Two sources, one path
Every migration batch starts in Customer migration from either Search & filter Shopify (read the current store) or Upload a file. The path is the same after that: choose → group → review → import → result. The step strip names them for you — Choose customers, Group, Review, Import, Result, or Upload, Map columns, Group, Review, Import, Result when you start from a file.
Search Shopify customers
Read directly from Shopify builds a selection from supported customer fields. Combine native fields with AND, OR, NOT or exactly-one groups, and add nested groups for complex selections.
- Customer tag (with the tag suggestions read from your store), Email, First name, Last name
- Address country code, Address company name
- Native company membership — No company yet, Has a company or Multiple companies
- Existing company ID, Assigned purchasing location ID
- Number of orders, Total spent (shop currency)
- Customer created date, Customer updated date
- Selected grouping metafield value, reading one
customer.metafields.namespace.key
Each condition joins one of the operators the field supports — is, is not, contains, does not contain, starts with, ends with, has a value, is empty, and, for numbers and dates, at least / at most / on or after / on or before. Add conditions with Add condition, nest a level with Add subgroup or Add OR group, and every group chooses how its own children combine: all (AND), any (OR), none (NOT) or exactly-one. One search accepts up to 64 conditions and subgroups in total, at most 6 children in one group, and at most 4 nested levels; a search expression that grows beyond that is refused with Use up to 64 conditions/groups and four nested levels. Native conditions narrow the Shopify query and the app also evaluates the whole expression on the returned rows.
Three Quick filter presets save typing: Wholesale or B2B tags, Customers without a company and Existing company buyers.
Select Find matching customers to run the search — it is explicitly a read-only step, and the panel says Read only · select customers before preparing a plan. The result table lists Customer, Existing company and Tags, twenty rows per page with Previous / Next, and a count line such as "12 matches in 340 scanned customers. This is a bounded source batch, not a store-wide count."
A search scans at most 500 source customers and a selection cannot produce more than 500 rows; the app refuses a larger set with Selected memberships produce more than 500 rows. Select fewer customers. When the scan stopped at the limit you also see The first 500 source customers were scanned. More exist; narrow your filters and use separate batches. An incomplete page is not a store-wide count, so do not read a truncated list as "all wholesale customers".
Upload a CSV or XLSX file
A file must be UTF-8 CSV or XLSX, at most 5 MB, with a header row and 1–500 data rows. Header names must be non-empty and unique, and formulas are rejected — export plain values. Files larger than 500 rows must be split into batches. The drop zone states the same limits, and there is a Download button for a CSV template built from the current field list.
After upload, Map columns matches source headers to app fields and shows real sample values from your first source row so you can verify each mapping before grouping. Mappings are reviewed one category at a time, in four groups:
- Buyer identity — existing customer GID, email, first and last name, new customer external key, and the column that marks a row as a new customer
- Company & purchasing access — company grouping key, company name, existing company GID, location grouping key, location name, existing location GID, role GID
- Location address — country code, street address, address line 2, city, province or state code, postal code
- Existing business context — existing tags, legacy workflow dependencies, legacy pricing agreement, legacy currency
Each field can be set to Do not import, and the category header counts how many of its fields are mapped so you can see at a glance what is still unmatched. Columns you leave unmapped are never rewritten in Shopify: the saved source keeps the original values, so marketing consent, lifetime spend and similar export columns are not replaced by this import. Select Apply mapping and continue to turn the file into reviewed rows.
Each row keeps its source row number. When a row is blocked during review, use that number to correct the file or the mapping instead of guessing an identity.
Two other ways to start
Start without a file → Add a customer opens a single empty row for one buyer you already know. Nothing about it is saved until you group and review it like any other row.
The optional hosted application page is a separate intake path for prospective buyers and is not part of this release. Buyer applications are optional, not automatic access explains what an approval does and does not grant.
What the source does not decide
A customer ID in a file is treated as a claim, not proof. Existing native records are reused only after the app revalidates them, and a customer who is already a buyer of that same company is simply reused — the review marks it Reuse company and Reuse location rather than creating a duplicate. Renaming a company in the Group step cannot move an existing membership; that is an access change for Companies & locations. What is blocked is one customer being assigned to two different companies, and two rows claiming the same email with different external keys.
Group companies, locations and buyers covers the next step.
Frequently asked questions
Can I bulk import Shopify B2B companies, locations, and contacts from CSV or XLSX?
Yes. Kaupvik B2B Company Onboard supports uploading UTF-8 CSV or XLSX files of up to 500 rows and 5 MB per batch, with a four-category column mapper for Buyer identity, Company & purchasing access, Location address and Existing business context. Unmapped columns are left unused, and every row can be edited inline before any native Shopify record is written.
What file limits apply to a Company Onboard upload?
Each upload must be a UTF-8 CSV or XLSX file of at most 5 MB with a unique, non-empty header row and 1–500 data rows; larger files must be split into separate batches, and formula cells are rejected so plain values must be exported first.
How many customers can one Shopify customer search select?
One search scans at most 500 source customers and a selection may not produce more than 500 rows, so a bigger store needs narrower filters and separate batches rather than one attempt to read the whole customer base.
What happens to columns I do not map in the file?
Columns you set to Do not import are not used and not written back, so values such as marketing consent or lifetime spend in your Shopify customer export stay exactly as they are in Shopify.