Entrepreneurs & the Scientific Method

Years ago Bill Sahlman, professor of entrepreneurship at the Harvard Business School, told me that “Startups are a succession of small experiments.” I’ve never forgotten those words of wisdom and have passed them on to many mentees. Of course, the mentees at MIT, all being engineers or scientists, do not need to have the scientific method explained to them, but it’s still good to see how it is applied to business. I was pleased to see Reed Sturtevant of Project 11 make this same point, very eloquently at a meeting of MIT Sandbox mentors and student entrepreneurs yesterday. Here’s my version:

  1. Make your hypothesis – per Professor Sahlman – make this something small that can be tested quickly.
  2. Design the experiment that will test your hypothesis. You don’t need to write a lot of code to do this. You can mock things up on paper and do an A/B test of a design element, for example. Or give prospects a choice of a purchase model: subscription vs. a la carte.
  3. Gather your data. Needs to be quantitative. Where qualitative judgements are involved, like in a design, get your test subjects to use a Likert scale or another method of turning qualitative into data.
  4. Analyze your data.
  5. Present your data and analysis to the decision makers on your team.
  6. Come to a conclusion: did you prove or disprove your hypothesis? Do you need to runs more or different experiments? What are the next steps?

Obviously you need a product roadmap and your experiments need to fit into that road map and into your big hypothesis: “Customers will love our product because it helps them do X far better than their current solutions, which will enable us to build a successful, profitable company.”

B2C – build in virality or bust


My experience  with consumer -facing products is limited to iShop, a mobile shopping app developed by Smartworlds in the early 2000’s, and Endorfyn, from PopSleuth, Inc., a web app designed to keep fans up to date on new releases from their favorite artists, developed three years ago.

iShop never got any further than a very positive article, with accompanying cartoon, in
The Wall Street Journal, and Endorfyn, while nicely executing on its mission, remains a niche product.

So what have I learned from these experiences, as well as advising and mentoring various B2C companies over the years? One thing: the cost of customer acquisition is very high, close to astronomical, which is one reason why you see companies like Uber raising billions of dollars; the second is is a correlative of the first: you must build virality into your product from from the get-go. It can not be pasted on later and no amount of Tweeting or SnapChatting can make up for lack of built in virality.

So what do I mean by virality? It’s simply the oldest and best method of marketing, word of mouth, digitized, and its effect amplified by several orders of magnitude by the power of social media marketing, led by Facebook, a colossus with well over a billion connected users the world over.

Connection is the secret sauce of virality. We never designed Endorfyn with virality built it. It’s an anonymous service, there’s no direct way to share your favorite finding on Endorfyn with your friends, no way to import your address book, no way to Like a finding and share that across your network, and many other ways we could have, but did not, built in virality. I had originally viewed Endorfyn as a utility, much like Evernote and like Evernote, it totally lacks viral features.

Another key aspect of virality is sharing. If you think about old fashioned word of mouth what does it mean? Sharing your opinion with other people. The huge successes in social media: Facebook, Twitter, SnapChat, Instagram, et al all have built in sharing and highlight and reinforce that feature.

The third key factor in virality is following – the ability to get notifications on new posts, comments, likes, ratings or whatever from people you care about – from friends and relatives to celebrities. Following is a key feature of Twitter and Instagram, for example.

One of my visions for Endorfyn was that we would get the artists themselves to use it, and thus people would sign up to Endorfyn just to find out what their favorite musicians were listening to, what their favorite authors were reading, what films their favorite actors and directors were watching, etc.

Notifications are an ancillary, but important part of virality. They make sharing, recommendations, and other viral features frictionless. Instead of having to hunt down the latest photo from a friend, you can opt-in to being notified on your smartphone when that friend posts a new photo.

A fourth component of virality is personal brand building. YouTube has made stars of many of its users, as it has brilliantly facilitated their ability to build their personal reputation or brand. This creates a virtual  circle, where the  more attention a YouTuber gets, the more video they create, the more views and followers they accumulate.

A fifth component is ratings. Facebook has made “Like” ubiquitous. eBay makes its users far more comfortable with doing business with strangers through its buyer and seller rating system. Uber has copied that feature: riders rate drivers and vice versa.

One of the most powerful drivers of virality, utilized by virtually all the successful social networks, is user contributed content. The content cost is virtually zero and contributors are motivated to share their photos, blog posts, videos, and other contributed content with their friends and followers. YouTube and Instagram would cease to exist if they shut down UCC. UCC is their growth engine and also enables personal brand building.

Recommendations are another driver of virality as again it is is digitization of typical consumer behavior. When I discover a great restaurant or watch a great movie I naturally want to recommend it to my friends. And vice versa. Given the ever growing plethora of media, finding stuff you like – which is the mission of Endorfyn – gets harder and harder. Yet we didn’t have the resources to add this viral component to the app.

Identity is vital to virality. Endorfyn requires a login and password, but like Evernote, your identity is a secret. There’s no way you can discover friends who are using Endorfyn and share your likes about favorite artists, their new releases or news about them. Whether that identity is “real” like on Facebook, or created by you like on Reddit, having a known persona is critical to enable all the others aspects of virality, from sharing to ratings.

Finally the important of all forms of virality – and one you have no direct control over -is the network effect. Metcalfe’s law states that the  value of a network equals the square of the number of its users. The law was originally applied to telecommunications systems and “nodes”, but has since been found to apply to social networks. By building in virality you create the opportunity to benefit by the network effect. And the network effect drives growth more effectively than any marketing technique ever created, and growth is the number one goal of all startups.

I’m far from an expert on either consumer customer acquisition or social media, so I’m sure there are other ways to build in virality. If you’re not a an expert on both those subjects it may be very worth your while to consult experts before you build your product.

Because if you get anything out of this post, it’s that virality must be built in from the get go. You can progressively add viral features, but you’ll never acquire enough users to get that far without going viral from the start.

Stages of a company

As MIT mentors we usually see very early stage companies, in contrast to the mature organizations at the Social Innovation Forum. Here’s a rough guide to a startup’s life stages. In startups many things go on in parallel, not serially – such as product development and customer development. So this list is fluid, your MVP might get you paying customers. Companies have been bought before even releasing an MVP! So keep in mind that this is just a list of phases and the order may well not apply to you as listed.

The Business idea

Typically ideas are either how to solve a problem or how to create an opportunity. Google solved the problem of finding stuff you were looking for on the ever growing Internet, quickly and accurately. Facebook created the opportunity to connect with friends, former classmates, acquaintances, friends of friends and keep up on their activities.

As Bill Gates and others have said, “ideas are cheap, it’s all about execution.” I tend to agree with that, though occasionally I do see an idea that looks valuable and the execution straightforward.  The canonical example of this was Hotmail, the idea being creating a
web-based email system. The founders knew they had a hot idea, kept in total stealth mode, and coded like madmen to get to market before anyone else. Evidently in a matter of days it was up, went viral and then was bought by Microsoft for $4oo million if I recall correctly.

Entrepreneurs tend to worry too much about others stealing their ideas and far too little about finding customers. While big companies do steal ideas, it’s much more common that they wait until the idea becomes successful, then build it into their products. The classic example of this are operating system vendors like Microsoft and Apple that spot useful utilities and then build them into their OS.

So get out and develop customers and worry less about your protecting idea, unless you’ve truly got an invention, in which case you should probably consult a patent attorney.

The Presentation and Executive Summary

While you are developing your product, and I hope testing the idea with potential customers to validate your idea, it’s common practice to produce a slide deck and a
one-page executive summary. I’m not a fan of decks – people have seen far too many of them, most cram too much information into every slide – and take up valuable time, time better used to create products and customers. What I do recommend, and I just did this myself, is to create a one-page process flow diagram of how your business works and how it delivers benefits and makes money. You can print this out on a extra large piece of paper at Staples or elsewhere, then talk you way through the diagram with your prospective customer, partner or whomever is sitting beside you. I promise they won’t fall asleep while you do this as they might during your slide show.

I do recommend a leave-behind one-page executive summary with the standard elements of a business plan. It should be terse, clear, and preferably illustrated. The exec sum and flow diagram can serve as pass alongs within your target company, acting as information emissaries on your behalf. Your goal is to arm anyone who likes your product with the tools to champion it within their organization, as your initial contact will rarely be the decision maker.

The Demo

I have a simple definition of a demo: it’s a simulation of your product that usually requires you to present it, as it will have bugs, missing features etc. that only you or one of your colleagues know how to navigate around. The sooner you can show, rather than tell, how you solve a customer problem or create a great opportunity the better. Video demos have pros and cons. The pro is you know there won’t be any glitches in the demo, the con is that the audience will probably assumed you just did it all through clever editing. But if you have a complex and dynamic process to demonstrate, like some new type of water filtration system, a video demo may well be the way to go.

The Prototype

Unlike a demo, a prototype is a functioning version of your product that a prospect can test drive for themselves. That does NOT mean it’s feature/function complete NOR free from bugs, cosmetic and/or functional. The idea is to have the prospective customer experience your product for themselves.

The sequence through these phases so far is tell, show, experience. If you’ve validated your idea to the extent you and your colleagues and backers feel confident in continuing to invest in product development, then do.  But sometimes at this stage it may be time to pivot, e.g. change direction, or even fold your cards.


Back in the old days of assembly language, then C and C++, when the cornucopia of tools and open software that exist to day weren’t even a glimmer in the eye of the most farsighted software seer, beta testing was a very distinct phase. I believe Netscape pioneered today’s mode of a constant stream of releases, letting the customers act as the QA department.

Of course, with products like medical devices the testing, validation and acceptance phase is vitally important and may even be government regulated. So how much you test, how and how long, are largely determined by the type of product you are producing.


Whether or not you’ve done beta testing, by this stage your customer development process should give you the confidence to create what is called the Minimal Viable Product. The MVP includes only enough features to be a useful product and to test the market, with your constant aim being market validation and customer acceptance. The MVP is showtime!

Product Launch 

Assuming your MVP meets with success, you’ve gotten feedback from your early adopters about what is lacking or flawed in the MVP and have used your customer base to help your prioritize your feature list, your bug list, optimizations, extensions, etc. Once you have added your top priorities to the MVP you are ready for your full product launch, which means a go-to-market strategy. That usually means social media, PR, trade shows, analyst meetings, Facebook ads, etc. Obviously your go-to-market strategy has to be developed in parallel with product development and take into account what’s been learned about customers during the MVP phase.

First Customer

No one likes to go first! So getting that first customer is orders of magnitude harder than 2 through N. You should bend over as far backwards as you can without injuring yourself to make that first customer happy. Especially now in the heyday of social media, there’s nothing better than a satisfied customer and nothing more deadly than a dissatisfied one – never has it been easier to complain about a product or service to thousands or even millions of people. So under promise and over deliver. Double and triple check your product or service. Provide the highest level of support you can. As you grow you’ll probably have to move from, say phone support to email or forum support, but you won’t have a chance to grow without getting that vital first customer.

First Revenue

There’s a reason you see those dollar bills framed in old mom and pop shops. No bigger milestone than first customer revenue! Just be careful about payment terms and cash flow. The bigger the company, the slower they pay. 90 or even 120 days is not uncommon. So remember that booking, billing, and receiving cash in the door are three different things. And as the saying goes, “in startups, cash is king.” It sure does help to meet payroll!


Ok. so you have your first customer and your first revenue. You just made it through spring training and now the real season begins. If you have investor money they are interested in two things: growth, then a liquidity event. So figuring out how to scale your business, get economies of scale, drive down customer acquisition costs, optimize your supply chain, get volume discounts and more are the new challenges once you have a product that sells.Not to speak of the need for customer support and the rise of competition.


Profitability hasn’t been in vogue since Netscape went public as a money-losing company decades ago. But with the recent tech bubble seeming to be deflating it may come back into vogue. Knowing when to take a profit and when to re-invest it into the company’s growth is a complex topic for another day. Suffice to say, generating a real – not an accounting-generated – profit, is perhaps the ultimate milestone short of a liquidity event.


Liquidity refers to turning the company’s assets – products, customers, IP, etc. – into a form that can be easily transferred. That usually means cash, but it could mean stock in a publicly traded company. If you have taken investment from angels, VCs or other investors they expect and deserve a return on their investment. Depending on investor patience you’ll have roughly five to seven years to return their investment with interest, either by selling all or part of the company to another party, or by going public.

Check out the book Nail It then Scale It: The Entrepreneur’s Guide to Creating and Managing Breakthrough Innovation by Nathan R. Furr and Paul Ahlstrom for help in finding your product/market fit, then growing profitably. Highly recommended by my serial entrepreneur friend, Giuseppe Taibi.



Usability testing


There’s nothing like watching a total stranger attempt to use your product to induce shock and awe in developers! I know. I’ve seen it. Over a decade ago my friend Giuseppe Taibi and I developed one of the first mobile shopping apps, called oddly enough, iShop.

Somehow, through means I don’t recall, I found a Bentley College student who was working on UI projects and had access to the Bentley User Experience Center. She set up a testing session for us with a number of Bentley students who didn’t know us and we didn’t know them. This is very important in usability testing: too many entrepreneurs rely on colleagues, friends, friends of friends, family, and acquaintances to get feedback on their product. You must use total strangers to eliminate bias.

The second important aspect of usability testing is to give the testers a set of tasks to accomplish.

The third and most important thing is to watch your testers attempt to use your product to accomplish those tasks!

Careful observation behind one-way glass, along with recording the usage sessions is vital. It’s the best way to learn how well your UI works, where users gets lost, confused, frustrated or even, on occasion, pleased.

It’s also vital to have a facilitator from the lab – no one from the company should do any more than observe and take notes. Again, the goal is to eliminate bias.

One of the big problems with doing market research through asking customers what they want or how they like your product is that most people a) don’t know what they want until they see it, as Steve Jobs told Business Week : “A lot of times, people don’t know what they want until you show it to them.” and b) what people say and what they do often tend to vary dramatically. Watching what they do with your product is the best way to learn about the user experience. And that’s what you’re selling, right?

As Henry Ford said, “Ask people what they want and they’ll tell you they want a faster horse.” But show them a car and they’ll quickly forget about horses altogether, and there goes the buggy whip industry!

There are a number of UI experts who specialize in running usability tests. I highly recommend you seek one of them out and get their help in designing your usability tests – and I do mean plural, because repeating usability testing during development can keep you on track in developing a product experience that satisfies and delights your customer. Contact me for names if you like.







Your company as a program



Disclaimer, I am not a programmer! I’ve struggled to learn FORTRAN, COBOL, BASIC, and Logo way back in the last century to no avail. Not enough brain power, persistence, or training, and very poor attention to detail.

But – it occurs to me that one way you entrepreneur/programmers can think of your company is as a program:

  1. Architecture – your company’s architecture is it’s organizational design. Are you a partnership like a law firm or design house? Or are you are a hierarchical corporation? Are you organized functionally like Apple or by business units like Google? (This last question is obviously one for mature companies, but it’s worth thinking about early on.)
  2. Inputs – data: the old saying “garbage in, garbage out”. Your inputs are your ideas for a new product or service. Like data, your inputs need to be clean, not impinge on other’s right, but sufficient in scale, etc.
  3. Code – that’s how you instantiate your ideas and manipulate your inputs, your data. In terms of your company, that translates into execution -operations. Whether it’s a manufacturing process or supply chain, it’s a dynamic process which like has loops, branches, and even recursion. Many mentors, myself included, believe ideas are cheap and plentiful, success comes down to execution.
  4. Testing – just like you test and validate your code, release beta versions,  find bugs, and fix them, the same goes for your company. You may test your hires (at my companies we used to have 3-month probationary periods for all but senior staff to see if there was a fit or not).
  5. Optimization – Once you have a team you are not done – in fact building a company is never done, it’s a continuing process. But you need to optimize, meaning making sure everyone is doing what they do best, tasks are properly delegated, everyone has a clear idea of the vision and direction and their day-to-day responsibilities, etc.  Companies need to move fast these days, so you’ll need to optimize your staffing and org design for speed of execution.
  6. Outputs – the main output of a company is its products, just like the main output of a program is a result. But there are other outputs to be aware of, mainly “brand” – company reputation. A good reputation will help you attract job candidates, partners, investors, and of course, customers. If your outputs are not meeting your expectations you may have to rewrite the company code – how it operates – to generate the desired outputs. Or you may have to go back to your inputs – is your idea valid and worth pursuing?

I could stretch the analogy a bit further by saying we are talking system software here – infrastructure, rather than application software – when comparing your company to a program.

But the bottom line is companies like software, need to be designed. Letting them just grow organically is like planting a garden and never weeding or tending it – you’ll end up with an overgrown mess.

Means vs. Ends

A typical problem I see with first time entrepreneurs is confusing means and ends.

Your company (or product) should start with the “ends”. Seems counter intuitive to start at the end, rather than the beginning, but it works. The “ends” are the benefits delivered to your target customer. Either the problem or problems you solve for your customers or the new opportunity you create for them.

Then you need to work backwards to figure out the “means” the “how” of reaching the desired ends. See my post on the The journalist story telling framework, for more of the what, how, why, etc. of your company.

The problem with this method is obvious – virtually no entrepreneur starts with the customer, with the exception of the entrepreneur who invents something to solve his own problem, as Dan Bricklin, then a student at HBS did. He was frustrated by having to recalculate the financial models created in his classes every time a single variable changes, so he invented the first spreadsheet, VisiCalc, which automatically recalculated the entire model when any variable changed. Dan was his own first customer.

Typical entrepreneurs start with an idea or business concept, not with a customer who has a problem they want to solve. If they stay in this mode we often see their product as  “a solution in search of a problem.”

So you need to move quickly from the idea the “ends” – what is the benefit your product or new business is going to deliver and what are the steps, measurable milestones to get there?

This is the power of the MVP (Minimal Viable Product) as it gets you to market faster and enables you to see if you are really delivering the ends you think you are.

But be careful as a typical technologist not to get lost in the trees of product development and miss the forest of the market you are going after.

A simple technique is to continually ask yourself while in product development, is this feature or optimization or add-in or whatever helping get you to the “ends” – delivering customer benefits? Or is it just cool, new, “innovative” and fun to code?


Boiling the ocean

One of the most common problems I see with the entrepreneurs I mentor is grossly over ambitious product plans. The common term for this is “boiling the ocean” – your reach exceeds your grasp. Another phrase for this is “overrunning your headlights.”

One solution to this problem is the product roadmap. Product roadmaps can start with the ultimate vision of your product – that Swiss Army knife that becomes a worldwide best seller for decades. But you need to start working backwards, one step at a time.

Corkscrews are pretty tough to design and manufacture  compared to a straight blade, so perhaps the corkscrew is the penultimate feature.  You can then hop back to the start of the project, perhaps the handle, or the straight knife blade. Then move to the next milestone, the serrated blade. The process is up to you. It’s the final result that counts.

Mapping out a product roadmap has both internal and external advantages. First, it can capture the creativity of your development and marketing teams, because your aren’t fighting them about scope creep – let them put those blue sky ideas on the draft roadmap, but way out there by product ship, not for next week.

The roadmap is a work in progress; the more engaged you get with the market, the more likely it is to change, based on what you learn about customers and what they start asking you for. There are various tools, such as Gantt charts, you can use.The tool doesn’t matter. What matters are two things: getting  buy-in from the team and continually maintaining it.

If you don’t know where you are going, all roads will take you there. Don’t put that roadmap in a virtual drawer somewhere – you need it as guidance. There are no GPSs for product development.

The external value of a roadmap is for investors and partners. Investors generally are interested in two things: growth and predictability. A roadmap can show vs. tell both variables. Partners need to plan just like you, sharing your roadmap with them can help them be better partners to you.

Forcing functions

As a manager you are responsible for meeting schedules, for software development, for product launches, for raising capital, for whatever. Schedules are a fact of life in startups. So is schedule slippage.

As a product manager faced with this problem I discovered a useful tool: the forcing function. What is a forcing function? It’s an external force that acts as a force multiplier to someone struggling to keep a project on schedule.

The best forcing function can be an external date, set by a 3rd party over whom you and your company have no control. For example, setting product release schedules based on launching at an industry trade show works very well. Those dates are set well in advance, rarely if ever change, and missing them can carry a high cost to the company. There are dozens of these every year. Make use of them.

Another forcing function is a VIP visit. Companies always want to show off their latest and greatest stuff to visiting VIPs. The bigger the VIP the further out that visit date usually is, and the more important having your product perform flawlessly.

New projects or products on the company’s roadmap can also be a forcing function, assuming your team is working serially. Development has got to end in order to start something new – on time. Developers tend to like building new stuff rather than debugging and optimizing old stuff.

Scheduling is tough. Keeping programmers, who by the nature of software, have difficulty projecting completion dates, to a pre-set time table is very difficult. Yet another reason to develop small modules quickly rather than large, monolithic products slowly. The shorter the interval, the higher the probability that the projected time to completion will be accurate.

Scope creep

One the greatest things about today’s software is the ability to build features quickly and easily. But that only feeds into one of the classic problems of the startup entrepreneur: scope creep. No, “scope creep” is not a high tech term for a creepy voyeur – it means the tendency of developers to keep adding “just one more feature” – until schedules slip, bugs appear, memory requirements soar, and software gets slow and bloated. Not that everyone one of those disasters will befall the well-meaning developer who tells the product manager – and I remember hearing this many times “but I can implement that over the weekend!”

As I was taught by Bob Frankston at Software Arts, development is all about rapid iteration: build something quickly, get it out, fix it or change it, release again – it’s a cycle. Focusing on your customers, beta testers, or just friends and family and responding quickly to their needs  and feedback (not necessarily their wants!) can help avoid developers being left to their own devices and adding features willy nilly.

While the days of 20 page specs are long gone and certainly not missed, that doesn’t mean having a spec or project roadmap with clear deliverables, milestones, and dates isn’t still the responsibility of the programming team and/or product management.

As Peter Drucker said, “If you can’t measure it, you can’t manage it”. That goes for software: you need to be able to manage your software by setting up measurable goals, resource budgets, and schedules.

Scope creep is a bit like weight gain, if you never step on the scale and have a tendency to gobble up sweet stuff on sight, you are going to gain weight. Before you know it, you’re fat. No one likes fat, slow, and late software. So beware of scope creep!