The POC as Business Case
Frame every pilot as co-creating an ROI model, charge for it smartly, and never anchor the commercial deal.
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 3
- Confidence
- 92%
Reframes the AI proof-of-concept from a technical-functionality test into a 30-day exercise whose sole purpose is co-creating a business case with the buyer. You charge for the POC to qualify serious buyers, but deliberately decouple that price from the eventual commercial deal, and deflect premature price demands with value-contextualized ranges.
Origin
From Madhavan Ramanujam's Scaling Innovation, addressing the two questions AI founders most often ask about early commercial motions.
Core principles
- 01The entire goal of a POC is to create a business case — functionality, integration, and fit are consequences of that, not the point.
- 02Charging for a POC qualifies serious buyers and filters out tire-kickers who burn 30-90 days and never buy.
- 03The POC price must not become the anchor for the commercial deal.
- 04When pushed for a number, contextualize on value or give a range — never a single point estimate.
How to run it
- 1
Reframe the POC as a business-case exercise
Position it explicitly: 'a 30-day pilot for co-creating an ROI model and building a business case. If we see value at the end based on the business case, we can get to commercial discussions.' This avoids talking price while focusing both sides on value.
Watch out Framing the POC as a proof of technical functionality sets the wrong expectation and forgoes the commercial narrative you need.
- 2
Charge for the POC — smartly
Attach a price tag so both sides signal seriousness and you qualify leads. But make explicit that the POC fee (e.g. $10k for 30 days) is only for building the business case and is not an indication of the eventual commercial deal.
Watch out If you quote a $10k POC without decoupling it, you have anchored the buyer at roughly $120k/year for the full deal.
- 3
Deflect premature price demands with value or a range
If pushed for a number before the business case exists, either contextualize on value ('for customers such as yours we've unlocked ~$10M in similar situations, and our pricing is 1:10 on ROI') or give a budgetary range ('final pricing would be anywhere from $500k to $1M, and we'll pick a point that justifies the value we co-create').
Pro tip The value-ratio framing tells the buyer you're roughly $1M without saying it, and justifies the number as 1-in-10x ROI.
Watch out Giving a single point budget (e.g. '$200k') is the worst thing you can say — it becomes the ceiling.
In the wild
When a buyer insists on a price during the pilot, the founder responds that for similar customers they've unlocked around $10M in value and price at roughly 1:10 on ROI — implying ~$1M without stating it, framed as justifiable.
→ The buyer gets a defensible indication of price anchored to value rather than a raw number to negotiate down.
Common mistakes
Running free POCs open to everyone
Curious buyers who just want to see if the AI works will engage for 30-90 days, burn your resources, and never buy — a price tag filters them out.
Letting the POC price set the commercial anchor
Failing to state that the pilot fee is separate silently establishes the annual deal value in the buyer's mind.
Is it for you?
Best for
Early-stage AI founders running enterprise pilots who must protect scarce time and avoid under-anchoring big deals.
Not ideal for
Products with no meaningful pilot phase or self-serve motions where a POC concept does not apply.
From the transcript
“The P should be framed as the entire goal of the P is to create a business case. Period. Full stop.”
“the reason you need to charge for a P is you start isolating people who are just tire kickers versus serious buyers. It becomes a…”
“you have to be clear that the 10K is only for building a business case. Commercial discussions will follow after that.”
“Give them a range. You can say something like look the final pricing would be anywhere from 500k to a million.”
From the episode
Pricing your AI product: Lessons from 400+ companies and 50 unicorns
Madhavan Ramanujam