# MarginPulse Kit v0.1 A practical worksheet for founders and small SaaS teams pricing LLM-backed features. This file is the first downloadable kit artifact: useful enough to validate demand, but still honest that no paid demand has been proven yet. ## What this kit helps decide - Whether an AI feature belongs inside an existing plan, a higher tier, or an add-on. - What call cap protects target gross margin under baseline and heavy-user scenarios. - Which usage language to publish before customers discover the edge cases. - Which vendor/model prices must be verified before quoting customers. ## Non-negotiable rule Do **not** paste vendor prices from memory. Before using this commercially, verify current model pricing on the vendor's official pricing page, then enter those numbers into the calculator. MarginPulse is a pricing guardrail, not a live price database. --- ## Step 1 — Define the buyer promise | Question | Draft answer | |---|---| | Customer segment | | | AI feature name | | | Job the feature performs | | | Current plan price / customer / month | CHF | | Is AI usage unlimited, capped, or metered? | | | Target gross margin after AI and other COGS | 80% default / other: | | What would make this feature unprofitable? | | Decision checkpoint: if the AI feature is core to the promise and usage can spike, default to caps or tiering before promising unlimited use. --- ## Step 2 — Scenario table Fill prices in CHF per 1M tokens after checking official vendor pricing. Then run each row through `index.html`. | Scenario | Calls / customer / month | Avg input tokens / call | Avg output tokens / call | Input CHF / 1M | Output CHF / 1M | Other COGS / customer | Plan price | Gross margin result | Action | |---|---:|---:|---:|---:|---:|---:|---:|---|---| | Baseline expected usage | 400 | 900 | 350 | | | | | | keep / adjust | | Heavy but legitimate user | 2,000 | 1,200 | 600 | | | | | | cap / tier / add-on | | Free trial abuse case | 1,000 | 1,000 | 500 | | | | 0 | | throttle / require card / limit | | Team account with shared workspace | 3,000 | 900 | 350 | | | | | | seat rule / workspace cap | | Higher-quality model switch | 400 | 900 | 350 | | | | | | reserve for paid tier | | Cheaper fallback model | 400 | 900 | 350 | | | | | | fallback if quality holds | Action rule of thumb: if heavy-user margin is below target while baseline is healthy, do not raise the base price blindly. Add a fair-use cap, tier boundary, or add-on trigger. For a guided version of these cases, open `scenario-library.html`: it turns the table into six copyable launch-review scenarios, including a purchase-intent qualifier for validation replies. --- ## Step 3 — Required price and cap notes | Plan / tier | Included AI calls | Required price at target margin | Planned public price | Gap | Decision | |---|---:|---:|---:|---:|---| | Free / trial | | CHF | CHF 0 | CHF | | | Starter | | CHF | CHF | CHF | | | Pro | | CHF | CHF | CHF | | | Team | | CHF | CHF | CHF | | Use the calculator's “Required price for target” as a guardrail, not as the final product price. Final price still depends on customer value, alternatives, support load, and conversion. ### Copyable pricing decision memo After running each scenario, copy the calculator memo into your launch/pricing doc. The memo should include: plan price, modeled AI calls, cap, LLM cost per customer, total COGS, gross margin, required price at target margin, recommended next action, and the explicit warning that vendor prices were user-entered and must be verified. Minimum decision rule: if the memo says “underwater” or “fragile,” do not publish unlimited AI. Choose one of: higher tier, add-on, hard cap, soft cap with upgrade conversation, cheaper fallback model, or narrower AI feature scope. --- ## Step 4 — Usage-cap policy templates Choose the least surprising rule you can enforce technically. ### Soft cap template > Your plan includes [X] AI actions per month. Most customers stay below this. If usage is consistently above the included amount, we may contact you to move to a higher tier or add usage-based billing. Best when: you are still validating usage distribution and do not want to break workflows abruptly. ### Hard cap template > Your plan includes [X] AI actions per billing month. After the included amount is used, AI actions pause until the next billing cycle unless you upgrade or buy additional usage. Best when: API cost exposure is material and the feature is not mission-critical during overage. ### Fair-use / abuse template > To keep service reliable and pricing fair, automated, abusive, or unusually intensive AI usage may be rate-limited. We will avoid limiting normal human use without notice where practical. Best when: free trials, bots, or very uneven usage can create loss-making accounts. ### Model-tier template > Standard AI actions use our default model. Higher-quality or longer-context actions may be limited to [Plan] or billed as add-ons because provider cost differs materially. Best when: the product needs both cheap routine outputs and expensive premium outputs. --- ## Step 5 — Launch decision checklist Do not ship the pricing page until every line is either answered or deliberately deferred. - [ ] Current official vendor prices checked and copied into the scenario table. - [ ] Baseline margin meets target at the public plan price. - [ ] Heavy-user case has a cap, tier, add-on, or model fallback. - [ ] Free trial has a usage limit or anti-abuse rule. - [ ] Team/workspace behavior is defined: per-seat, per-workspace, or pooled allowance. - [ ] Public copy does not promise unlimited AI unless unlimited usage is economically safe. - [ ] Support has a plain-language explanation for overage or throttling. - [ ] Metrics exist to observe AI actions per account after launch. - [ ] Model-switch plan exists if vendor pricing or quality changes. --- ## Step 6 — Demand validation record Count only concrete demand signals. Do not count likes, impressions, or vague encouragement. | Date | Source/community | Signal type | Exact request summary | Counts? | |---|---|---|---|---| | | | reply / opt-in / purchase intent | | yes/no | Validation gate for a paid checkout: 10 concrete replies, opt-ins, or purchase-intent messages asking for the worksheet/template pack. --- ## Changelog - 2026-05-22T07:37:28+02:00: Added `scenario-library.html` guided scenarios and purchase-intent qualifier to support validation replies. - 2026-05-22T07:35:14+02:00: Added copyable pricing decision memo guidance for launch/pricing docs. - 2026-05-22T01:18:34+02:00: Created v0.1 downloadable kit artifact in the Mercury Loop workspace.