Importing old records: what to clean first
Photo: Avila Ops Tecnologia
Before importing old records, clean seven things: duplicates, inconsistent names, empty required fields, mixed formats (phone, ZIP code, amounts), test records, inactive records, and data you no longer have a reason to keep. In that order. Dirty imported records are dirty records with a higher cost.
For small businesses, old records accumulate years of rushing: customers noted in WhatsApp, products entered twice under different names, phone numbers with and without area codes. None of this hurt in the spreadsheet because spreadsheets accept everything. The system will reject it or, worse, accept it and multiply it.
The seven problems, in cleaning order
| Order | Problem | What it looks like | What to do |
|---|---|---|---|
| 1 | Test records and junk | “Test,” “asdf,” customer with phone 000 | Delete |
| 2 | Inactive records | Product unsold for two years, customer with no purchases for over two | Archive outside the system; do not import |
| 3 | Exact duplicates | Same row repeated | Keep one |
| 4 | Different spellings | “Maria Silva,” “maria silva,” “M. Silva,” same phone | Merge into one, keeping all history |
| 5 | Mixed formats | Phones with and without area code, ZIP codes with and without hyphens, amounts “R$ 79,00” and “79” | Standardize each column to one format |
| 6 | Empty required field | Product without price, customer without contact, order without date | Fill in or decide to exclude |
| 7 | Data with no reason to retain it | CPF tax ID of someone who bought once in 2022, unused birth date | Do not import |
Order matters because each step reduces the next step’s work. Removing tests and archiving inactive records before merging duplicates can halve the list.
Find duplicates when names differ
Names do not identify people. Phones and email addresses do. Sort customer records by phone: two names with the same number are the same person. Then sort by email. Records remaining without either are customers you cannot contact, raising the question of whether to include them.
For products, the identifier is the barcode or internal code. Without a code, use name plus variant. “White mug 300ml” and “Mug 300 ml white” are the same product, and the system will create two unless you merge them first.
One limitation: merging requires deciding which record is primary and what happens to the other’s history. Orders from “M. Silva” must move to the merged “Maria Silva,” not disappear.
What to exclude, and why it is a decision
Item 7 is often skipped, yet is the most important. The LGPD requires a purpose for all personal data and retaining only what that purpose needs (article 6). A CPF from a one-time buyer three years ago, with no invoice retention requirement, is unlikely to have a purpose. Importing increases what you must protect.
A practical rule: for each personal data column, ask, “What do I use this for today?” If the answer is “It might be useful someday,” do not import it. Keep it in the frozen spreadsheet archive and import only what is used.
The same applies to inactive records. Customers with no purchases for over two years become unused contacts appearing in every search and count. Archiving is more honest than importing.
What to do now
Export your customer list and sort by phone. Rows with repeated phone numbers measure the cleaning still needed. If they exceed 10%, clean before any import. Avila Ops includes cleaning in the migration scope, delivered within 48 hours: avilaops.com/sistema-para-pequenas-empresas or contact us on WhatsApp: wa.me/5517997811471.
Frequently asked questions
How do I clean duplicate customer records?
Sort by phone and email, not name. Records with the same contact are the same person. Choose one as primary, move the others’ history into it, and delete the rest. Do this in the spreadsheet before importing.
Must I import every old customer?
No. Import customers who bought in the last 12 to 24 months and those with active relationships. Archive the rest in the frozen spreadsheet. Smaller clean records are more useful than large dirty ones.
Can I import CPF numbers and personal data from old records?
Only what has a current purpose: invoicing, delivery, or service. A CPF for someone who has not bought in years, without a tax retention obligation, should not be imported. Under the LGPD, data without a purpose should not be retained, and the new system is a chance to reduce that risk.
Turn what you learned into your next step.
Tell us about your company and what you want to improve. We can explore the right tools, integrations or custom project together.