Should I run a pilot? Paid vs free pilots and pilot-to-paid conversion
Charge for pilots when work, data risk, support, or legal effort start. Set price, scope, and a short path to annual contract.
The short version. Run a pilot only when it can prove live buyer value and move straight into annual paper. Charge for it when your team will touch data, tune product, support users, or carry legal risk. Free only fits a bounded sandbox or a strategic buyer who gives real access first.
Before you start:
- Charge when the pilot uses live data, custom work, support, or security review.
- Free fits a narrow sandbox or a buyer who already grants rare access.
- Keep pilots short, buyer-owned, scoped to one live workflow, and tied to annual paper.
- Judge pilot-to-paid by annual wins, stall reasons, buyer access, and repeatable use proof.
Step 1: Qualify the pilot ask before you price it, and define what done looks like
Say no early. A real pilot starts after the buyer names the live workflow, the hands-on owner, the budget path, the legal path, and the date when an annual order can be signed. Enterprise buying rarely belongs to one person; Gartner says the typical B2B buying group has 6 to 10 decision makers at https://www.gartner.com/en/sales/insights/b2b-buying-journey.
Free trials hide weak deals. If the prospect cannot give data access, a user group, a buyer sponsor, a success metric, and a path to the annual contract, the ask is research for them and services work for you. When your first meeting came from outbound, sanity check the ask against Does cold outbound still work in 2026? Benchmarks and what gets replies.
- Named workflow and user owner
- Data access before kickoff
- Budget owner on the call
- Security path already known
- Annual contract terms sketched
Step 2: Charge for risk, labor, buyer proof, and time, and define what done looks like
Charge when you will touch data, tune models, build workflow, train users, or carry support risk. Price creates gravity. For most seed-stage enterprise pilots, I would start around USD 5,000 to USD 25,000 for 30 to 60 days, and I keep that paid-pilot stance close to the founder-sales bias SaaStr publishes at https://www.saastr.com.
Zero can still work for a narrow sandbox, a public-data demo, a design partner already under contract, or a senior buyer who grants unusual access. Cash tests the buyer. A paid pilot also gives legal, finance, and security a smaller version of the real deal before the annual paper lands.
- Sandbox only: free can fit
- Live data: charge
- Custom work: charge
- Buyer asks for exclusivity: charge more
Step 3: Keep the pilot short, scoped, buyer-owned, and measured, and define what done looks like
Time kills pilots. A sharp enterprise pilot runs long enough to see the real workflow and short enough that the buyer cannot drift into vague learning. My default is 4 to 8 weeks for a workflow tool or AI copilot when the buyer can supply data fast, with the paid-pilot stance above anchored by https://www.saastr.com.
Scope one workflow and one user group with one success metric. Founder energy must stay on learning the buyer's path to revenue, away from babysitting a loose trial. If this is still founder-led, borrow the handoff logic in Scaling out of founder-led sales without breaking your pipeline before you let a rep inherit pilot chaos.
- One live workflow
- One buyer sponsor
- Named daily user group
- Shared data checklist
- End date on calendar
Step 4: Write the mutual plan before kickoff, and define what done looks like
Paper beats memory. Before kickoff, write a one-page plan with owner names, dates, data needed, success metric, support limits, price, and the annual contract path in plain language their boss can read. Legal may still take time; that smaller paid order gives the buyer's team a rehearsal for vendor setup, invoice flow, security review, and redlines.
Mutual plans fail when founders write them alone. Buyer names matter. Put the buyer's project owner beside your product owner, then name the exec sponsor, security contact, legal owner, finance contact, and the person who can stop the deal. For early teams, the same habit belongs in your operating dashboard beside The only GTM metrics that matter before Series B.
- Kickoff date
- Data owner
- Success metric
- Annual price range
- Exit meeting date
Step 5: Measure pilot-to-paid, learn fast, and define what done looks like
Folklore lies. Track pilot-to-paid as a stage in the same board where you track source, buyer role, contract value, next step, and close risk. No public benchmark tells an AI founder the clean typical paid-pilot-to-annual rate; your useful bar is whether paid pilots become annual deals more often than they stall.
Below that bar, fix entry gates before you add volume. A founder can create a fake win rate by saying yes to free proofs, friendly sandboxes, unpaid research, and vague innovation teams. Strong pilots start with a buyer who can buy, a problem that hurts, data you can access, and a path to live use.
- Paid pilots opened
- Paid pilots won
- Free pilots opened
- Free pilots won
- Stalls by reason
What goes wrong?
Founders confuse access with intent. A famous logo can drain weeks when the team wants market scan, free consulting, internal theater, a benchmark for another vendor, or a slide for next quarter. Beware polite applause. If nobody argues about price, data, risk, or rollout during the pilot talk, nobody has started buying.
Another trap is the underpriced custom build. Your team bends the product for one account, the buyer praises the effort, and the annual contract dies when the real owner sees a narrow patch. A paid pilot should fund learning, expose deal risk, prove repeatable use, and protect team focus.
- Free pilot with no buyer
- Paid pilot with no scope
- Long pilot with no end
- Happy users with no budget
Common questions
Should an enterprise pilot be free?
Usually charge when the pilot touches live data, custom work, user training, security review, or support from your team. Free can fit a sandbox where the buyer self-serves and you learn without burning founder hours. Any free pilot still needs a named buyer, a live use case, and a path to annual paper.
How much should I charge for a paid pilot?
For a seed-stage enterprise product, I would test USD 5,000 to USD 25,000 when the pilot runs 30 to 60 days, with the same paid-pilot bias SaaStr writes around at https://www.saastr.com. Higher annual value, custom work, heavy data risk, or exclusivity should push the fee up.
How long should an enterprise pilot run?
Keep it short: 4 to 8 weeks usually gives enough time to see live workflow proof when the buyer can supply data fast, and that range fits the paid-pilot stance tied above to https://www.saastr.com. Longer pilots invite drift. A short pilot forces scope, buyer access, and a real end meeting.
What is the typical conversion from paid pilot to annual contract?
No reliable public typical rate exists across SaaS, AI services, and internal tools. Your bar should be simple: a paid pilot should become an annual deal more often than it dies. When wins fall below that bar, tighten entry gates, scope, sponsor access, and price before adding more pilots.