So I’ve been uninvolved in the day-to-day operations of Synapp.io since February, mainly doing custom development for folks (some call it “fractional startup CTO” but I’m not actually really sure what that means). While this pays the bills, I’ve been looking for what’s next for me, and I’ve been exploring healthcare IT.

This space is interesting to me for a number of reasons, but it really boils down to these two facts: a) I inhabit a body and b) many of my friends and people I care about also inhabit bodies. So things that make those bodies work better, longer, etc. are interesting to me, and if I can somehow marry that with a business, so much the better.

The problem with healthcare IT is that it’s an incredibly broad market, with way too much to explore in a reasonable period of time. So I’ve been trying to narrow things down. One avenue I’ve been exploring a little is bringing some of the retail analytics technology and science I learned while working at Predictix to healthcare providers, particularly in helping them optimize their inventory of consumables and replenishment strategy to drive costs down. While there’s been _some_ interest in this, I’m not sure I feel the authentic demand that everybody at Flashpoint talks about.

Last week, in seeking to learn more about the market, I attended Health Connect South. It was an incredibly educational experience, and I got a chance to meet a number of people that I’m certain I never would have run across outside that context. One thing that I noticed during the conference is that nobody was talking about the costs of delivering healthcare, or inventory, or stuff like that, at all. That could be simply due to the prevalence of practitioners in attendance (Doctors don’t generally get into medicine to become business managers), but it made me reconsider supply chain as a focus.

What people *did* talk about a bit, particularly in the “Big Data and Healthcare” panel, were issues that were all related to electronic health records (EHRs):

– While an EHR may have all the data a doctor needs to deliver care, it’s not particularly well-organized or accessible at the point of care. (My doctor may not find my Grandmother’s breast cancer relevant to my strep throat infection, for instance.) Bringing the most relevant information to the front seemed to be something missing from existing solutions.
– Consumers on their own, and sometimes patients at the directions of their doctors, are starting to collect a _lot_ of data using wearable devices. Often, this is not even present in the EHR, and when it is, it can get lost in a jumble of unstructured data. (If my Fitbit has continuous heart rate monitoring on me for the last 6 weeks, why on earth is the nurse taking my pulse at the beginning of my annual physical?)
– *Precision medicine,* customizing treatment to the individual and their own genetics and biochemistry, is becoming more and more of a hot topic. Any kind of precision medicine, however, relies on actually having a place to store the data on the patient as a first step.
– Clinical trials are perpetually in search of participants who meet certain criteria, and filtering and screening applicants is currently a time-consuming process that really _should_ be easy to do (IMO) with an appropriate EHR solution.
– A fragmented EHR market means that patient intake is a much more painful process than it needs to be. When I start seeing a new specialist, they may go through a number of tests and questionnaires that collect unnecessary, redundant information that *already exists in another EHR*.

Now a good bit of work is being done to prod healthcare providers and EHR vendors to make their systems work better together. In particular, the CMS Meaningful Use standards (https://www.healthit.gov/providers-professionals/meaningful-use-definition-objectives) are connected to financial incentives. HL7 and FHIR are making some strides in defining common APIs and data formats, but it’s still a fragmented market, with the technology over-promising and under-delivering.

Immediately my mind went to the idea of One EHR to Rule Them All, a portable record that went with the patient from provider to provider. But of course the problem with an idea like that is that every EHR is trying to be the One EHR. (“The best thing about standards is that there are so many to choose from.”) So assuming I can’t do that, what else could solve the problem?

One idea would be to provide an EHR to EHR connector service. Think of it as ETL for healthcare, or Zapier for healthcare. There are certainly lots of barriers in place to a business like that (one of the largest EHR providers charges tens of thousands of dollars per site for export functionality to be enabled). So I did a little googling and found out that there are already businesses – mostly consultants – who will build your interoperability for you. Now a good business (IMO) that could be built here would be one that develops such integrations and interoperability as a service, but does it in such a way as to collect a library of IP that makes each new project just a little bit easier.

All that being said, I’d love to make that pivot in my own “fractional CTO” work, so if you know of anyone who’s doing any EHR integration or interoperability work, I’d love to talk to them to see if I might be of service. I’d also love to hear any opinions on whether building the EHR-EHR connector service solves a real problem for healthcare providers, or if I’m just imagining that it would.

For those who missed it, here are the earlier parts of the story:

• The Story of Synapp.io, Part 1
• The Story of Synapp.io, Part 2
• The Story of Synapp.io, Part 3

Flashpoint at Georgia Tech

Tim Ferris, in The Four Hour Workweek, states that

A person’s success in life can usually be measured by the number of uncomfortable conversations he or she is willing to have.

If this is true, the four months spent Flashpoint will move you ahead in life by four or more years.

To kick off our time at Flashpoint, we (and the other teams in our batch) were told to go out and do some (10? 15? 20? I don’t recall exactly) customer interviews, which we would report on later. OK, so that’s uncomfortable for me, self-professed geek, but I can do this. Mike and I set out across the street to the Atlanta Technology Development Center, knocked on a few doors, and asked a few questions. We were proud of our “mad customer discovery skills,” and more convinced than ever that “Big-data analytics yielding actionable insights to help optimize your digital marketing” was a winning idea. Then we presented what we learned.

To say that Merrick Furst (founder of Flashpoint) was unimpressed would be an understatement. Of course, he was unimpressed with all of the teams, equally. But Mike and I, we figured we were really smart and thus not subject to the requirement to learn the hard way. But the truth is that to get into Flashpoint, you’re already really smart, you know you are, and Merrick isn’t going to waste his (or your) time telling you how smart you are. He’s going to tell you what you’re missing, what you’re blind to in the market, and the whole host of cognitive errors you’re making. I don’t remember exactly the feedback that Merrick gave us, but to give you a flavor, here are some of the things that were directed at our team from time to time:

I’ll stop interrupting you when you say something that doesn’t sound like you just made it up.

 

Please sit down. You have nothing of value to share with us tonight (in front of a room with all the other teams, mentors, and investors)…. Actually, please stand back up; let’s deconstruct why you didn’t make more progress this week.

 

I know it’s hard. Do it anyway.

It sounds (and feels) brutal. But the process is necessary. Illusions are being shattered, and that just kind of sucks. You have put your life outside this business on hold because you’re convinced you have a world-changing idea. And you just found out that your idea is almost certainly a loser. Eventually you learn that all the ideas start out as losers. It’s not a loser because you’re dumb; it’s a loser because you haven’t done the work yet. Merrick has added a bit more nuance to his feedback over the years since our batch, but the overall journey for teams still visits the same emotional peaks and valleys today; they’re just eased into it a bit better.

Over the course of our time at Flashpoint, Merrick made countless contributions to us and the other teams, but what always stands out to me is his rare (unique?) ability to put himself into a state of utter skepticism as a way of helping teams. He watches you work, present, and pitch, and focuses on finding the flaws, never being drawn in by the story. And he tells you that it’s exhausting, that he wants to buy into your story, and that he’s telling you you’re full of it because he actually wants you to succeed. (You start to actually believe that a little bit around month 2.5/4.)

“Pivot” Time

So Mike has various business & family interests in Romania, and he takes trips there from time to time. He was on one of those trips when I had office hours with Merrick. Alone. I tried to talk about how we were going to do digital marketing optimization, starting with email, and it was going to be awesome. Somehow, though, the fact that it was me telling the story and not Mike made it a bit less plausible. Ok, it sounded completely implausible. Lesson? If you need the guy who invented the idea to tell the story convincingly, it’s not a good story.

So I would say something, Merrick would push back, I would restate, more pushing back, and all the while I was wishing Mike was here to have this fight instead of me. Eventually, I said something to the effect of “Fine. We have a business idea that solves a known, urgent problem with ready buyers who we can serve effectively at scale. But it has competition, and it’s not that interesting.” So I described email validation (i.e. removing undeliverable email addresses from marketing lists without actually sending to the list – which is what we actually do today), why it’s important to email marketers and email service providers, and the fact that we even had a little micro-product that could do it. His response, as I remember it, was something like “Then just do that.”

So I kinda betrayed Mike’s vision for marketing optimization in that meeting, but things in the business started to pick up. Eventually, we would acquire the domain DataValidation.com from Mike’s old company and the stream of customers that came with it. By the time we left Flashpoint, we had an actual business with revenue that could actually support me full-time (Mike was still doing other things to supplement his income.) We would hire 6 people by the end of the 2013, raise a bit of money from angel investors, and be on the path to profitability. Of course, the story is never quite so simple, and for the next chapter you’ll have to check out part 5…

For those who missed it, here are the earlier parts of the story:

• The Story of Synapp.io, Part 1
• The Story of Synapp.io, Part 2

Flashpoint

So Mike’s new idea from Part 2 was to apply to Flashpoint, a startup accelerator here in Atlanta led by Merrick Furst. Apparently, to hear Mike tell it, Merrick was present at the Startup Weekend where Mike pitched the original idea for Synapp.io, and Merrick chased him down to suggest that he apply to Flashpoint. My thoughts were, “hm, ok, sure, why not?” I didn’t know much about the Flashpoint way of doing things (Startup Engineering), about Merrick personally, or really much about startups at all. I know that another of my fraternity brothers Rob had gone through Flashpoint, so maybe it was valuable.

So Mike put together an application, and then we had an interview. I know that both Merrick and Sig Mosley were in the interview with us asking questions, and that there were others, but some of the details are somewhat unclear 3 years out. What I do remember are two things that Merrick said. First was after Mike pitched the idea: “We will use big data to derive actionable insights to help small and medium businesses optimize their digital marketing.” Sounds cool, I guess. It was definitely enough to convince me to join, after all. Merrick’s response, as near as I can recall, was “that sounds compelling and plausible, which is how I know that you’re almost certainly wrong.” Hm. That was unexpected. (Merrick is good at “unexpected,” I would come to learn.)

The second thing about the interview that stood out happened during an exchange concerning my work history. At one point, I had worked for a small local company (it’s bigger now) building retail analytics software. I had some technical disagreements with the CEO that I felt pretty passionately about, and decided that it would be best if I resigned and went elsewhere. As it turns out, that was one of the things that pushed Flashpoint to accept us into the program. Huh?

If you’ve met Mike, you may have guessed why that was a plus for us. Mike enjoys the gift of presence when he enters a room. When he pitches, you want to believe him. When he debates, his opponent starts to doubt all his own premises. And when he says “this is the way we should go,” the natural response is to follow. What Merrick knew is that while that is undoubtedly a gift when running a well-defined business, it can be a liability when discovering a business because you start to believe your own story. Flashpoint is many things, but some of the most important things you learn there are techniques to fight your almost unlimited capacity for self-deception.

So why was my resignation from my old employer important? Because it meant that I had some chance of actually standing up to Mike if the data contradicted his story. This doesn’t mean that my ideas are less wrong than Mike’s, it just means that all the cognitive errors I make (or anyone makes) aren’t 100% correlated with his. So if we both agree an idea is bad, we probably won’t consider it, and if we agree that it’s good, it has a chance of being good (and should proceed carefully). If we disagree, it’s time to stop and think, because one of us is definitely wrong (and quite often, both of us are). It turns out in hindsight that most of the progress we actually made at Synapp.io, particularly at Flashpoint, came from those moments of disagreement.

So that was the application process. As you may have guessed, we got in, without much of any idea what to expect. What did we find? A difficult, sometimes harrowing journey that was nonetheless invaluable in moving Synapp.io from a pitch to a product to a business. But you’ll have to wait till next time to hear that story.

For those who missed it: The Story of Synapp.io, Part 1

Starting up our startup

When Mike and I hitched our wagons together to form Synapp.io, Mike suggested an equity split of 75/25 in Mike’s favor. Not having any experience in such things, I sought advice from the Google as to how much equity a technical co-founder should receive. Obviously this is a contentious topic, and unfortunately there are no clear answers. What I found when I searched, however, was a nifty flowchart on Nathan Hurst’s blog. Following the chart, depending on exactly how I answered questions, it looked like 20-30% was about right, so I accepted the offer, and we got to work.

I had read Eric Ries’ The Lean Startup, and I figured that Mike had already done the necessary customer discovery, so we set out to make a minimum viable product (MVP) based around a few key ideas:

1. Customers would “link” their email service provider (i.e. ESP — e.g. Mailchimp, Constant Contact, etc.) accounts with us
2. We would look at what they’re doing and tell them how to improve (“actionable insights”) using all the stuff their accounts tell us about them (“big data”)
3. ???
4. Profit!

Let’s build stuff!

So we set out to build a multi-ESP optimization platform. (If we’d actually been insightful at that stage, it would have been clear to us that we just gave up the “M” in “MVP” — and probably the “P” as well, since we weren’t 100% sure how to charge.) At the time, we had a couple of other folks helping us part-time: my fraternity brother who initially did startup weekend, and another friend of his who was going to do the front-end of the platform. We had no money, so we offered to figure out an equity compensation at some time in the future. (It turns out, when you don’t pay people, they have to spend most of their time doing things for people who do pay them, so the other 2 eventually had to drift away.)

Now Mike had a business before Synapp.io that was tangentially related called DataValidation. What they did is remove undeliverable email addresses from databases, reducing bounce rates and getting more email in the inbox. Seemed pretty simple, and as a developer, I figured I’d attack the problem as our first “optimization.” I will say that, while it’s not rocket science, email list cleaning is not trivial, either, especially at scale.

Baby steps in the Atlanta startup scene

So in early 2013, we were in build-build-build mode, meeting at coffee shops, Regus business lounges, and wherever else we could meet. Eventually Mike mentioned that there was this thing called Startup Gauntlet at the Atlanta Technology and Development Center (ATDC, a local startup accelerator/incubator) that we should check out. Basically we got up in front of people and said what we were doing, and were told we didn’t know what we were doing (fair point, and it was the same for everyone). We did a few customer interviews as a result of the Gauntlet, but eventually the 7am start time was too much for Mike and me (we both live a bit north of the city), so we dropped out, thinking that we’d discovered something (but not knowing exactly what we’d discovered).

So Mike’s first idea of a “startup thing” (the Gauntlet) we could do didn’t quite lead to riches and glory. He had one more idea that he’d been sitting on since startup weekend that he’d bring up to me next that made all the difference….

Background

I’m a geek. Some computer hardware design, but for most of my career software. I always thought I’d be the “individual contributor” who brought in so much value that he became rich and retired early to do some as-yet-unspecified Important Work that didn’t pay much (or anything). That was around 1996, when I first graduated from Georgia Tech.

A couple of jobs and a Master’s degree later, I ended up in a startup doing software for computer hardware design. I was a “Senior Systems Engineer,” but that was mainly due to title inflation at startups. I discovered that I liked the startup vibe. I liked knowing that my work was appreciated and made a huge impact on the trajectory of the company.

I did not like running out of money and getting laid off. Nor did I particularly enjoy failing at making a go of the technology of the business on my own. And I especially didn’t like going back into corporate America, into a career track that terminated as a Corporate Software Engineer. Ick.

So I worked a couple of other jobs. I read The Four Hour Work Week. I wanted to build a business now, not just work in one. I started speaking at (software) conferences and wrote a book. I quit my job and became a consultant/trainer, hoping to find that golden idea that would make everything come together. I wrote another couple of books. It was the fall of 2012, and the “pipeline” of contracts started to dry up. Was I going to need another job?

Enter the Business Guy

Out of the blue, I get a call from a fraternity brother who had just done a Startup Weekend. He’s also a geek, but he’s a geek with 4 kids (I only have 2) and a steady job, so he couldn’t take the idea further. He offered to introduce me to his partner Mike for the weekend, where they took 2nd place despite being a team of 2.5 (their other programmer left). They had a business model. It was lean and validated. Sounded great, so we decided to get together and I’d hear him out.

Mike came in with an idea for marketing optimization. Not automation, but optimization. The idea being, lots of folks are automating their digital marketing but have no idea whether they’re even moving in the right direction. Mike had run a test, and it said he should be able to get $20k revenue per month. I did the (very simple) math — assuming a 50/50 split (it wasn’t) and immediate $20k revenue per month (it wasn’t), I’d have plenty to live on from day one. So Spring of 2013, we formed an LLC and started work on this thing which we called Synapp.io (like synapse, but with applications, and the .com was taken).

Stay tuned for what happened when the idea met reality….