Before You Import 1,000 Leads: The CRM Data Import Checklist
The spreadsheet looked ready: 1,046 rows, green email-validation marks, and a “status” column. Then the sales manager asked: Which rows are companies and which are people? Does “active” mean researched, assigned, or contacted? If two people share a domain, are they one account or two leads?
Nobody could answer without reopening the source files. That is the point of a CRM data import checklist: proving another salesperson can understand each record after the researcher is gone.
SaleAI CRM Management provides visible controls for contact and company records, tags, importing, exporting, assigning data, choosing columns, and filtering records. Those controls are useful after the file has a clear structure. They do not remove the need to decide what a row represents, which field is authoritative, or whether a lead is ready for sales action.
This guide makes those decisions before a large upload. The 12-row test, mapping sheet, stop rules, and post-import checks are editorial methods, not automatic SaleAI behavior.
The 90-second answer
Before importing a B2B list, split company facts from contact facts, choose stable identifiers, define blank rules, use reusable tags, and assign an owner only when a next action exists. Load 12 normal and awkward test rows. Then use SaleAI's filters, column settings, record views, and assignment controls to check the result.
Do not start with all 1,000 rows. A clean small test tells you more than a large successful upload notification.
What the SaleAI CRM screen actually showed
A read-only inspection of the logged-in SaleAI backend on August 14, 2026 found Contact Management, Company Management, and Tag Management. Contact Management displayed Help, Data Import, Data Export, Data Assignment, Column Settings, Add Contact, several filter types, and common owner and contact fields. Company Management followed a parallel account-level pattern. Tag Management showed reusable tags with names, colors, timestamps, and edit or delete actions.
The inspection did not establish the items in the right column below.
| Verified in the SaleAI interface | Not verified as an automatic SaleAI capability |
|---|---|
| Separate contact, company, and tag areas | Guaranteed duplicate prevention |
| Data import and data export entries | Automatic contact-to-company matching for every file |
| Data assignment controls | Automatic owner selection |
| Column settings and several filter types | A universal field-mapping template for every source |
| Owner, company, email, phone, social, stage, tags, and last-contact fields | One-click rollback of an incorrect import |
This boundary matters. If a product demo later confirms more import behavior, document it then. Until that point, use a reversible test and inspect the result.
First decision: is the file allowed to enter CRM?
The most expensive import errors are often visible before anyone touches the Import button. Start the CRM data import checklist with a stop gate, and halt the job when any of these conditions is true:
- A row mixes a company name, a person's email, and a status whose meaning is undocumented.
- The file cannot distinguish an empty value from “not checked” or “not applicable.”
- Several researchers used different spellings for the same market, source, or stage.
- A tag contains a one-off search query that will never be reused.
- No one owns the import, the review, or the correction file.
- The list contains personal data that has no defined business purpose or retention decision.
- A “qualified” label exists without the evidence used to reach it.
The UK Information Commissioner's Office explains data minimisation as keeping personal data adequate, relevant, and limited to what is necessary. It also treats accuracy as an ongoing responsibility. Those principles are external governance guidance, not SaleAI product settings, but they are a sound reason to remove speculative personal fields before uploading a sales list.
Build two clean records, not one crowded row
A company can remain a target account after an employee changes roles. A person can move while the original account remains relevant. Treating the two as separate records preserves that reality.
Company record
Keep legal or trading name, domain, market, company type, source evidence, product relevance, owner, stage, and review date together.
Contact record
Keep name, work email, appropriate phone or public profile, role, employer reference, owner, stage, suppression note, and review date together.
Connect the files with a shared value. A normalized domain is often steadier than a display name, but groups and regional sites may use several domains. Retain the source value for review.
HubSpot's official import documentation gives a useful general lesson: contacts and companies need a common column across files, while unique identifiers help avoid duplicates. It describes HubSpot, not SaleAI, so confirm SaleAI's template and association behavior in the rehearsal.
The field-mapping sheet that prevents quiet damage
Do not rename columns directly in the source workbook. The field-mapping section of the CRM data import checklist needs one line per source field. This becomes the decision record for the import.
| Source column | Intended SaleAI destination | Transformation | Blank rule | Reviewer |
|---|---|---|---|---|
business_name |
Company name | Trim spaces; retain original spelling in source note | Reject if blank | Research lead |
website |
Company/domain reference | Lowercase host; remove tracking path | Manual review if absent | Data reviewer |
contact_person |
Contact name | Preserve human-readable order | Keep unimported if unknown | Sales ops |
work_email |
Contact email | Lowercase; do not invent | Leave blank and mark email-unverified | Data reviewer |
source_type |
Tag or source note | Map to a controlled list | Reject undocumented values | Research lead |
lead_status |
Customer stage | Translate with an approved stage dictionary | Default to review queue, not qualified | Sales manager |
sales_rep |
Owner | Match an active owner list | Leave unassigned for review | Sales manager |
checked_on |
Review date or note | Convert to one date format | Reject if freshness is required | Data reviewer |
The mapping sheet should also state what will not be imported. Raw scraping notes, guessed seniority, personal email addresses with no approved use, and temporary calculation columns usually do not belong in a working CRM record.
For the file itself, follow conservative CSV conventions: one header row, one record per line, the same number of fields on each row, and quotation marks around values that contain commas or line breaks. RFC 4180 documents these common CSV rules and the text/csv media type. It does not define SaleAI's upload limits, so the live importer remains the authority for supported file types and size.
Run a 12-row rehearsal
Twelve rows are enough to expose most mapping mistakes while remaining easy to inspect manually. This is the failure-finding part of the CRM data import checklist, so do not choose the first 12 rows. Assemble a miniature test set deliberately.
| Test row | Why it belongs in the rehearsal |
|---|---|
| 1-3 | Ordinary complete company-contact pairs |
| 4 | Company with two contacts |
| 5 | Contact with no verified email |
| 6 | Company name containing a comma or quotation mark |
| 7 | Non-English characters in a company or person name |
| 8 | Missing domain but credible source evidence |
| 9 | Possible duplicate domain with a different display name |
| 10 | Unknown owner |
| 11 | Stage value that does not match the approved dictionary |
| 12 | Record that should be rejected, proving the stop rule works |
Keep the source unchanged. Date the test filename, record its row count, and name the approver. If a value changes, compare the source, prepared file, and imported result.
Inspect the result in SaleAI, not just the upload message
This sequence uses observed SaleAI controls. Confirm the active account's prompts, limits, and mapping screen before a production import.
Check the contact and company split
Open Contact Management and Company Management separately. Check that company facts did not land in person fields and that each contact still points to the right account. Use Column Settings to review the needed fields together.
Filter for blanks and edge cases
Use condition or advanced filters to find missing owners, missing companies, unexpected stages, and the rehearsal tag. Test one known positive and one known negative; a zero-result filter may simply be wrong.
Review ownership as a decision
Data Assignment should follow a defined next action. A record without one can remain in review; assigning every row merely to remove blanks creates false accountability.
Export the test cohort
Export the 12-row cohort and compare names, domains, owners, stages, tags, punctuation, and non-English characters with the prepared file. This catches a bad assumption before scale makes it costly.
A worked example: one account, three people, two source paths
An industrial sensor exporter finds NordWerk Process GmbH and three public professional profiles. The source rows use three company-name variants; two people share a domain, one lacks email, and the trade evidence is six months old.
The company record uses one normalized name and domain, retains the source task, adds market and product family, and sets the stage to “research reviewed.” It does not claim current purchase intent.
The contact records keep names, roles, available work details, and review dates. The person without a verified email stays in research; no address is invented. All three point to the same company identity.
The owner receives the account after a next action is defined: verify distributor fit and identify the buying group. One tag records the source family; dated evidence and uncertainty stay in notes.
After import, the reviewer filters the rehearsal tag and checks company, owner, stage, email, last contact, and tags. Unclear records trigger a mapping revision before the full import.
Scale in batches, with a correction path
Once the 12-row rehearsal passes, the CRM data import checklist moves from preparation to controlled scale. Import the larger list in named batches. A batch might represent one market, one source cohort, or one review date. Keep it small enough that a correction does not require rebuilding the entire CRM.
For each batch, record:
- source file and prepared file names;
- row count submitted and row count accepted;
- mapping version;
- owner and approval time;
- controlled tag used for the batch;
- known exclusions;
- exported verification file;
- issues found and the correction decision.
Do not assume an import can be rolled back automatically. The August inspection did not verify a rollback control. A clean source file, a small batch, a batch tag, and an export are the safer correction path until account-specific behavior is confirmed.
What good looks like 48 hours later
Import quality is not finished when the rows appear. The last part of the CRM data import checklist happens two days later, when the team asks whether the data supports real work.
First, can a salesperson explain why an account is present without opening the original research task? Second, can the team filter for unassigned or unreviewed records? Third, are stages describing an agreed sales state rather than the mood of the importer? Fourth, do tags group records in a way that will still be useful next month? Finally, can a contact change roles without erasing the company history?
If the answer to any of these is no, pause the next batch. More rows will not repair the model.
SaleAI brings the relevant operating surfaces together: research outputs can be retained in Data Assets, reviewed contacts and companies can be organized in CRM Management, and downstream work can be separated from the research stage. The B2B lead source tracking article explains how to preserve provenance. The B2B lead qualification article covers readiness decisions, while B2B data revalidation is useful when an older list needs a second look.
The practical rule is simple: import only the structure your team can explain, test, filter, assign, and correct. That is what turns a CRM data import checklist from a spreadsheet exercise into an operating safeguard.
Teams evaluating the workflow can review SaleAI CRM, compare SaleAI pricing, and request a walkthrough focused on the exact import template, supported identifiers, association behavior, duplicate handling, and correction options required by their data.
Frequently asked questions
What is a CRM data import checklist?
It reviews record types, identifiers, mappings, blank rules, stages, owners, tags, test rows, and post-import checks so loaded data stays understandable and correctable.
Which SaleAI CRM functions were verified?
The August 14, 2026 backend inspection showed Contact Management, Company Management, Tag Management, Data Import, Data Export, Data Assignment, Column Settings, multiple filter types, and common ownership and contact fields.
Does SaleAI automatically remove duplicate contacts or companies?
That behavior was not verified. Use stable identifiers, test duplicate edge cases, and confirm account-specific handling before a large import.
Should contacts and companies be kept in separate files?
Separating them during preparation makes mapping easier to audit. The active SaleAI template determines whether they are uploaded separately.
How many rows should I test before importing a large list?
Use 12 deliberately varied rows as an editorial test, not a SaleAI limit. Add more when languages, sources, owners, or field types vary widely.
What should happen to blank fields?
Define it before import: unknown, unchecked, not applicable, or withheld. Never turn every blank into a default sales value.
Should I assign an owner during import?
Assign an owner only when a next action exists and the owner list is verified. Otherwise, use a visible review queue.
How should SaleAI tags be designed?
Use a controlled vocabulary for market, segment, source family, or review state. Keep queries and dated evidence in notes.
Can I assume an incorrect SaleAI import can be rolled back?
No rollback function was verified. Use small batches, preserve source files, label the cohort, and export the test result.
What should be excluded from a CRM import?
Exclude fields with no defined purpose, guessed contact details, undocumented scores, temporary formulas, obsolete values, and personal data the team is not prepared to govern. Preserve necessary source evidence without turning every raw research field into a CRM property.
