Data-Readiness Checklist Before Adopting an Operations System

Why Prepare the Data First

An operations system is only as good as its data. Import duplicated, incomplete, inconsistently formatted records as-is and the new system will simply show the same mess faster — and store staff will lose confidence quickly. Cleaning data before go-live is usually far less work than fixing it afterwards. Start by giving each data area an owner and a definition of done:

Customer Records and De-duplication

• Normalize formats: strip spaces and symbols from phone numbers, make full-width and half-width characters consistent, trim extra spaces from names, and use one date format.
• Flag candidates, confirm by hand: for example, treat records sharing name + mobile or name + date of birth as possible duplicates, and have staff who know the customers confirm before merging. Don't merge fully automatically.
• Keep the source: when merging, record the original store and the legacy system ID so you can trace records later.
• Watch for shared phones: older customers often give a son's or daughter's number, so one number may belong to several customers.
• Bring only what you need: use the migration to decide whether data you no longer use should be kept at all.

Audiograms and Test History

Test records are often spread across paper, scanned files and audiometry software. Before go-live, take stock:

• Every record needs at least: customer, test date, tester, ear (left/right) and test type (pure-tone, speech, etc.).
• Label structured data (thresholds by frequency) separately from images (scans or photos). Images can be kept intact and linked to the customer first; there's no need to convert them all to numbers right away.
• Decide how much history to bring: everything, or only the most recent tests — decided with the audiologists based on clinical need.
• Spot-check: after import, have audiologists compare a random sample against the originals.

Device Serials and Inventory

• Every device: serial number, left or right, customer, fitting date, warranty expiry and repair history.
• Physical count: each store counts stock and reconciles it with the books; label trial, demo and loaner units with their own status.
• One consumables catalogue: standardize item names and units so the same item isn't called different things in different stores.
• A final count before cutover: this becomes the opening balance in the new system.

Staff Roles and Access

• List every user: name, store and role (audiologist, store staff, store manager, HQ).
• Grant access by role on a least-privilege basis: store staff see their own store's customers, HQ sees cross-store summaries. Restrict data-export rights tightly.
• One account per person, no shared logins — so you can tell who did what, and when.
• Leavers and transfers: define who disables or changes an account, and when.

Consent Records

Under Taiwan's Personal Data Protection Act (PDPA), when collecting personal data you must inform the individual of matters such as the purpose, the categories of data, how it will be used and their rights, and use it within the scope of that purpose. Hearing test results relate to health and may fall under stricter rules. Before go-live, review:

• What existing customers signed — content, version and date — and whether it covers how the new system will use the data (for example cloud storage or sharing across stores).
• Whether marketing contact (SMS, calls, messaging apps) has its own consent, and whether it reliably stops when a customer objects.
• How consent is stored: paper forms must be traceable to the customer; electronic consent should keep the time and version.
• Who receives requests to access, correct or delete data, and how they're handled.

Start With a Pilot Store

• Pick one or two stores: with representative workflows and a manager who is willing to invest time.
• Write down success criteria first: for example data completeness, daily use by staff, and trials and follow-ups handled in the system — stated in measurable terms.
• Set a cutover date: decide the day from which new data goes only into the new system, so you don't end up with two half-complete records.
• Review weekly: collect issues from the floor, adjust fields and workflows, and expand only once things are stable.
• Reuse the checklist: apply the pilot's data-cleaning method and this checklist to each new store.

In the stores, SONUS AI handles customers, appointments and trials; at HQ, SONUS MANAGE handles staff, inventory and reports. Prepare the data first and both start on day one with records you can trust.