Role Guide: Data Entry and Data Quality Specialist
For Narqix specialists who enter, check and clean client data. After reading you will know what accuracy really means at Narqix, how your work is measured, and how to prove your numbers are right.
Track: Back Office & Admin · Role: Data Entry and Data Quality Specialist · Time to complete: 15 min
What this role is
Most people think data entry is typing. It is not. Data entry is moving the truth from one place to another without changing it. The typing is the easy part. The real job is proving that what you entered is correct.
A data quality specialist goes one step further: you find what is already wrong (duplicates, bad formats, missing fields, impossible values) and fix it safely.
Clients pay for this role because bad data is expensive. One wrong invoice amount can mean a wrong payment. One wrong phone number is a lost lead. One wrong status can trigger the wrong automation for a whole list.
What you own
- Entering records from source documents (forms, PDFs, invoices, spreadsheets, scanned papers) into the client's systems.
- Checking every batch with a structured method before you call it done.
- Cleaning lists: formats, duplicates, missing fields, stray values.
- Keeping a Held list of anything you could not enter with certainty, with the reason.
- Reporting your counts and your measured error rate.
- Spotting work that should be an import or an automation, and raising it.
Your daily rhythm
| Shift time | What you do |
|---|---|
| Start | Check in. Read instructions and any answers to yesterday's held items. Post start of day report with today's batch targets. |
| First block | Highest-risk data first (money, legal, anything with a deadline). Fresh eyes catch more. |
| Every 25 to 50 records | Stop and run the batch checks (see Data Entry Accuracy). Do not carry an error forward. |
| Mid shift | Logged break. Send held questions to the client in one batched message. |
| Second block | Lower-risk entry and cleaning tasks. |
| End | Comparison pass on today's critical batches, counts reconciled, end of day report with numbers. |
How you are measured
| KPI | Target |
|---|---|
| Field accuracy on financial data (amounts, account numbers, totals) | 100% |
| Field accuracy on contact data (names, phones, emails, addresses) | 99.5% or more |
| Field accuracy on status, tags and categories | 99% or more |
| Batches with record count matching source | 100% |
| Held items with a written reason | 100% |
| On-time batch completion | 95% or more |
| QA score on sampled batches | 90% or more |
How field accuracy is calculated: correct fields divided by fields checked. If QA checks 40 records with 10 fields each, that is 400 fields. Two wrong fields means 398 out of 400, which is 99.5%. Three wrong fields is 99.25%, which misses the contact data target.
Coach's note: speed is never your first KPI. A specialist who enters 300 records at 100% is worth far more than one who enters 500 records at 98%, because the client must recheck all 500. Accuracy first, then speed comes with practice.
Tools (generic)
Spreadsheet, CRM, accounting system (entry access only), shared drive, scanned document viewer, task tracker, team channel, password manager and your campaign's time tracker.
Handoffs
- You receive: source files from the client or the client's team, instructions with a field map, answers to held items.
- You pass on: completed batches to the client or the system owner, held lists with reasons, a weekly data quality note (what kept going wrong at the source and how to prevent it).
Quality standards
- Enter from the original source. Never from memory, never from a copy of a copy.
- Write the setup first. Source, destination, field map and "done" definition at the top of every task.
- Batch and check. Groups of 25 to 50 records, checked before you start the next group.
- Double entry for anything financial. The Double Entry Comparison method is mandatory for money and account numbers.
- Never guess. Unclear handwriting, missing digits, conflicting values: put it on the Held list.
- Count everything. Records in equals records out, plus held.
Escalation rules
- Stop and tell the client and your lead within the hour: you find you entered a batch into the wrong place, you see an amount that looks like fraud or a duplicate payment, or source files contain data you should not have (for example full card numbers).
- Batch into one message: held items and format questions.
- Weekly report: recurring source problems.
Never do this: never "fix" a number in a financial record because it looks wrong. You do not change source truth. Enter exactly what the source says, flag the item as held, and let the client decide.
Common mistakes
| Mistake | Result | Fix |
|---|---|---|
| Visual checking only | Misses most real errors | Use double entry or control totals |
| Leading zeros lost from ZIP codes or IDs | 02134 becomes 2134 | Format the column as text before pasting |
| Mixed date formats | 03/04 means March 4 to a US client and 3 April to a UK client | Agree one format; write months as words when unsure |
| Extra spaces | Duplicates that look different, broken matching | Trim spaces in every batch check |
| Correcting "obvious" typos in names | You change a real spelling | Enter as written; flag if unsure |
| Working late at night on money batches | Fatigue errors | Do critical data in your first block |
Check yourself
- QA checks 50 records with 8 fields each and finds 3 errors. What is your field accuracy, and does it pass for contact data?
- A receipt total is unreadable in the scan. What do you do?
- Why is "I looked it over" not a valid check?
- Which data type requires 100% accuracy, and which method do you use for it?
© Narqix Academy