. Depending on faith and progress, they either get more capital or canceled. If done well and in the spirit of searching for truth (without ego and attachment), the company will eventually iterate to the bets with the highest potential and probability.

      1. JLM

        .Good thoughts.When things are “hard” to do is exactly when they need to be done, particularly if it is hard to develop the data.When things are “easy” to do, instinct takes over and the work is not as critical. Instincts built on experience can be powerful.JLMwww.themusingsofthebigredca…

  18. JLM

    .Any system of making decisions is better than no system.My math spider sense tells me that a 4-variable differential equation is a little too chainsawish as one variable can have too big an impact.The key is to use it consistently and then circle back and see how the results compare to the analysis.”results may vary”The old fashioned “after action report” is a great check on the effectiveness of any system.JLMwww.themusingsofthebigredca…

    1. PhilipSugar

      If you do this you have to be able to make decisions that you were wrong.

      1. JLM

        .One of the first things I do with the CEOs I work with is to ask them, “What percentage of your decisions turn out to be either correct or the results are good?”The really good ones, “About 30-40%.”The not so good ones, “90%.”The challenge is not usually how good are your good decisions, but how good are your bad decisions?JLMwww.themusingsofthebigredca…

        1. JaredMermey

          Just need to be right 53% of the time to beat the vig.

        2. LE

          The not so good ones, “90%.”An example of this is my credit checking procedures in the 80’s at my first company. Had a hard time extending credit unless it appeared someone was truly credit worth as a business. D&B, call references (often pre selected by the potential buyer because they gave the right answers) but in the end you got a good idea if you should extend credit. Or not.But here is the thing (to your point ‘90%’). In the end the fact that there was so little bad debt at the end of the year indicated to me that we weren’t taking on enough risk and were losing business. That we needed to loosen up because the increase in sales from more risky prospects would more than take care of the losses we incurred from easier lending. [1][1] There is more to it than this of course. The easiest accounts to get were the ones who had stiffed others so it is more difficult than it appears on the surface. Plus nothing like staying up at night worrying about money that is owned by a deadbeat and it’s right off the bottom line if not paid.

    2. sigmaalgebra

      Yup, an initial value problem for a one variable, first order, linear, ordinary differential equation.Impact: One Saturday afternoon in Memphis, kept two representatives of BoD Member General Dynamics from leaving FedEx. Their leaving would have killed FedEx.The equation?y'(t) = k y(t) (b – y(t))Assume have y(0), k, and b and want y(t).Derive the equation? Let t denote time, say, in days. Do some revenue projections. Then t = 0 is the current time, and y(0) is the current daily revenue. Assume that know b, the daily revenue of the business at its full potential.Assume that at each time t the rate of growth in the business, d/dt y(t) = y'(t), is proportional to (1) the number of current customers talking about the business, the y(t) factor, and (2) the number, the (b – y(t)) factor, of target customers not yet served. Then k is the constant of proportionality and needs some data or a guesstimate.Then, yes, there is a closed form solution. Don’t even need a course in differential equations, and just the first year of calculus is sufficient.I’ll leave the solution as an exercise!Sure, the solution is a lazy S curve that starts growing slowly like an exponential, soon starts growing rapidly as an exponential, then slows its growth as it approaches b from below (runs out of new customers). So, it’s a first-cut, simple model of viral growth, viral in the sense of growth from word of mouth advertising, that is, current customers talking to new customers.It’s crude and simple minded with only two numbers of data, the y(0) and b. But in some situations, like that at FedEx, is much better than anything anyone else had. And, first cut, eyeball, it looks a lot like some classic product growth scenarios.

  19. Collin

    I’m a big fan of using a variation of the 80/20 rule. Use 20% of the resources you’re allocating to growth for a big swing that could deliver 80% of all results. It can be adjusted as you see fit.

  20. creative group

    FRED:”I generally would like to see our portfolio companies taking at least one or two big swings a year. These are high impact, often high effort, and often low confidence. They don’t score that well as a result.” – FredAre you communicating this directly to the USV portfolio companies, this is the communication, you allow the portfolio companies the autonomy unless your on their BOD? Your process of providing advice with the probability of low success rate in this instance.

  21. rfreeborn

    Here’s a methodology that we created a few years ago to prioritize the development of different SW features. After looking at a ton of different options, this simple one seemed to fit the bill the best, feel free to use as needed.https://docs.google.com/spr

  22. Ric Merrifield

    Swinging for the fences can be a macro company-wide effort, and it can be departmental – marketing could be swinging for the fences this month for something that doesn’t touch R&D. As long as everyone has a stretch goal at least episodically that touches their own accountabilities – that helps everyone understand their contribution to the whole. It also depends on the team – when we did the Disney MagicBand – every month was swinging for the fences for about a year and that was tough to sustain, but because we had an amazing team – we pulled it off.

  23. Steve

    Add a metric that there must be at least 5-10% chance of having a game changer amongst the projects (could be multiple <5% ones)Maybe that 10% of total effort on high risk?All about balance in the KPIs / planning

  24. karen_e

    Fantastic post. Such a typical nugget in the sense that it can only be formed through the crucible (isn’t that a JLM word?) of 30 years of experience.

    1. JLM

      .Think that was Jerry Colonna’s word. I rented it. My favorite word today is legerdemain, but watch out cause I’m tricking y’all.JLMwww.themusingsofthebigredca…

      1. karen_e

        But can you pronounce legerdemain?

        1. JLM

          .In two different languages.JLMwww.themusingsofthebigredca…

  25. Mike Zamansky

    Nothing specific to add but wanted to say that I found this post particularly inspirational.

  26. CThomps

    I’m a fan of putting projects into frontiers – there should be something in the “third frontier” which fits the description of the post here. We like breaking the outlier/big swing projects into a small ‘minimum viable tests’ with KPIs they must achieve to sanity check them, which creates a stage-gate the continued resource dedication to them or direct it elsewhere, to either sobering blocking & tackling or to other exciting outlier “3rd frontier” projects.I think even for the big swings, early data points that are very unfavorable very rarely make a dramatic comeback, unless the market changes. This is particularly true for demand tests for consumers or B2B if this project is going to solve a legit $ funded problem. If you put a minimum viable test in front of them and there is low/zero interest, it’s unlikely to recover in that same form. This is where teams want to put their head in the sand and keep doing the fun part of development and ignore the dramatically low demand for the minimum test that has been done so far. Should at least engage with that bearish data point and take a hard look at the project’s potential and note why the data point is being masked or cast aside, if it is.

  27. sigmaalgebra

    Under some fairly well understood circumstances, it is possible to “swing for the fences” with high probability of being successful. For projects based mostly on technology, the history of US national security for the past 70+ years is awash in examples, sure, e.g., the SR-71 — a big swing for the fences but once Kelly Johnson’s design work was done low risk and high payoff.> But if you aren’t going for it with at least some of your resources, you can get mired in a rut of small wins and that can be a problem for your growth trajectory, morale, and overall business mojo.Generally employees are darned reluctant to do such things because the consequences of anything like failure can ruin their job and maybe hurt their careers.And among both the low level worker bees and their management, not very many people know how to do such projects at all well. Evidence: In the STEM fields, for nearly all grad students, the research for a Ph.D. is very challenging. Teaching how to do research projects — or anything both significant and new — is tough, and not many people are good at it.I’d suggest, for big successes, first, quietly, where little or no blame could be assigned, do the planning very, very well. For an information technology project, to be more sure the planning is solid, the crucial core of it should be some applied math complete with rock solid references and, often, some math derivations or even some theorems and proofs. Want this stuff rock solid, iron clad. The work product should be in a well written paper done with D. Knuth’s TeX. Most organizations will have to learn TeX, and that will take some time. If the organization isn’t comfortable with some high end theorems and proofs, and that basically requires a good ugrad pure math major, then maybe they should focus on first base instead of the fences.Then, also quietly, do some prototype work, say, small data, small objectives. See how that goes.Only when things look quite good and low risk, swing for the fences.Yes, definitely DO swing for the fences, but do it safely, for careers, and effectively for the company.

  28. John Collins

    Hi Fred,Thanks for sharing our RICE framework. We share your concerns about focusing on low and medium impact work. One of our co-founders, Des Traynor, wrote a post about the importance of avoiding snacking, which you might find interesting: https://blog.intercom.com/f

    1. fredwilson

      That is great. Thanks

  29. James Ferguson @kWIQly

    I think all entrepreneurs can do themselves a favour by reading and internalising the project finance sections ofPrinciples Of Corporate Financeby Richard A. Brealey, Stewart C. Myershttps://en.wikipedia.org/wi…The argument for NPV is inarguable – none-the-less the tough skill is to abstract that from Gut Feel.I think these frameworks for thought assist in that – but it is important not to ignore the underlying rationales – eg principles of sunk costs.Yes its heavy going – but its a keeper,