Page 1 of 2

InferMargin — Research Questionnaire

~15-20 minutes
I'm conducting independent research on how AI-native founders currently measure their inference economics (LLM COGS / Revenue ratio) and benchmark themselves against peers.
Your answers help me build a confidential benchmark network — the missing layer between your observability tools (Helicone, Langfuse), your finance stack (Stripe, Maxio), and your investors.
In exchange for your time, you'll receive:
1. A personalized mini-report based on your answers 2. The aggregated report from 20+ AI-native founders once complete (anonymized, percentiles only)
3. Early access to the InferMargin private beta
All answers are confidential. No company-level data will ever be published.
— Fouad, Founder of InferMargin
SECTION A — Product context

What does your AI product do? (1 sentence)

Current MRR/ARR range

Primary AI use case

Primary AI use case
A
B
C
D
E
F

If Other, please specify:

SECTION B — Current AI costs

Approximate monthly LLM API spend

Which LLM providers do you use? (approximate % split optional in text below)

Which LLM providers do you use? (approximate % split optional in text below)

Optional: approximate % split per provider

Can you currently attribute LLM costs to specific products or use cases?

Can you currently attribute LLM costs to specific products or use cases?
A
B
C

If yes or partially, how do you do it today?

Approximate LLM spend as % of monthly revenue

SECTION C — Pain

How do you currently know if your AI infrastructure costs are "normal" compared to peers? (Free text — be candid)

What metric does your board or investors ask for regarding AI economics? (Free text)

Have you ever made a decision (model swap, pricing change, architecture change) based purely on cost guesswork because you had no comparison data?

Have you ever made a decision (model swap, pricing change, architecture change) based purely on cost guesswork because you had no comparison data?
A
B

If yes, briefly describe.

SECTION D — Solution fit

If you had a private benchmark showing your inference cost vs similar companies, what specifically would you do with it? (Free text)

Which data sharing modes would you find acceptable?

Which data sharing modes would you find acceptable?

Would you pay for a tool like this?

Would you pay for a tool like this?
A
B
C
D
E

How comfortable would you be generating a read-only Admin API key for this kind of tool? (1 = Not at all, 5 = Very comfortable)

How comfortable would you be generating a read-only Admin API key for this kind of tool? (1 = Not at all, 5 = Very comfortable)
A
B
C
D
E

Are you open to a 25-min follow-up call to discuss your situation?

Are you open to a 25-min follow-up call to discuss your situation?
A
B
C

Your name

Your work email

Your company name