Problem Statement
Non-technical stakeholders often struggle to understand where a critical KPI originates, how it is calculated, and what its limitations are. This workflow produces a clear, friendly, and accessible KPI lineage explanation. It converts technical inputs into a narrative, a source/transform table, an FAQ, and ownership details—all while avoiding proprietary names and keeping the explanation concise.
Steps
Ingest KPI Details
INGESTCollect the KPI name, optional dashboard link, upstream sources, transformations, and common confusion points. Normalize these into a structured internal representation.
Retrieve Context and Validate Inputs
EXECUTECheck for missing components such as absent dashboard links or unclear transformation descriptions. Retrieve any supplemental context from the provided notes to ensure completeness.
Generate KPI Lineage Narrative
EXECUTEDraft a plain-language explanation describing how the KPI is produced, referencing upstream sources and describing transformations in non-technical terms while staying under 600 words.
Summarize Sources, Transforms, and Caveats
EXECUTEProduce a table-like text block listing sources, transformations, and known limitations using placeholder names instead of proprietary labels.
Validate Tone and Constraints
EXECUTEEnsure the tone is clear, reassuring, and teachy but not condescending; confirm no proprietary table names appear and that the narrative is within word limits.
Prompt Configurations
KPI Lineage Explanation Prompt
You are an analytics educator helping non-technical stakeholders understand a KPI. Write in a clear, reassuring, teachy-but-not-condescending tone. Use only placeholder names for data sources and transformations. Produce the following sections: (1) plain-language lineage narrative under 600 words, (2) a simple text table summarizing sources and transformations using placeholders, (3) an FAQ addressing common confusions, (4) known limitations, (5) ownership and update cadence. Do not invent data—use only the materials provided. Rephrase any technical jargon into accessible language.
Example Inputs/Outputs
KPI name: Customer Retention Rate. Dashboard link: optional. Upstream sources: subscription events log, customer master file. Transformations: filter to active customers, compute monthly churn then invert. Confusion points: why it differs from marketing's retention metric.
Narrative: This KPI measures how many customers remained active from one month to the next, using placeholder sources such as <CustomerEvents> and <CustomerProfile>. It first identifies all customers marked as active, determines which accounts churned during the month, and then calculates the retention percentage based on those who stayed. Table: Sources: <CustomerEvents> (event data), <CustomerProfile> (status fields). Transforms: filter active accounts, detect churn events, compute retention. FAQ: Why does it differ from marketing’s number? Their definition counts email-engaged users, while this KPI counts product-active customers. Limitations: Data depends on accurate activity tagging. Delayed event ingestion can shift values slightly. Ownership & Cadence: Owned by BI Analytics; updates daily at 6 AM.