How to Import Leads into Pipedrive Without Creating Duplicates
Pipedrive's import is quick and its data model is opinionated. If your file doesn't match the shape it expects, it won't refuse — it'll create records that technically exist and are quietly wrong.
The duplicates are the part people notice a month later, usually when two salespeople ring the same business in the same week.
Pipedrive's Data Model Is the Whole Story
Pipedrive has four objects: organisations, people, deals, and leads. Everything about a successful import comes from mapping your file onto the right ones.
An organisation is a business. A person is a human who belongs to one. A deal is a specific opportunity with a value and a pipeline stage. A lead is an unqualified prospect that sits in the Leads Inbox until you promote it.
A scraped local business list is organisations. Possibly organisations plus leads. It is almost never deals — importing prospects straight into your pipeline as deals inflates your forecast with businesses that have never heard of you, and it's a habit that takes months to unwind.
The Leads Inbox is the right destination for cold prospects. It exists precisely so unqualified contacts don't pollute pipeline reporting, and promoting a lead to a deal later is a single click.
Prepare the File First
Pipedrive will import what you give it, so the cleaning has to happen upstream.
Phone numbers in E.164, so Pipedrive's caller integration works and the same number is never stored two ways. Websites normalised and stripped of tracking parameters. Empty cells actually empty. Business names trimmed of the page-title junk — "Bright Smile Dental Practice | Leeds | Book Online" is a title tag, and if it goes in as an organisation name it's on every email you send.
Deduplicate before import rather than after. Pipedrive's merge tool is decent but it's manual, and merging four hundred organisation pairs by hand is a day you won't get back. The deduplication guide covers matching on identifiers and normalised domains rather than raw names.
How Pipedrive Decides Something Already Exists
This is where the duplicates come from, and the rules are not the ones most people assume.
Pipedrive's import documentation sets out different criteria per object. For organisations, the field it uses to identify a duplicate is the address — its own guidance is that you can have organisations with the same name as long as you always specify the address field. For people, it looks at organisation, phone, and email, and treats a match on any one of those as a duplicate. For deals there is no duplicate identifier at all, so importing three deals with the same name gives you three deals.
Two things follow. The first is that the address field is doing far more work than its position in the form suggests. If it's blank, two identically named organisations are simply two records. Populate it, and populate it consistently, even when you don't otherwise care about the address.
The second is that Pipedrive's documented criteria don't include website or domain for any object. Domain is the most reliable key you have for local business data, and it isn't one Pipedrive will use on your behalf. Deduplicating on a normalised domain is therefore work you have to do in the file, before importing.
The importer does let you decide what happens when it finds a duplicate — recognise and merge them, or bypass the merging step and create separate records. Choose deliberately rather than accepting whatever is preselected, particularly on a top-up import into a CRM your team has already been working in.
The Field Mapping
Organisation name, address, and phone are straightforward. Two things are worth attention.
Pipedrive's address field is a single combined field with optional structured subfields. If you have address components in separate columns — street, town, postcode — map them to the structured subfields rather than concatenating them into one string. Structured data is filterable; a concatenated string is not.
Website isn't a default organisation field in all Pipedrive configurations. If it's missing, create it as a custom field before importing rather than mapping the URL into a notes field, where it's useless for filtering and matching.
Custom fields must exist before the import runs. Create them first, then import — the mapping screen won't offer you a field that doesn't exist yet.
Import in the Right Order
If you're importing organisations and people together in one file, Pipedrive handles the association automatically as long as both are in the same row.
If you're importing them as separate files, do organisations first. People reference an organisation by name, and if the organisation doesn't exist yet Pipedrive creates one from that name — usually a subtly different one from the record you're about to import, which is another duplicate.
Test With Ten Rows
Same advice as any CRM import, for the same reason. Cut ten rows into their own file, run it, and look at what arrived: are organisation names clean, did the address subfields populate, did the phone number format survive, did anything land in the wrong object?
Deleting ten test records takes a minute. Unpicking three thousand takes an afternoon.
Exporting a Pipedrive-Ready File
LeadSnipe has a Pipedrive export preset that names columns to match Pipedrive's fields and applies the formatting first — E.164 phones, normalised websites, trimmed names, consistent empty cells — plus a non-destructive dedup pass on place identifier and domain, which matters here precisely because those aren't keys Pipedrive checks itself. The HubSpot guide covers the equivalent for HubSpot, and the clean CSV checklist covers preparing a file by hand.