Context-Rich KPI Decision Report
Turn raw business metrics into a prioritized plan with actions, owners, and KPIs
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 95%
The Context-Rich KPI Decision Report begins with a raw export of the business's operating data rather than a generic strategic question. The user identifies two or three outcomes to improve and provides relevant company context, constraints, and definitions. A prompt-preparation model can organize this material before a reasoning model evaluates it. The requested report should rank initiatives, explain the mechanism behind each recommendation, propose experiments, identify cross-functional contributors, define leading and lagging KPIs, and surface risks or trade-offs. The final step is validation: compare every recommendation against the original metrics and operational reality. This produces a decision aid that can be translated directly into accountable quarterly work.
Origin
Extracted from Marketing Against The Grain, where Sully Omar demonstrates feeding SaaS metrics into a reasoning workflow to produce a prioritized action plan for churn, retention, conversion, and revenue.
Core principles
- 01Start with real operating data
- 02Optimize a small number of priority metrics
- 03Ask for tactics as well as strategic headlines
- 04Require measurable KPIs for proposed initiatives
- 05Include organizational context to improve relevance
- 06Seek caveats and pushback rather than blind agreement
How to run it
- 1
Collect the metric baseline
Export the metrics that describe the current business, including relevant usage, retention, conversion, and revenue measures. Preserve definitions and time periods where possible.
Pro tip Raw dashboard text is acceptable if it retains labels and values.
Watch out Incorrect or inconsistent metrics will distort the entire report.
- 2
Choose priority outcomes
Limit the assignment to two or three outcomes for the planning period. State both the current baseline and desired movement when known.
Pro tip Prioritize constraints or bottlenecks rather than the most visible metric.
Watch out Trying to optimize every metric at once prevents meaningful prioritization.
- 3
Add business context
Supply the business model, customer profile, product constraints, team capabilities, and current initiatives. Include information that may change whether a tactic is feasible.
Pro tip More relevant context enables more specific cross-functional recommendations.
Watch out Context volume cannot compensate for unclear goals.
- 4
Request an actionable report
Ask for prioritized strategies, specific experiments, implementation details, expected metric impact, and risks. Require the model to explain dependencies and caveats.
Pro tip Request both the tactic and the operational how.
Watch out Generic strategy headings without execution details are not a usable plan.
- 5
Attach measures and owners
For each initiative, identify the KPI, expected direction, review interval, and participating teams. Convert recommendations into accountable work.
Pro tip Include early indicators that move before the quarterly outcome metric.
Watch out Avoid treating speculative target dates or forecasts as facts.
- 6
Validate and sequence
Check recommendations against the original data and organizational constraints. Select the smallest set of initiatives that can be tested without overloading the team.
Pro tip Reject recommendations that cannot be connected to a metric mechanism.
Watch out A detailed report can still contain assumptions requiring human judgment.
In the wild
A SaaS founder supplies monthly and daily active users, churn, retention, conversion, and revenue metrics, then asks for a plan to improve two or three quarterly outcomes. The report recommends onboarding changes, behavior-triggered messages, target KPIs, and customer-success contributions.
→ The founder receives a measurable cross-functional plan rather than a generic list of growth tips.
A product team provides acquisition channels, signup completion, trial activation, feature usage, and paid conversion. It requests ranked experiments with owners, measurement windows, risks, and stop conditions.
→ The team can choose a focused experiment sequence tied to observed funnel constraints.
Common mistakes
Submitting only the desired outcome
Asking to reduce churn without supplying the current metrics and business context invites generic recommendations.
Optimizing too many metrics
A broad request creates an unfocused report with competing initiatives. Restrict the planning objective to a few priorities.
Accepting detail as proof
Granular recommendations may appear authoritative even when based on an incorrect input or unsupported assumption. Validate them against the source data.
Is it for you?
Best for
It is best for founders, analysts, and leadership teams choosing which growth or retention levers to prioritize.
Not ideal for
It is not ideal when the underlying metrics are inaccurate, undefined, or too sparse to support informed prioritization.
From the transcript
“what I've been doing is I've been deferring anything revolving my business around KPIs, metrics, anything I need to do around my business where I'm…”
“I want to optimize, you know, two to three key metrics for this quarter.”
“the more context you throw, the better these models do.”
From the episode
Which AI Model Should You Use? (Claude vs GPT & O1 Pro Live Prompt Guide)