Wednesday, September 07, 2005

The Best Startup Advice I Have

When are company forecasts reliable, and when is it time, based on those forecasts, to ramp up the sales force? Investors and entrepreneurs face these two critical questions, respectively, in every venture-backed company. The wrong answer will either blow a startup's precious capital, or miss the opportunity.

So at Bessemer, when we assess investments and direct their pace of investment, we always keep in mind the Sales Learning Curve (SLC). The SLC is the invention of Stanford Business Professor Mark Leslie, who also happened to have founded and led two Bessemer portfolio companies, one of which is dead and forgotten, and other of which is Veritas! Prior to Leslie's formal description of the SLC, VC's like us just stumbled along, making clumsy judgments about the quality of forecasts and a company's readiness to "ramp it up" based upon the patterns we could glean from our own anecdotal data.

Leslie describes the SLC in a paper (pending publication) titled something like (it's been a while since I've read it) "It Always Takes Longer and Costs More." This paper attempts to explain why it is that Silicon Valley, which is so good and experienced at inventing new technology and building new companies, can't ever seem to craft business plans that track reality.

Leslie's approach springs from the observation that factories would never presume to produce widgets in volume until they have brought the cost per widget down to a profitable level through a commonly recognized manufacturing learning curve. Only once you start manufacturing something can you identify hurdles and opportunities, ultimately learning how to reduce cost, improve quality, and raise yields.

Likewise, technology companies shouldn't expect to sell their 1.0 product without undergoing a learning curve around the needed feature set, the competitive response, the right type of channel and sales rep, optimal pricing, etc. And yet, nearly every business plan I have seen shifts gears directly from Development to Sales, staffed by a growing number of experienced sales reps charged to sell their regular quota of product.

The consequence of skipping the Sales Learning Curve in a business plan is all too common: the company fails to meet the plan, sales reps are fired, eventually the VP Sales is fired, and the company has to raise a highly dilutive down round of capital to stay in business.

Leslie's prescription for success and capital efficiency is a plan that includes the Sales Learning Curve. Accept that until we're out there selling, we can't know what we don't know about the selling process. The goal at this point is to maximize runway, since you can't rush science. As my erudite partner and Harvard Business School Professor Felda Hardymon likes to instruct, "run the business like a one story whorehouse" (with no fucking overhead). Hire only two or three creative, guerilla-style reps and a VP who knows how to experiment with multiple channels.

It may take 4, 7, or 10 quarters to climb the curve. Only then, when your sales force is a profit center (with two quarters where contribution exceeds twice the sales costs) is it time to flick the switch. At this point, raise lots of capital and hire as many talented reps as you can possibly find! This kind of prescriptive framework for clearly distinguishing the phases of a company is so much more actionable than the wishy-washy approach of hesitantly adding more and more sales reps over time.

The SLC is least helpful in markets like telecom carrier equipment, in which it's rather straight-forward to determine the needs and buying habits of the customer. That's why these companies tend to accrue a lot of value after winning just one or two large customers.

But for enterprise-focused startups, the SLC is critical. No amount of up-front due diligence can yield the clarity that comes from actual field sales. Enterprise companies are particularly vulnerable to the delusion that they have cracked the code, thanks to some early wins. But without a proven, profitable sales force, it is reckless to ramp up the sales and marketing expenses.

The extreme value of the SLC is seen in consumer markets. Focus groups or not, you just never know. That's why, as I discussed in this earlier post, we try not to fund User Behavior Risk.

Further reading: Professor Leslie's slides on a case study of applying the SLC to the Nano-Optical Customer-Adaptive Software/Hardware (NOCASH) company.


  1. Anonymous10:14 AM

    This makes total sense. It is how best known practices get developed for new industries. And it always takes a few tries before getting it right. Glad to see that Bessemer is clued in :=).

  2. I see that you linked to Mark's presentation that he gave at the CEO briefing Altus Alliance hosted where Mark Leslie presented the Enterprise Sales Learning Curve that was very well received. Mark's whitepaper & presentation are posted here The paper is well worth reading.

  3. Anonymous3:44 PM

    Funny. You link to a 37 page powerpoint to what most good business people would think of as intuitive and obvious!

    Can you link to my 51 page powerpoint that proves that shit stinks?

  4. Anonymous: everything is obvious once it is demonstrated.

    It must be very tiresome for you to be so much more intelligent than the rest of us. Building successful companies must come very easy for you. Please be charitable with your billions!

  5. Anonymous5:31 PM

    As an entrepreneur, I find this site very interesting... but most interesting is the little cat scratches! meeee-ow! You go, girls!

  6. Anonymous4:42 AM

    some people learn from reading,
    some learn from watching,
    and some people just have to piss on the electric fence to get it.

  7. Anonymous9:50 PM

    I had the glorified title of National Sales Manager (sales trainer) in a startup that made just that mistake.

    We had a product folks loved just out of alpha testing. We hired sales teams and trained reps, and then had to fire them.

    By then we had burned through our owner's capital.

    I fired the owner, but it was me that took the walk.

    Good experience, bad results.

    Thanks for sharing the warning.

  8. Anonymous2:30 PM

    I have just distributed the url to your site to a number of friends who will appreciate SLC as I do. Thanks for your generosity.

  9. Couldn't agree with the SLC more - it's all about "learn before you burn."

    It seems that engineering teams have figured out short learning cycles and rapid iteration ("Extreme Programming"), but the rest of the company functions have yet to catch up.

    For a good read on learning about your customers before you crank things up to fast, check out the self-published book by my former boss at E.piphany, Steve Blank

  10. Anonymous1:17 PM

    few enterpreneurs can fend off the pressure from VC's to "scale," i.e., grow the sales team, and hence-ahem-sales. some vcs hire "experienced" CEO's to come in and scale it up (shd i say screw it up?)

    thanks to Mark Leslie by offering entrepreneurs a tool push back and trust their instincts.

  11. I think the SLC concept is very interesting. Selling to Enterprises can sometimes be a very slow Sales Cycle.

    I found the comment: "If there are only one or two vendors, such as major telecom vendors, then the SLC concept is less useful"

    In this situation- how do you assess the risk that your customer will not develop a similar product?

  12. John Spiers4:39 PM

    I get it and agree, but what if you just sold your last company that was a huge success and start a new company with the same/experienced people addressing the same target market, channel and customer base with a product that fixes all the things that customers complained to you about at your last company + some new game changer stuff all at a competitive price. I’m thinking the SLC should be much shorter.

  13. Hi David,

    Since you are a singer, I wanted to ask if you ever do seed investing in music technologies? I have a prototype (early stage), and am looking for capital to help with development costs.

  14. I did invest in the seed round of Smule, which makes the Sing! app. But I think that is one enough!

  15. Nice read, and helpful to see some possible realities for the uphill climb. We are about to launch a biometrics platform using reaction biometrics to help users find matches for dates and friendships.