Shopify Lead Scoring [534K-Store Study]

TL;DR

• The April 21, 2026 extraction covers 534,514 stored domains with recorded estimated traffic and a latest stored snapshot.
• 198,771 domains had a stored score of 95+; 67,228 met the combined 50K+ estimate, 95+ score and stored verified-status gate.
• These are heuristic distributions and filter intersections, not validated probabilities of replies, budget or purchases.
• The proposed 100-point worksheet below is separate from the measured StoreInspect score and needs offer-specific testing.

What the measured score means

The retained query and summary read each store's existing stored Lead Fit score; they do not calculate a new score. The exact historical formula, weights, version, saturation and input ages are not retained with this extraction. A current formula cannot retrospectively establish which version produced every April value.

Traffic, app depth and pixel depth overlap with the kinds of signals used in a fit heuristic. Their relationship with score is partly circular, not independent proof of lead quality. The 95+ group's 99.5 mean suggests heavy saturation near the ceiling, which limits ordering within that group.

The query replaces missing app/pixel lists with empty lists and selects the latest snapshot without an age ceiling. Contact coverage uses at least one stored contact, not a currently reachable inbox. Verified coverage requires a contact email entry marked verified, without a separate filled-in-value, age, suppression or role requirement. Plus is a stored flag with missing values counted as no; it is not confirmed billing.

There is an additional query limitation: the displayed bucket rule falls through to “95–100” for a missing raw score, while the selected score value is set to zero. The retained output does not report how many missing scores existed. Read the bucket table as the historical extraction's classification, rather than a guaranteed complete ordering of known scores. Explicit intersections below that require a score of 95 or more are separate gates.

Historical bucket composition

The 85–94 group has 43.0% stored verified coverage, above the 40.8% in 95–100. A higher score therefore does not monotonically improve this contact metric. Thresholds of 85 or 95 can be trial filters, but the study does not establish universal cutoffs or justify deleting lower-scoring real businesses.

Traffic and score

The 50K+ combined group averages 96.9 versus 60.2 under 50K. This describes the stored heuristic and estimated traffic distribution. Traffic is not a merchant's revenue, margin or ability to afford a retainer.

Offer-specific research pools

Each “no covered match” combines selected app categories and vendor slugs. For example, the analytics filter covers analytics/attribution categories and Triple Whale, Northbeam, Elevar and Littledata identities; lifecycle includes email categories plus Klaviyo, Mailchimp, Omnisend, Privy, Attentive, Postscript, Drip and Iterable. Native Shopify functionality, backend integrations, custom workflows and unsupported signatures may be absent from this detector.

These rows overlap. They define questions to investigate, not proven problems. For an analytics candidate, inspect the current consent and measurement setup with permission. For lifecycle, inspect actual signup and message flows. For CRO, inspect the intended offer logic. For design, identify a specific storefront task: a free theme and a Plus flag alone do not prove the design is inadequate. A subscription signature does not establish a recurring-revenue amount or an unfinished retention program.

How gates change the shortlist

The final row has 100% stored verified status because that status is a selection condition. It is not evidence that scoring improved deliverability. No retained output reports the count after current verification, bounce/unsubscribe suppression, unique-person review and offer-relevant role checks.

An untested 100-point worksheet

This is an editorial starting template, not StoreInspect's historical formula. Assign each criterion once, record unknown inputs separately, and do not treat a detector-negative category as confirmed pain.

The weights and any negative penalties are hypotheses. An absent contact should block a send, rather than be offset by app points. A conflicting geography or product fit should trigger review rather than an automatic budget assumption. Keep unknown values visible so incomplete data does not masquerade as a weak business.

To calibrate, compare the worksheet with a simple category/traffic baseline on held-out accounts using the same offer and outreach process. Record qualified replies, meetings and opportunities per researched account. Predefine cutoffs and monitor false exclusions before treating a score as a prediction.

Freshness and next action

Record the score version, detector version, latest snapshot time, contact verification time and any change-event interval on each current shortlist. The April extraction does not retain their age distributions or measure decay. Monthly review may be a starting policy; set the actual interval from observed changes and the time sensitivity of the offer.

Use lead qualification to inspect the account, then contact-quality checks before sending. A useful scored account has a relevant current task and a reviewed contact; the numeric value determines research priority only after its limitations are understood.