Experiment 1 Results: Cold emailing businesses

Last Tuesday I undertook my first experiment. The idea was to email 50 businesses cold, to see if it could be a useful tool for discovering an interesting problem to solve. I sent two variations of my cold email, a polite version and impolite version.

I had two hypotheses.


  1. If I email 50 businesses, I will get 4 positive responses. Out of those 4 positive responses, I will get 1 meeting.
  2. The Impolite Version will have a better response rate than the Polite Version.


  1. I ended up sending just 46 emails, and got 5 responses. Of these 5 responses, I was able to convert one into a phone call and on-site interview.
  2. The Impolite Version got 3 responses, whereas the Polite Version got 2 responses. Not exactly a landslide. So I’m calling it a wash.

Onsite interview

This was cool. One of the respondents invited me to their office to chat about their business. I went, and learned a bit about their industry, and a lot about how they run their day to day business operations.

Unfortunately I didn’t hit on any big problems in talking to the business owner. I have one idea I’m going to follow up on related to, suprise suprise, making filling out governemnt paperwork easier for this industry. The owner also passed my name to another company to talk to.

A little more data

I had a response rate of 10.8%. That’s better than I was expecting. Of those five respondents, I got one meeting. That ain’t bad percentage wise either, though the numbers are so small at that point they don’t mean much. I could have tried a little harder to convert those respondents by calling them. Frankly, it was a little chicken of me not to.

I was suprised how much time it took to do this all. Here’s an hourly breakdown:

  • 2 Hours Obsessively drafting and re-drafting email templates
  • 4.5 Hours Trolling the web for prospects to email, building spreadsheet
  • 2.5 Hours Sending 46 emails
  • 2.5 Hours Responding to 5 emails back/forth, phone interview with 1 customer
  • TOTAL: 11.5 Hours


It went better than I thought it would, but that’s not saying much. It took me 11.5 hours to get one on-site interview with a business owner. Super inefficient. I could definetly optimize the email sending process, I did it mostly by hand with the help of Textexpander. More difficult would be optimizing the research that went into building a spreadsheet of prospects to email. That was mind-numbing work that took way too long. Maybe that’s something where a virtual assistant could help.

I have to think that cold calls convert better than emails. That’s probably the real answer here. Get on the phone and start calling. It’s a scary prospect for an introverted nerd like me. It comes down to this: The hustle must overcome the nerd. And it shall.

Bureau of Labor Statistics (BLS if you’re down)

Amongst the many chicken/egg based scenarios I find myself in, none is more poultry-like than this one: how do you find a niche with a problem that needs solving, when you don’t even know which niches exist?

There are so many vertical industries that I have never heard of, it’s stupefying. How do you start knowing, what you don’t even know? Research. You need to begin general, then hone in on the specific.

In Start Small, Stay Small, Rob Walling suggests using the Bureau of Labor Statistics to check out a specific industry. That makes sense if you have a specific niche in mind. It’s easy to find info.

But what if you got ‘nuthin? No idea what vertical you want to address. Perhaps you want a list of every job category known to man, in one skimmable 85 page list, sorted alphabetically and cross linked with industry-specific stats?

Well bless my lucky tax dollars, such a thing exists.

Occupational Outlook Handbook, 2010-11 Edition

If you read this index, and I mean really read it, you will come away with one unmistakable conclusion: there are a lot, a LOT of proper nouns to describe someone who works in the logging industry. Sorry, inside BLS humor. You’ll have to read the index to get it.

I did read this beast. And I came away with some industries to check out that I would not have thought of, here’s some:

  • Agricultural inspector
  • Asbestos Abatement
  • Blood Bank technologists
  • CNC operators
  • Community Health Educator
  • Conference interpreter
  • Elevator repair
  • Environment auditor
  • Family therapist
  • Gas and water service dispatcher
  • Occupational Health and Safety Specialists
  • Language pathologists
  • Lead abatement professional
  • Massage therapist
  • Sports scouts
  • Tenant selector
  • Wholesale buyers

The trick is that some of these jobs map really nicely to one industry. And some don’t at all. Elevator repair people work at elevator servicing companies. Easy one. But wholesaler buyers? They work in retail I guess, but retail is about as much of a vertical as small businesses are.

So there’s still legwork to do, to flesh out and investigate each category, to get specific. But at least this gives me something to start with, and I can end my stare down with Mrs. Chicken and Mr. Egg.

Experiment: Cold emailing for ideas

My first experiment starts today. Cold contacting businsesses to find a problem worth solving. You can read a little more about what’s going on here in this post.

So the first thing I’m going to try is cold emailing potential customers. But before we get into that, what am I talking about when I say “the first experiment starts today?”

Experiments as mental scaffolding

This whole deal of finding customers with a problem is part of the customer discovery process. Sure, I’ve read books about this process. But reading about something and actually doing it are worlds apart. Will it really work for me? How can I know if what I’m doing is even working? And how do I know when it’s time to try something different?

I realized I need a little bit of structure, a way to make decisions about what stuff is working for me, and what’s not so much. So I’m imaging each new thing I try is an experiment. Each experiment will have three parts:

  • An action I’ll perform.
  • A hypothesis that I can prove or disprove.
  • And a deadline for completion.

I realize my criteria for what makes an experiment won’t pass muster in the academic world. That’s fine. Again, this is really just a way for me to break down and better understand each new thing I’m trying as I develop my business.

So with that in mind, let’s lay the ground rules for my first experiment.

Experiment 1: Cold email businesses to discover a problem

Email 50 businesses. Tell them who I am, how I found them, and what I’m trying to do. Ask them “What annoying problems do you have in your company?”

I’ll send two versions of my cold call email. The Polite Version where I say who I am and what I’m doing up front, and the Impolite Version where I don’t say what my angle is until towards the end of the email. The Impolite Version is also shorter.

1. If I email 50 businesses, I will get 4 positive responses. Out of those 4 positive responses, I will get 1 meeting.
2. The Impolite Version will have a better response rate than the Polite Version.

The action portion of this experiment must be completed by end of day Feb 22nd. I’ll evaluate results the follwing Tuesday, the 28th.

We’ll see how it goes

I don’t know what to expect with all this. These numbers (50 businesses, 4 responses, 1 meeting) are just me making it up as I go along. And I’m cool with that. Who knows if these numbers are anywhere close to reality, but I have to start somewhere, so I can compare where I get next.

Have problem? Will travel.

My goal is simple. I have to find a customer with a problem that needs solving.

The problem should be:

  • Something that can be solved by software
  • An “I will pay you money to fix this right away” level problem

And the customer should be:

  • A small business (they’re approachable, faster decision makers, usually underserved)
  • In an industry that employs more than 10,000 people
  • Within driving distance of me

How do I find them?

This is the first huge hurdle I have to clear. How do I even find these customers? There are so many potential businesses and verticals to approach, where do you start? I can think of three ways of getting this going.

  1. Use my personal network to find customers
  2. Cold contact (email or phone) customers after researching them
  3. Use search engine keyword tools to see what products businesses are searching for

I’m going to try all three methods. Each way is an experiment for me. So I’m going to approach them that way, with a clear hypothesis, that I can prove or disprove.

I think I’ll first take a crack at method number two, cold contacting prospects. Look for a post on that soon.

Keylime Lean Canvas Iteration 1

What is all the mumbo jumo in this post’s title?

Keylime is the web-based help desk software I built.

The Lean Canvas idea was created by Ash Maurya, I believe he adapted it from this book. It’s an alternative to a traditonal business plan document. It gathers up all assumptions and beliefs about your business in one easy to read page. Unlike a traditional bizplan document, which no one really reads or updates, it’s simple to update and revise as you learn more about your business.

Iteration 1 because this is the first version I’ve made. There will be many, many more as I test and learn from my hypotheses about my business.

Here’s a link to my first iteration of a Lean Canvas for Keylime.

View Keylime Lean Canvas Iteration 1

I’m using my friend Clifton’s webapp Revisu to share it. Revisu is cool because you can easily leave feedback on documents.

Foggy understanding

Creating this canvas made me uneasy. I really have a lot of unverified assumptions about the help desk market. And the assumptions I do have, I’m not even that confident they’re right. Here’s a few areas of concern.

Broad customer segment

Small businesses is not a market. It’s way too broad. I need to go crazy narrow to start. That’s the source of the rest of the confusion here. How can you know the needs of such an impossibly large market?

I think things will start to fall into place if I can just find a particular customer support vertical to work with.

Uncertain about the problem

In the canvas I list what I believe are the top three problems with customer support. Are they though? I’m not really sure. The first two problems (cases get lost, cases hard to delegate) really apply to people who are using email as their only support tool. The third problem (help desk software is complicated) applies to software like Autotask or Zen Desk, which try to be all things to all people, and are generally really fiddly.

Again, it would be more interesting to hone in on one customer support niche and undrestand what their specific problems are. I have a feeling I’ll discover a completely different set of problems.

Unique value proposition not so unique

It’s hard to be confident in a UVP when you don’t understand the problem. These UVPs reflect that. They’re kinda wishy washy.

You gotta have a baseline

I’m sharing this, let’s admit it, crappy first version warts and all because I want a baseline to compare against. You have to know what you don’t know, before you can know what you need to know (is that you Rumsfeld?).

I’ll continue to publish new versions of this canvas as I iterate and learn.

Last Day

Today was my last day at the old IT job. I’ve been doing IT support for ten years, so a chapter of my life is closing.  Whatever comes next, it will be interesting if nothing else.

Also, it feels very very strange not having to check my cellphone for new email every five minutes.  I am okay with this.

Starting from scratch, now what?

My last few posts have looked back about how I got to this pivotal point. I’m here. What am I going to do about it?

I see two options.

  1. Take my Keylime Help Desk software, and customize it for a vertical market. E.g. help desk software for veterinarians.
  2. Build a new product from scratch. Find a class of businesses that have a problem, and fix it.

The common denominator here is finding an actual living breathing customer that has a problem, and that will give me money to solve it. At this point I’d be happier to have a customer and no product, then a product and no customer. You can build a product, but you can’t build a customer.

The customer development way

In contemplating how to move forward, I’ve been doing a lot of reading. One author who has been deified in startup circles and is having a deep impact on my way of thinking now is Steve Blank, who created the concept of customer development.

The core idea behind customer development, as opposed to product development, is you work with potential customers to test your product vision before writing a line of code. This sounds like simple requirements collection, as is done in the old school waterfall development approach. But it is very different for two reasons.

The learning loop

First, customer development is all about iterating. You come to the customer with your initial hypothesis of what their needs are, and a vision for how to fix them. By talking to customers, you test that hypothesis, and it probably gets shot all to hell. That’s good! You just saved months of coding a product customers didn’t even want.

So you refine your hypothesis, focusing on the 20% you got right, and have further rounds of discussions with customers. Each time you refine your vision. Maybe things start to get real enough that you can build a very basic prototype, purely as a learning tool to further test your vision with customers.

The point is that there is a learning loop. You have a product hypothesis to test, you test the idea with your customers, you learn from the test, and you reformulate your hypothesis based on what you learned. Lather, rinse, repeat. That’s a far cry from the collections requirement way of building a product.

One customer is anecdotal evidence

The second way customer development is different than traditional requirements collection phase of product development is, you don’t talk to just one customer. Learning from just one customer is not enough data to go on. You have to meet with several.

Only when your product hypothesis is validated against multiple customers is it safe to proceed. In his book The Four Steps to the Epiphany, Steve Blank suggests you work with a minimum of five customers. He gives a very practical guide to finding customers, and suggest you might have to call 100 customers before you find five visionary customers who are the kind you want to work with in these early stages.

I need to find five companies

So that’s my goal. To find at least five companies in a particular industry which have the same problem. It has to be a problem that hurts enough that they are willing to pay to have it fixed.

I’ll be on high alert for companies that have problems related to help desk or customer support needs. But no problem is off the table at this point.

How do I actually start finding and talking to companies? That’s kind of an important point. I have some ideas, stay tuned.

What went wrong with my web app

In the Maoist tradition, it’s time to undergo self-criticism. I started building a webapp in 2008. I released it in 2011 (!). No one is paying for it. It failed. Fuh-hay-ulled.

Sure, I’m bummed about it. But I’m also obsessed with making a new product that succeeds. In the spirt of of learning from my mistakes, here are a few reasons my webapp failed.

Scratched my own itch

Building a product that you’d plunk down your own cash for seems like sage advice. And it can be. But what if there are lots of credible itch-scratchers out there already? Are your nails sharper? You’d better be sure.

I took it as an article of faith that if I built for myself, customers would show up at my doorstep. In my crowded market, it didn’t happen.

Didn’t set aside work hours

If this project was a child, it would have some serious daddy abandonment issues. I’d go months at a time without working on it. Building something by yourself after hours is tough. Without a strict work schedule, it’s crazy tough.

Failed too slow

It took me way too long to realize how crowded the market I was entering truly was. If I had done some tests before hand, say create a Coming Soon sign-up page coupled with modest adwords buys, I would have felt the pain much sooner. Next time I’m failing faster.

Didn’t validate product with customers

Talking to customers tests your product vision. If I had talked to customers earlier, I might have discovered some niche vertical where competition isn’t so fierce. Then I could have expanded from there. From a position of strength.

Moving on

Phew, glad that’s through. This is my last “big things that went wrong” post for a while. But I felt like I had to clear the air. Expect many more “little thing that went kind of right” and “little thing that went sort of wrong” posts soon.

Quitting my job with vague plan, no product.

Hi, I’m Ian. I’m just another dude leaving the corporate world to start my own company. Here’s my deal.

Feb 17th is the big day

T-minus 14 days to…bold new business opportunity? Unemployment?  Yes.

I have two weeks left of the ‘ole IT support day job. I’ve been in the industry ten years, but got burned out. It happens.

The first spark of burnout came three years back. I needed a path out. So I taught myself to program and started building. A webapp, natch. Something I knew pretty well, help desk software. That’s the software folks like me use to keep track of customer problems.

Three years ago that market was just sad. Help desk software was hard to use and it was slow. I saw an opportunity to build something better

Driving below the speed limit

Say hi to Keylime. I released it.  But I did it on the side, very very slowly. Life kept interrupting me too. I moved to a new state, got married, bought a house, and stayed crazy busy with my day job.

Man, I paid a price for being so slow. While I wasn’t paying attention, the help desk software market exploded. Everyone was building simple help desk software now. Pay attention at your next family gathering. It’s entirely possible your Great Aunt Nellie has written a web based help desk system that you can try free NOW for 30 days, no credit card required!

There’s a problem with several well-funded competitors getting there before you did.  You have to fight harder for a smaller piece of the pie.  Why not find a different pie instead?

Pivot is a five letter word, not four

I decided it’s time to pivot.  Pivot is a magical word. It’s way more concise than saying “Abandon old deluded plan, find new deluded plan.” So I’m pivoting.

I can pivot in two directions. Pivot left, and I take my help desk software and apply it to a specific kind of business. Help desk software for dog groomers? Or I can pivot right, and build a new product for a new industry.

But why not just keep working part time? Well, on top of being too slow, I made a second mistake. I didn’t get out of the house and talk to customers before I started building. For this this pivot to work, I’m going to have to do some serious door knocking, pavement pounding, old school hustling. I need to get out there, talk to businesses, understand their problems, and see if I’m the man with the vision to fix’em.

That’s the plan. I couldn’t see doing all that part time, so I quit. I got a vague plan and no product. But I have an itch to take a risk, and to strike out on my own.

Here goes something.