# Six People and a National Signup Flow

> HealthCare.gov launched on 1 October 2013. By the morning of 2 October, war-room notes recorded six successful enrolments.

Canonical: https://www.diffraction.sh/blog/six-people-national-signup-flow

## The morning it went live

On the morning of 1 October 2013, HealthCare.gov went live.

It was supposed to be the front door for Americans in states that had not built their own insurance exchanges. Create an account. Verify identity. Check eligibility. Pick a plan. Enrol.

Millions of people showed up.

Almost nobody finished.

By the morning of 2 October, internal "war room" notes later released by the House Oversight Committee had a number that still makes engineers stop scrolling: six successful enrolments.

Six people had made it through the whole path.

Notes from that morning meeting name a handful of issuers across those early wins. By an afternoon meeting the same day, enrolments were "approximately 100." By the end of 2 October / morning of 3 October, the cumulative total in the notes was about 248.

You do not soft-launch a national signup flow at six completions.

That is a product that could not finish the job it was built to do, on the day it had to do it, while the country watched.

## What users hit

People got outages. Cryptic errors. Pages that crawled.

A lot of them could not create accounts at all. Others got stuck mid-flow.

The war-room notes from those first days read like an incident channel you would rather not be on. High capacity problems. Direct enrolment not working. Experian credit-check confusion. Waiting-room estimates on the order of 40,000 people showing up in the notes. Direct enrolment was still broken into the afternoon of day two.

The funnel was not one page. It was a chain.

Account creation failed. Identity verification confused people. Eligibility and plan selection never arrived. Enrolment never happened.

The site lost users at every step. Then it lost the ones who remained to infrastructure that could not carry the load.

A waiting room is not a feature when it becomes the product.

HHS Secretary Kathleen Sebelius later told Congress the initial consumer experience had not lived up to what people expected, and was "not acceptable." She apologised for a flawed launch. The hearings ran for hours. The phrasing was careful. The reality underneath was not.

## The money under it

This was not a skunkworks experiment.

CMS had obligated hundreds of millions of dollars across contractors for marketplace IT. An HHS Office of Inspector General overview of sixty related contracts found nearly $800 million obligated for the Federal Marketplace by February 2014, with roughly $500 million already paid. Original estimated contract values across that set totalled about $1.7 billion at award. For many contracts, obligations had already overrun those estimates. Costs ballooned. Vendor coordination was a known weak spot. Later OIG contracting reviews criticised the lack of an overarching acquisition strategy, and missed chances to plan for a lead systems integrator before launch.

The Government Accountability Office later spelled out why day one looked like that.

Inadequate capacity planning. Coding errors found and not fixed before go-live. Planned functionality still incomplete. Testing inconsistent: systems were not fully tested before launch, and test docs lacked key pass criteria. Readiness reviews did not catch what end users found in the first hour.

Deadline pressure won.

End-to-end product verification lost.

## The conversion problem

Engineers like to split "infra incidents" from "UX failures."

HealthCare.gov erased that line.

Capacity planning is a product decision when the product is a signup flow. Incomplete functionality is a product decision when the incomplete bits sit on the critical path. Inconsistent testing is a product decision when nobody has walked a stranger from a blank browser to an enrolled plan under something like real load.

The war-room notes keep the human scale honest. Six enrolments by the morning after launch. Roughly a hundred by afternoon of day two. Two hundred forty-eight by the next morning's roll-up.

Public reporting talked about millions of visits in the early window. Visits without completions are not a soft win. They are a funnel with a cliff where the button should be.

For a while, officials were not ready to publish reliable enrolment totals even while contractors' dashboards were getting talked about inside the war room. What leadership can say and what the product can do are different problems.

Multi-step flows hide failure modes that single-page demos never show.

Each contractor module can pass its own tests. Account creation can "work" in isolation. Identity can "work" in a lab. Plan shopping can render in staging with ten users.

The product only exists when all of those steps succeed for the same person, in order, under real traffic, with real third-party dependencies (credit checks, issuer systems, waiting rooms).

Large government IT is hard. Everyone already knew that.

The sharper point is this: a conversion path can get treated as a programme of contracts and milestones while the actual user journey stays unproven. When the only hard gate is a calendar date, the soft gates (capacity, defect burn-down, end-to-end rehearsal) get negotiated away.

Organise a launch around the date, and you may ship a date.

Organise it around completed journeys, and you ship a product.

## After day one

The site eventually stabilised. Capacity went up. Bugs got attacked in waves. Enrolment numbers rose in later months.

That recovery is real. It still does not rewrite day one.

If the product's job is to move people through a multi-step path, the only launch metric that counts is how many people complete that path.

Traffic is an input. Press conferences are an input. Obligated dollars are an input.

Completions are the output.

Six people completed it when the nation opened the door.

The rest of the story (the war rooms, the hearings, the GAO findings, the OIG contract tallies) is what happens when deadline theatre outruns end-to-end proof.

## Sources

https://www.nbcnews.com/politics/politics-news/only-6-able-sign-healthcare-govs-first-day-documents-show-flna8c11509571

https://oversight.house.gov/release/102-healthcare-gov-meeting-notes-6-enrollments-occurred-far/

https://www.gao.gov/products/gao-15-238

https://oig.hhs.gov/reports/all/2014/an-overview-of-60-contracts-that-contributed-to-the-development-and-operation-of-the-federal-marketplace/

https://oig.hhs.gov/oei/reports/oei-03-14-00230.pdf

https://www.nbcnews.com/news/us-news/sebelius-acknowledge-glitches-pledge-action-obamacare-testimony-flna8c11488703