JPMorgan's margin engine had only ever served JPMorgan. In 2022 new margin rules pulled mid-market hedge funds into scope, and most couldn't justify $2M to $4M to build their own. My MD gave me a blank sheet and twelve weeks to work out how to sell ours.
$2M is the sum of annual contract values at signing, reached by month 9. Revenue recognised in year one was about $1.6M because contracts started on different dates. The one client that didn't renew was bought by a fund with its own engine.
Phase 6 of the uncleared margin rules, in September 2022, brought funds with $8B to $50B in average notional into bilateral margin for the first time. Building a SIMM engine meant 12 to 18 months of work and quants on $300K salaries, plus a recalibration every year that nobody wanted to own.
Before calling a single fund, I interviewed 15 Sales coverage officers who managed those relationships. Their clients were getting by on spreadsheet approximations or paying consultants for one-off runs, and neither would satisfy a regulator for long.
I sized the market bottom-up. From the firms entering scope I removed anyone who already had an engine, then kept the regions where JPMorgan had Sales coverage. That left about 80 firms, worth $8.8M a year if every one signed.
Then Sales set up eight calls. I had a two-page concept brief and no product. Six funds wanted to pilot, and two put it in writing.
The commercial case, the product Engineering built and the sales motion, on one map. Pick a thread and step through it. Dotted pieces were deferred or scoped for later.
Make the case · weeks 1 to 12
The product · scoped with Engineering, built by them
Sell it · months 4 to 9
OverviewFrom fifteen interviews to a funded product. Press play or step through.
over the $60K Sales asked for
Sales wanted the lowest price they could get, to close faster. At $60K, reaching $2M took 28 clients, 35% of the whole serviceable market in one year, and Sales agreed that wasn't happening. At $80K it took 20.
over building a sales team for one product
Sales already knew more than 60 of the 80 firms. Each deal went from an intro to a 30-minute demo to a proof of concept on the fund's own file. 18 of 21 real conversations ended in a contract.
over a schema of our own
Every fund in scope could already produce CRIF files, so integration took about three weeks instead of months. It also made leaving us easy, and that made choosing us feel less risky.
over standing behind clients' regulatory filings
Legal's question was who is liable if a client files a wrong number. JPMorgan provides the calculation and a validation report the client can attach. The filing stays the client's. Legal signed the third draft.
over Engineering's eight-month plan
The extra three months were Tier 3 analytics. I moved them to a second phase, paid for by the first clients' revenue, and deferred the streaming API Engineering also wanted.
I treated this as two decisions. The model shapes how clients behave for years, and the level is one number you can revisit later. Pick each one and see where it lands against what funds were already being quoted.
Selected
Chosen
"We have the best SIMM engine on the Street. Figure out how to sell this."
"We budgeted $1.5M to build this in-house. You saved us 18 months and $1.3M."
A $12B fund, a $9B quant fund and a fintech lender hedging its book, at $40K for 90 days. Integration averaged 3.2 weeks against a four-week target. The engine matched clients' own numbers 99.7% of the time, and the gap was CRIF formatting, so we added an input validator.
All three converted at full price, and two moved up to Tier 2 for the regulatory reports. Their savings, anonymised, went into the ROI calculator Sales used from then on.
Fifteen more clients in three months. By month 6 the average client ran 180 calculations a business day. The strategy document became the template for selling other internal risk tools.
Ideas from after JPMorgan. Not built there.
The year-one model was built for adoption, and the usage threshold did its job: most clients never thought about the meter. It did nothing to lock in a second year or to reach funds below the $8B line.
Back to the pricingList price for one year and 15 to 20% off for three. Year one needed flexibility. A second year needs predictable revenue and lower churn.
Around $15K a year with a small allowance, as a way in for smaller firms. We never priced for that long tail.
We metered every call and never used the data to sell. A client running 180 calls a day could be shown exactly what Tier 2 would save on overages.
Turning something built for internal use into a product people pay for? I'd like to hear about it.
pattanayakspandan@