Good recruitment data helps consultants identify candidates, progress placements and maintain relationships. It also gives managers a reliable basis for reporting. Use parsing and connected communication to reduce repeated entry, then address gaps and duplicates that affect matching or business decisions. Data quality is not a one-off clean-up. Availability changes weekly, documents expire and contacts move roles, so the aim is a set of habits that keep the records people rely on accurate enough to act on.

Prioritise the information people use

Agree the fields that support sourcing and booking: relevant experience, availability, contact details and role requirements. For reporting, agree status definitions and how ownership, dates and financial values are recorded.

Not every field deserves the same effort. A record with dozens of optional fields, half of them blank, is harder to use than one with a few that are always right. Group the information and set a standard for each group:

Matching fields: grade, speciality, relevant experience, shift and location preferences, availability and expected rate. If these are wrong, the right nurse or locum doctor never reaches the shortlist.
Contact and compliance fields: phone, email, professional registration details and document expiry dates. Errors here delay bookings or put an unready candidate in front of a client.
Reporting fields: status, owner, key dates and placement values. Errors here distort conversion, revenue and margin.

Use fixed value lists for grades, specialities and statuses wherever possible. Free text produces "HCA", "healthcare assistant" and "care asst" for the same role, and a search for one misses the other two. Keep compulsory fields to what each stage genuinely needs. Making everything mandatory teaches people to type placeholder values, which is worse than a blank because it looks complete.

Capture information where work happens

Let CV parsing capture profile information and connected communication retain the conversation history. Correct gaps or inaccurate details that affect matching, and approve relevant suggested call updates. Keep the resulting record available to the placement workflow.

The further information travels from the conversation it came from, the more likely it is to be lost. A consultant who learns on a call that a candidate has stopped working nights, and writes it on a notepad to update later, is relying on memory at the busiest point of the day. Capture it on the record during or straight after the conversation, or let the system propose the update from the source.

In RecSphere, call review suggests record updates from recorded calls, and a recruiter approves each one. Treat parsed CVs the same way: the output is a draft, so check the fields that drive matching and correct anything misread before the profile is used for a shortlist.

Refresh the information that decays

Some information goes out of date on its own. Plan for it rather than waiting for a consultant to discover it in the middle of a booking:

Availability: ask at every contact, and run regular check-ins by text with candidates who work shifts.
Compliance documents: track expiry dates and request renewals before they lapse, not when a shift is offered.
Client contacts: when a ward manager or staffing lead moves on, record the change and keep the history rather than overwriting the old contact.
Rates: review client rate agreements when they renew, so new bookings do not carry an outdated charge.

Handle duplicates carefully

Review likely matches before merging or removing records. Preserve the contact, compliance and placement history required for ongoing work and the agency's record-keeping process.

Duplicates have predictable causes. A candidate applies through two job boards with different email addresses, two branches register the same nurse, or a client contact is added again after changing roles. Find them by comparing several details, such as mobile number, email and date of birth. Two care workers can share a name, so a name match alone is never enough to merge. Before merging, check:

Which record holds the most recent contact details and availability.
Where the compliance documents, references and their history sit.
Which record the placements, timesheets and invoices are attached to.
Any communication preferences the candidate has given.

If in doubt, flag the records for review rather than merging them. A wrong merge is harder to undo than a duplicate is to live with.

Make ownership clear

Assign responsibility for correcting errors and maintaining shared values. A consultant should know where to raise an issue rather than creating another personal spreadsheet.

Ownership works at two levels. Each record has a consultant or team responsible for keeping it current. Each shared list, such as grades, specialities, client sites and compliance document types, has one person who approves changes. Without that second owner, lists grow by accident: one consultant adds a new spelling of a speciality, another adds a duplicate site, and matching and reports quietly fragment.

Make raising an issue easier than working around it. A short form or a shared channel watched by the list owner is enough, as long as issues get fixed and the person who raised them hears back.

Write a short data standard

A one-page standard stops much of the inconsistency at the point of entry. Keep it short enough that people read it, and cover:

How client names, sites and departments are written, so one hospital is never recorded three ways.
One format for phone numbers, including the country code for agencies working across the UK and Australia.
Dates in day, month, year order.
What each status means and the event that changes it.
Which fields must be complete before a candidate can be submitted or booked.

Use it when inducting new starters, and update it when the checks below reveal a recurring problem.

Check the reports against records

Select a small set of active candidates and placements each month. Check whether the reported status, owner and value match the underlying work. Use recurring errors to improve the workflow.

Keep the check small enough that it actually happens. For example, open a handful of active candidate records and ask the owning consultant whether the availability, status and preferences are current. Then take a few live placements and compare the value in the report with the approved timesheets. Record each error and its cause. If the same field is wrong on several records, the cause is usually the workflow rather than the individual, and the fix belongs in the process or the standard.