Count customers before publishing a customers-served claim

Write down what the sentence claims

Copy the exact sentence and its placement. 'Trusted by 5,000 researchers', 'over 10,000 orders shipped' and 'serving laboratories since 2021' are different claims with different denominators, and a record that supports one does not support the others. Identify the unit the sentence implies — people, organizations, orders or years — before opening any database, because the definition decides which records are even relevant.

Check what the surrounding page adds. A customer count beside logos or institution names implies that those institutions are customers, which is an affiliation claim needing its own evidence. A count beside a research-only statement does not become more verifiable; it just acquires a second claim to check.

Define the unit, period and exclusions before counting

Write the definition as a rule someone else could apply: for example, distinct billing identities with at least one completed, non-refunded order, between two dates, excluding staff test orders. Fix the period — all time, or a stated window — and record the as-of date. A number without an as-of date will silently become a claim about the present for as long as the page stays up.

Decide the edge cases now rather than during the count: whether refunded orders count as served, whether free or replacement orders count, how wholesale accounts are treated, and whether a person ordering for two organizations is one customer or two. There is no single correct rule, but the published sentence must match the rule actually applied. 'Customers' and 'orders' are not interchangeable, and swapping them to get a larger number is the failure this worksheet exists to catch.

Reconcile profiles with actual purchasers

Shopify's customer documentation describes profiles arising from purchases, abandoned checkouts, mailing-list signups and manual addition. On that platform, the customer list is a contact database, not a register of buyers: a newsletter subscriber and a completed purchaser both hold profiles. Any count drawn from profiles must therefore be filtered to the store's definition of served — typically profiles joined to qualifying orders — rather than read off the profile total.

Deduplication needs equal care. The same person can hold multiple profiles from different email addresses or guest checkouts, and Shopify's merge function is intended for profiles of the same customer. Merge only where records show the profiles are the same person, and keep a note of what was merged; merging to tidy the list and merging to reach a number are different acts, and only the first is defensible. If your store runs on another platform, find its equivalent distinction between contacts and purchasers before trusting any dashboard total.

Organizations, precision and the honest number

Research-only merchants often sell to laboratories and institutions. If the claim is about organizations, count distinct purchasing entities from order records, and do not let one institution's five departments inflate the count unless the sentence says departments. If the claim is about people, an institutional account still resolves to the purchasers your records actually show. Pick the unit the sentence uses and stay in it.

Match the precision of the sentence to the reliability of the count. A deduplicated record that yields 4,830 qualifying customers as of a stated date supports 'more than 4,000' comfortably and 'exactly 4,830' only while the as-of date is stated. Rounding down within the supported figure is a reasonable way to keep the sentence comfortable at the moment of publication, but it is not a guarantee of continued accuracy: the defined qualifying population can shrink after refunds, corrections or further deduplication. Every rounded claim stays tied to its as-of date and defined population, and the count is refreshed or the sentence reviewed whenever qualifying records change — rounding alone does not keep a claim true between reconciliations.

The FTC's substantiation policy requires an objective claim's evidentiary basis to exist before publication. A customer count is an objective claim, and its basis is the reconciled record plus the written definition. Keep both; a number without its definition cannot be checked next quarter, and a definition without the underlying records is just a plan.

Publish, date and assign the refresh

Publish the sentence with the supported unit and, where useful, the as-of date or period. Record the definition, the source system, the reconciliation date and the owner who will re-run it. Set the refresh condition: a scheduled re-count, or a review whenever the figure is reused in a new placement, since a claim copied from the homepage into an advertisement inherits the old number without the old check.

If the reconciliation shows the current public figure cannot be supported, correct or remove it while the gap is resolved, and do not replace it with an estimate presented as a count. All identifying records stay inside the store system; the worksheet holds definitions and totals, not customers. For a Prism website-review consultation, bring the sentence, its placement and a non-personal summary of the counting method; scope, responsibilities, fees and terms are agreed before work, and the review addresses the public claim, not your customer database.

Claim denominator worksheet

Complete one worksheet per distinct customer-count claim. Fix the definition first, then reconcile. The worksheet records definitions, rules and totals — never customer identities.

Worksheet entries are not submitted by Prism’s worksheet and are not saved by the site. Use record types, availability, anonymized observations, or match/mismatch results. Do not enter government identifiers, customer names or addresses, customer messages, receipt-access links, card or bank details, passwords, or keys. Send sensitive documents only through the provider’s verified secure channel.

These are temporary notes. Leaving or reloading this page may clear them. Worksheet entries are not sent automatically. If you copy notes into the consultation message and submit the form, Prism receives them as part of your request.