Data & Import

Data health score

How LiftSignals measures the completeness of your customer data and why it affects model accuracy.

What is the data health score?

The data health score (0–100) measures the completeness and quality of your customer records. A higher score means LiftSignals has more information to work with and can produce more accurate predictions.

Accessing your data health score

Go to Settings → Data → Health. Your score is shown with a breakdown by field.

What affects the score?

FieldImpact on modelsHow to improve
Email addressHigh — required for customer identificationUsually already high from Shopify
Last order dateHigh — primary recency signalUsually high from store connection
First order dateMedium — affects CLV and cohort modelsRe-import with first_order_date column
Phone numberMedium — needed for SMS exportsCollect at checkout; import via CSV
Product categoriesMedium — powers propensity and affinity modelsEnsure products are categorised in your store
Order countHigh — frequency signal for RFM and churnUsually high from store connection

Health score thresholds

  • 85–100 (Excellent) — All models perform at full accuracy
  • 70–84 (Good) — Most models work well; minor gaps in some predictions
  • 55–69 (Fair) — Some model accuracy degradation, especially CLV and propensity
  • Below 55 (Poor) — Significant data gaps; scores should be treated as directional only
Was this article helpful?