The Zoe Score
One number you can take apart.
A composite score asks to be trusted. This one is built to be interrogated: nine dimensions, weighted for the company’s stage, rolled into a single 0 to 100 score, all pointing back to the source data.
The nine
The nine dimensions of company health.
Each one is read from sources the company already runs, scored on its own evidence, then weighted for stage into the composite. Each dimension has its own ring and its own row. Select a row to read what it measures.
Close cadence, AR aging, invoice rhythm, burn consistency
Weighted for stage, the nine roll into one composite, and every point of it decomposes back to the target’s own data.
Weighted by stage
Same nine dimensions. Different stakes.
A Series B and a PE platform company do not fail the same way, so their dimensions do not weigh the same. The set never changes. The weighting does, and the composite recomputes from the same evidence.
The bands
Three words, calibrated.
A band is not the verdict. It names how seriously to take the number, and the word travels with the color everywhere it appears.
Behaving like a company of its size and stage should. Nothing to raise at IC.
A named, dated pattern worth a diligence question, and often a Founder Call question. Not on its own a reason to walk.
Thesis-breaking if it holds. Read the evidence trail before the IC, not after.
What it isn’t
Uncertainty is visible.
The score measures behavior and shows where it came from. The verdict is the signed opinion in the Readout Package, issued only when confidence clears the bar.
Zoe reads QuickBooks for cadence and concentration. Your accountants still issue the QofE.
Confidence is stated as a percentage and coverage as a tier on every readout. Pending renders as pending; gaps are shown, not smoothed.
The score is predictive in framing: it scores what is about to happen. The validation study is not done, so we do not claim proven predictive validity.
Frequently asked questions
Where does the data come from?
From the pitch deck and data-room files added to the case, and from the systems the target connects read-only in its own session: QuickBooks, HubSpot or Salesforce, GitHub, Jira, Slack. Read-only scoped OAuth, patterns not paragraphs in the connected systems, extract-and-discard by default. No surveys. One Founder Call closes the gaps the Claim Ledger leaves open, and each answer is scored answered, dodged or contradicted. Nobody fills out a form.
How are the dimension weights set?
Weights are stage-aware. A pre-seed company is judged differently than a PE portfolio company. Financial Vitality dominates at growth stage but C-Suite is heavier at pre-seed. The weighting shown above is a labeled sample: the set of nine never changes, only the weights do.
What if a dimension has missing data?
Each metric has a confidence score, and the readout states a confidence percentage and a coverage tier for the whole. Missing data lowers confidence on that dimension but does not zero it out. Zoe falls back on adjacent signals (e.g., delivery velocity proxies for execution if sprint data is missing).
Can I see the raw data behind a score?
Yes. Every score is traceable down to the underlying metrics, sources, and timestamps. Zoe is auditable end-to-end. No black box.
Does the score keep updating?
The score is computed when you run the Zoe Diagnostic and delivered in the Readout Package inside 24 hours. At Close Case you choose: seal the case, with a destruction certificate at 30 days, or on the Portfolio package keep the target’s connections live and monitor.
What are the bands? What counts as a "good" score?
Healthy is 70 to 100. Attention is 50 to 69. Critical is below 50. A band names how seriously to take the number, and the word always travels with the color.
Nothing held back.
A diagnostic that hides its evidence is just an opinion. The example readout carries the score, the findings, the IC memo, the signed opinion and every dated pull underneath them.