You will get asked for a number. Here is the honest version: there is no standard price we can quote you, and the published figures that do exist are mostly about labor.
The cloud is small. At one rare-disease biotech with a brand in market, Snowflake averaged a bit over $700 a month and AWS a few thousand more. The two vendors around it, billing hourly, cost roughly $30,000 a month. If you build a budget off the infrastructure invoice, you will be wrong by an order of magnitude.
The other large line is syndicated data: IQVIA, Symphony, specialty pharmacy, claims. It is usually the biggest single line, and we have no published figure for it to give you. It is also the line a data team cannot move. The cost lines on the in-market page leave it unchanged on purpose.
What the published figures say, by stage
| Stage | Team on record | What it cost or saved |
|---|---|---|
| Prelaunch | One data engineer for the first 1¾ years of a launch, two from nine months out. That platform averaged 1.5 engineers. | Field reporting ready six months before launch |
| One brand in market | Before: two hourly vendors, three to four FTEs. After: a standing team of about two. | About 58% lower annual cost of running commercial data |
| Several brands | Seven engineers and 10 to 12 analysts at Celgene, across Revlimid, Otezla, and Immunology | About $10 billion a year in US sales supported, no missed SLAs |
Sources: the launch case study, the rare-disease biotech story, and the Celgene figures.
Read the 58% carefully, because finance will. It compares the new operation against the ten reports the old one had already been cut back to, not the 41 it started with. By the customer’s own count, nobody read the other 31. The same team then absorbed a second indication’s data work while monthly hours trended down. What that adds is in the answer on adding a second brand.
The benchmarks you will be compared against
Three numbers come up in every budget conversation, and all three are in the cash-runway post.
Hiring your own team: six fully loaded commercial data FTEs in Boston or the Bay Area run $1.5 million to $2 million a year, plus a six to nine month recruiting cycle. Analytics consulting: billed at around $300 per analyst hour. A multi-year analytics agreement on that model runs higher than the in-house number and transfers nothing back to you. The third is the in-market case: about 58% off the annual run cost, from the rare-disease biotech story.
Our rate is flat and set at scoping. We don’t publish it, because it depends on your feeds and scope. It scales down on 30 days’ notice if a launch slips.
The costs that never get a line item
Two costs hide inside other lines, and both are larger than the cloud.
The first is analyst time. Data janitorial work absorbs 30 to 50 percent of a commercial analyst bench: reconciling the IQVIA refresh against claims, chasing a payer name that arrived three ways, rebuilding a territory roll-up. You pay for it at an analyst’s rate.
The second is process overhead. At the rare-disease biotech, a large share of the old bill came from running commercial analytics as regulated data, with clinical-grade change control on a weekly dashboard. Separating the two was a real part of the saving before anyone rewrote code.
What makes the number fall and stay down
Cutting reports and refresh frequency lowers the bill once. It does not change what each report costs, so the saving runs out when the reports do.
What changes the unit cost is automation: generated tests instead of a hand-written suite, AI writing the SQL transforms, and an architecture that does not need someone awake at 2am. At Celgene, only 18 percent of engineering time went to data loading. That is how seven engineers covered a $10 billion portfolio.
If you have a brand in market and need cash for the next one, the in-market cost page has the line-by-line version. If you are prelaunch, the launch timeline shows what the money has to buy, and when.