Mentor's startup guide

Building a healthcare analytics company — the honest playbook

How to consolidate Epic and other healthcare data into a product hospitals will actually pay for — and the real gaps in the market where a small, focused team can win.

Epic + multi-source dataReal-time operationsMid-market focusAvoid fatal mistakesLand your first 3 customers

Section 01

The competitive landscape

Before building anything, understand exactly who occupies the market and what space they don't cover.

The most dangerous assumption: "We do Epic dashboards and data consolidation" describes what Health Catalyst, Arcadia, Innovaccer, and Epic's own Cogito product do. If your pitch sounds like this, you're competing with companies that have raised $200M–$800M and have 200+ hospital clients each.

CompetitorWhat they doFunding / scaleTheir weaknessYour angle
Epic Cogito / ClarityBuilt-in analytics for Epic customers. Retrospective, batch reports.Epic's core product — included in EHRBatch only (24–48h lag), no cross-system data, limited operational BIReal-time + non-Epic data
Health CatalystEnterprise data operating system, population health, financial analytics$600M+ raised, public companyTargets 50+ hospital IDNs only. 12–18 month implementation. $500K–$2M ACVMid-market, fast to value
ArcadiaPopulation health, payer analytics, value-based care, risk stratification$200M+ raised, 900+ customersPrimarily population health focus. Weak on hospital operations and real-time.Hospital operations layer
InnovaccerUnified data platform, care management, network performance$375M raised, unicornEnterprise sales motion, long cycles, generic dashboards.Opinionated problem focus
Health system IT teamInternal build — 'we'll do it ourselves with Tableau'N/A — varies by health systemMid-size health systems claim they will, and almost never do. Chronically understaffed.Show speed advantage
Tableau / Power BI / QlikSelf-serve analytics. Hospitals use these for ad hoc reporting.Mass market — not healthcare-specificRequire expensive data engineering to connect to Epic. No pre-built healthcare data models.Pre-built Epic connectors

The positioning that works: You are a fast, affordable, operationally-focused analytics product for 3–20 hospital regional health systems — the market segment genuinely underserved by everyone above. That's your beachhead.

Section 02

The 8 real gaps — where you can actually win

Specific, painful, budget-allocated problems that hospital executives talk about in every operational meeting — and that no existing vendor solves well for the mid-market.

How to choose: Not the biggest market. Not the easiest to build. Ask yourself: which of these did I sit in a meeting about and think "this is embarrassing that we can't solve this"? That visceral frustration is your product.

Section 03

Advantages

With healthcare environment experience, there is advantages:

Epic Knowledge

Clarity, Caboodle, HL7 interface builds, Chronicles, the actual naming conventions in the data model — most startup founders waste 12–18 months learning what you already know. This is worth 18 months of competitive head start.

🤝

Good network inside hospitals

Your first 3 customers will come from people you know — a former colleague now at a different health system, a CMO you worked with, a VP of Operations who complained about the exact problem you're solving. Cold outreach in healthcare has near-zero hit rate.

🔧

You understand the real data messiness

Epic data is filthy in ways that only insiders know — unmapped charge codes, inconsistent provider NPIs across facilities, facility-specific encounter type configurations, Clarity views that differ by Epic version. Outside founders underestimate this by 10×.

🚀

You can move fast on implementation

The biggest complaint about Arcadia and Health Catalyst is implementation time — 12–18 months to go live. A founder who knows Epic can get a mid-market hospital live in 8–12 weeks. That speed difference is a competitive moat in itself.

The advantage you must protect: Your personal relationships inside health systems are non-transferable. Have verbal conversations with 5–8 people in your network. "If I built X, would you be willing to be a design partner?" Get a yes from two of them before you write a line of code.

Section 04

Build sequence — do not try to do everything at once

The companies that die in this space almost always do so because they built a platform before they found product-market fit.

Months 0–6

One problem, one customer

  • Pick one gap — just one
  • Find one hospital contact you know personally
  • Build it with them in the room
  • Do not write code until you have a verbal commitment
  • Charge $2K–5K/month from day one
  • Go live within 60 days of starting

Success signal

First paying customer live

Months 6–12

Repeatability check

  • Can a second hospital use the same product?
  • Document everything that took custom work
  • What took 8 weeks — can it take 4?
  • Write down the implementation playbook
  • Hire one clinical informaticist
  • Sign customers 2 and 3

Success signal

3 paying customers, $15K+ MRR

Months 12–24

Platform foundation

  • Build the shared data layer all modules will use
  • Add second analytics domain from gap list
  • Start Epic App Orchard application process
  • Hire a clinical informaticist as head of CS
  • Raise pre-seed or seed round ($1–3M)
  • 8–12 hospital customers

Success signal

$1M ARR, seed round closed

Months 24–48

Defensible platform

  • 4–5 analytics modules sharing one data layer
  • 15–30 hospital customers
  • AI narrative layer on all dashboards
  • Epic App Orchard listing live
  • Series A fundraise at $3–5M ARR
  • First payer or ACO customer signed

Success signal

$5M ARR, Series A closed

The rule that saves most founders: Speed to first paying customer beats everything. Every week you spend building without a paying customer is a week of burn with no signal.

Section 05

The mistakes that kill companies like yours

Every one of these has killed multiple well-funded, technically excellent healthcare data startups in the last five years.

Fatal mistake 01

Building a platform before finding a problem

Most healthcare data startups build a beautiful multi-source consolidation platform and then go looking for buyers. Hospitals do not buy platforms. They buy solutions to specific, painful, budget-allocated problems. 'We connect Epic and other systems and build dashboards' will get you polite meetings and no contracts. Find the problem first.

Fatal mistake 02

Underpricing to get in the door

If you charge $500 per month to land a pilot, you will never get to $5,000 per month with that customer. Hospitals anchor to the first number they see. Charge a real price — $3,000–8,000 per month depending on hospital size — from the very first conversation.

Fatal mistake 03

Selling to the wrong person

IT departments do not have analytics budgets. The CIO will be enthusiastic and will never actually fund you because they are a cost centre. Go to the operational leader who owns the problem and has a P&L: CNO for workforce, CFO for revenue cycle, CMO for physician analytics.

Fatal mistake 04

Hiring engineers before hiring a domain expert

Your second hire should be a clinical informaticist or former hospital analytics director — not a second engineer. That person gives you credibility in every sales conversation that no product demo or slide deck can buy.

Fatal mistake 05

Trying to compete on features with incumbents

You cannot out-feature Health Catalyst or Arcadia. Win on speed (live in 10 weeks vs 15 months), focus (you solve this one problem best), price (60–70% less), and relationship. Those are advantages you can sustain. Feature parity is not.

Fatal mistake 06

Ignoring HIPAA and BAA requirements from the start

Every hospital will ask for a signed Business Associate Agreement before giving you access to any patient data. Use AWS HealthLake, Azure Healthcare APIs, or Google Cloud Healthcare API from day one. This is a door requirement, not a differentiator.

Section 06

Revenue model — how to price and structure deals

The right model for a healthcare analytics company targeting hospitals is a subscription fee (annual, billed upfront) plus a one-time implementation fee.

Right model

Annual subscription (per-hospital or per-bed)

Charge a flat annual fee based on hospital bed count. Billed annually upfront. Range: $36K–$180K per hospital per year depending on size and module depth.

Right model

One-time implementation fee

Charge $15K–50K separately for the Epic interface build, data validation, and training. Waive for your first 3 pilot customers in exchange for a reference commitment.

Avoid

Fixed-bid project pricing

You absorb all scope creep risk. Healthcare integrations always expand. Epic upgrades change interfaces mid-project. Never use this as your primary model.

Year 1

$300K–600K ARR

3–5 pilot customers

Subsidised implementation fees. Primary goal: case studies and reference customers. One well-documented win at a 200-bed regional hospital is worth more than 10 undocumented pilots.

Year 2

$1.2M–2M ARR

10–15 customers

Implementation fees now at full rate. Second analytics module live. First regional health system (3+ hospitals) signed. Pre-seed or seed round closed to fund growth.

Year 3

$4M–7M ARR

30–45 customers

Epic App Orchard listing drives inbound. 3–4 analytics modules. NRR above 120% from module expansion within existing accounts. Series A fundraise at this ARR level.

Year 5

$15M–25M ARR

100+ customers

Platform with 6+ modules. AI narrative layer live. First payer customers signed. Path to acquisition by Health Catalyst, Oracle Health — or Series B for independence.

The number that matters most: Net Revenue Retention (NRR). Design your product and pricing from day one so that expansion is natural — a hospital that starts with operational analytics should be able to add workforce analytics, then revenue cycle, then the AI narrative layer.

Section 07

The mentor's direct take

Not the polished advice. The things a mentor who has seen this market says directly when nobody else is in the room.

On your core idea

"Consolidate Epic and other data, build dashboards" is a category, not a product

That sentence describes what every competitor in this space does. The companies that win narrow it to: "We give CNOs a real-time view of nursing labour cost versus patient acuity by unit, updated every 15 minutes, with automated alerts when any unit goes above their target labour ratio." That is a product. You can demo it in 8 minutes and close a pilot in 3 weeks.

Your first job is not to build a platform. It is to find the one sentence that makes a hospital leader say "yes, that's exactly my problem." You don't find it by thinking about it — you find it by calling 10 people you know inside hospitals and asking what problem they wish they could solve with data.

On competition with incumbents

Your size is an advantage — until you forget that it is

When competing against Health Catalyst for a deal, you will lose on features, depth of product, and number of reference customers. You will win on speed (live in 10 weeks vs 15 months), price (60% less), personal relationship, and the fact that your CEO answers the support ticket personally.

The moment you start hiring to match incumbents on features instead of dominating them on speed and focus, you will spend yourself into insolvency. Stay small and fast for longer than feels comfortable.

On the path forward

The question that determines everything

Which of the 8 gaps above did you personally sit in a meeting about and think: this is embarrassing that we can't see this data? Not which one is the biggest market. Not the easiest to build. Which one made you frustrated, personally, because you knew the data existed somewhere in Epic and nobody had connected it into something useful.

Pick the gap that made you angry. Then call the person you know who has that problem. Then build exactly what they need. Everything else follows from those three steps.

Section 08

Your first 90 days — exactly what to do

Not a strategy. An action list. Every item is specific enough that you know whether you have done it or not.

The first 90 days, step by step

1

Make 10 calls in the next 2 weeks

Contact 10 people you know inside hospitals. Ask one question: 'What is the one thing you wish you could see in a dashboard that you currently can't?' Do not pitch anything. Just listen. Write down every answer verbatim.

2

Find the answer that comes up 3 or more times

After 10 conversations, you will hear some version of the same problem from at least 3 people. That repetition is your signal — structural problems have budgets attached to solving them.

3

Get a verbal design partner commitment before writing code

Go back to the 2–3 people who described the problem most clearly. Ask them to be a design partner — give you test environment access, meet every 2 weeks, and if it works, become a paying customer. Do not write code until you have this.

4

Set up HIPAA-compliant infrastructure on day one

Before you touch any real patient data: sign up for AWS (or Azure/GCP), enable healthcare compliance configurations, prepare your BAA template, and get a HIPAA-compliant hosting attestation. This takes 1–2 weeks and costs almost nothing. Doing it retroactively takes 3–6 months.

5

Build the simplest possible version in 30 days

Not the full platform. The one dashboard that shows the one metric your design partner said they need most. Connect to their Epic Clarity or Caboodle environment. Build one chart. Show it to your design partner. Watch them use it.

6

Charge for it by day 60

By day 60, convert your design partner to a paying customer. Even if the product is basic. A customer who pays $3,000 a month is telling you something true. A customer who uses it for free is telling you nothing.

7

Find your second customer using your first customer's referral

By day 90, ask your first paying customer to introduce you to one peer at another health system. A warm referral in healthcare closes 5–10× faster than cold outreach.

8

Do not raise money until you have 2–3 paying customers

Go to VCs with 3 paying customers, a documented implementation playbook, and a clear answer to 'what would you do with $1.5M?' Raising pre-revenue in healthcare SaaS in 2025 is extremely difficult.

The 90-day goal in one sentence: By the end of 90 days, you should have one paying customer using a real product connected to real Epic data, and a verbal commitment from a second hospital to become a customer.

Resources to study

Organisations and communities worth your time

  • HIMSS — largest health IT conference, best for building vendor relationships
  • CHIME (College of Healthcare Information Management Executives) — where CIOs and CMIOs network
  • HFMA (Healthcare Financial Management Association) — where CFOs and revenue cycle leaders go
  • AMIA (American Medical Informatics Association) — clinical informatics community
  • Epic UGM (Users Group Meeting) — the single most important annual event for Epic shops
  • Health:Further, ViVE, and HLTH conferences — startup-friendly health tech events
Reading list

What to read before you build

  • The Innovator's Dilemma — Clayton Christensen. The hospital market is a classic Christensen disruption scenario
  • The Mom Test — Rob Fitzpatrick. How to have customer discovery conversations that don't lie to you
  • Lost and Founder — Rand Fishkin. Honest account of what building a SaaS company actually feels like
  • The Hard Thing About Hard Things — Ben Horowitz. What nobody tells you about running a startup
  • Epic's technical documentation for Clarity and Caboodle data models — knowing this is your moat
  • CMS quality measure specifications — HEDIS, Star Ratings, and value-based care metrics are table stakes