# Transparent Guarantee Design > Builds a sales guarantee that closes deals without exposing the agency to a metric it can't hit — pick one narrow, self-contained, measurable outcome (not a revenue promise dependent on the client's own conversion skill), define the exact tracking mechanism before the guarantee ships, and pair it with a fallback offer (a fee reduction until the metric lands) so the guarantee is risk reversal, not a coin flip. Use when close rate is stalling on trust, when an existing guarantee is vague enough that "did we hit it" is an argument instead of a fact, or when a guarantee promises an outcome (revenue, ROI) that depends on factors outside the agency's control. - Skill: `search-atlas-group/transparent-guarantee-design` (Agent Skill) - Install (CLI): `npx skillmds@latest add search-atlas-group/transparent-guarantee-design` - Raw SKILL.md: https://api.skillmd.com/api/skills/search-atlas-group/transparent-guarantee-design/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Finance & Business - Author: search-atlas-group (https://skillmd.com/u/search-atlas-group) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/search-atlas-group/transparent-guarantee-design --- # transparent-guarantee-design **The problem this solves:** a vague guarantee ("we'll grow your business") sounds generous in a sales call and becomes a liability the moment a client asks whether it was met. The agency has no clean answer, the client feels misled either way, and the guarantee that was supposed to build trust destroys it instead. A guarantee promising an outcome the agency doesn't fully control (revenue, ROI, "more customers") is even worse — those depend on the client's own conversion, staffing, and follow-through, not just the agency's work. This is the design method a member's agency uses to build guarantees that are narrow enough to always be measurable, and honest enough to survive being checked. > **A guarantee is not a promise of results. It's a promise of a specific, trackable > metric, with a pre-agreed fallback if that metric isn't hit on time.** --- ## Say this to your agent > "Design a guarantee for [service]. Pick ONE outcome we fully control the measurement > of — not revenue, not 'more customers,' something like a keyword ranking position or a > ranking placement. Write the exact tracking mechanism before we send the guarantee: what > tool measures it, what counts as a hit, what counts as a miss. Then build a step-function > guarantee: state the initial, easily-verified guarantee up front (e.g. measurable > improvement to the tracked metric), and only evolve it to a bigger promise (ROI, > revenue-based) once we've delivered against the first step and trust is established. > Pair the guarantee with a concrete fallback if we miss the deadline — a fee reduction > until the metric lands, not just an apology. Flag anywhere the guarantee's wording could > be read as promising something we don't directly control." --- ## The three rules (all three, not a subset) | Rule | What it does | What breaks without it | |---|---|---| | **Pick a self-contained metric** | Guarantee something the agency's work alone determines (a ranking position, a placement) — not something gated by the client's own sales/ops (revenue, "more customers") | The agency does everything right and still "fails" the guarantee because the client didn't close the leads it sent | | **Define the tracking mechanism before the guarantee ships** | Decide up front what tool measures it and what counts as a hit/miss | "Did we hit it or not" becomes a dispute instead of a lookup — the fastest way a guarantee damages trust instead of building it | | **Step-function the promise** | Start with an easily-verified, narrow guarantee; only evolve to a bigger outcome-based promise (ROI, revenue) once the first step has already proven out | Selling the big promise on day one either overstates what the agency controls, or gets walked back later and reads as a bait-and-switch | --- ## The fallback is what makes it risk reversal, not a coin flip A guarantee without a stated fallback is just an unenforceable promise. The member's version pairs the tracked metric with a concrete consequence if it isn't hit on time — a 50% fee reduction ("haircut") that stays in effect until the metric lands. That single addition is what turns "trust us" into a guarantee the client can actually rely on, and it's also what keeps the agency honest about which metrics it's willing to guarantee in the first place — nobody offers a 50% haircut against a metric they don't believe they can hit. ```text ## Guarantee — Tracked metric: Tracking mechanism: Timeline: Fallback if missed: Step-function next stage (only after this one is proven): ``` --- ## Two example guarantees from the same agency (why both work) - **The self-contained one:** commit to a top-3 Google Map Pack placement for an agreed keyword, for a local-service client, within an agreed window. Fully agency-controlled (rankings), fully verifiable (check the map pack), and it maps directly to something the client already understands is valuable. - **The action-guarantee variant:** commit to a specific dollar-recovery outcome ("we'll find and recover the $50,000 you've lost, here's exactly how") — this one leans more on things the *client* has to do, so it's framed as a partnership guarantee rather than a pure agency-delivery one. Used deliberately, alongside the map-pack guarantee, not as a replacement for it. Both examples share the same underlying discipline: a metric specific enough that "did we hit it" is a lookup, not an argument. --- ## Watch-outs (a guarantee that fails this design) | Signal the guarantee is unsafe | Fix | |---|---| | The guaranteed outcome depends on the client's own sales/ops execution | Reframe to the leading metric the agency actually delivers (rankings, traffic, placements), not the client's downstream conversion | | No stated tracking tool/definition before the guarantee is sent | Write the tracking mechanism first — if you can't define "hit" precisely, the guarantee isn't ready to sell | | The guarantee jumps straight to an ROI/revenue promise with no proven step before it | Step-function it: verified narrow win first, bigger promise only after that's delivered | | No fallback stated if the metric is missed | Add one — a guarantee with no consequence for missing it isn't risk reversal, it's marketing copy | --- *Sourced from a live member contribution shared in the 2026-08-20 AMM cohort session — the working guarantee-design method one agency uses to close deals without exposing itself to a metric it can't measure or control.*