AiCrmAiCrm

How to transfer your clients from Excel to a CRM without losing data or duplicating contacts

Team at AiCrmWritten with AI assistance24 min read

In summary

  • Excel supports 1,048,576 rows (Microsoft Learn, "Excel Specifications and Limits," consulted on October 25, 2026), so the problem is never size: it's that a sheet doesn't keep history, doesn't warn of duplicates, and doesn't know who touched what.
  • Migration almost always breaks due to four things: the CSV separator, the encoding of accents, the phone numbers that Excel converts to numbers, and the "status" column that everyone made up.
  • Before importing, normalize: one row per person, one header, phones in a unique format (+34…) and labels in a single column separated by commas.
  • Import in batches and with the error file in front: it's normal for between 2% and 10% of the rows to fail due to misspelled emails or empty names, and those rows are fixed in the CSV, not manually in the CRM.
  • Migrating is pointless if you return to the sheet in three months. The only way not to go back is for new entries to come in automatically from the web, the form, or the chat.

This article is a practical migration guide, not a brochure. We wrote it from AiCrm, the lightweight CRM with its own API from AIsToHire, because it is exactly the path taken by everyone who arrives at the app: a folder with three spreadsheets, two old exports from another tool, and the feeling that something is going to be lost along the way. You can see what the app does on the landing page of AiCrm, the lightweight CRM with API from AIsToHire.

The steps apply to any CRM that imports CSV. When something depends on how AiCrm does it specifically —limits, rules, error messages— we mention it with the number in front, so you can check it yourself.

Disordered spreadsheets passing through a funnel and turning into a single customer record with color-coded labels
Disordered spreadsheets passing through a funnel and turning into a single customer record with color-coded labels

When does Excel stop being useful as a CRM?

Excel stops being useful the day you need to know what happened with a customer, not just who they are. A sheet keeps the last state of a piece of data (name, phone, email) well, but it doesn't keep the conversation, who called, or when it’s time to write again: that ends up in a "Notes" column that grows until it's unreadable, or in someone's head.

The technical limit is almost never the reason. Excel supports 1,048,576 rows and 16,384 columns per sheet (Microsoft Learn, consulted on October 25, 2026); no SME comes close to that. What fails is the data model.

What the sheet does wellWhat starts to hurt
View and edit 500 rows at a glanceTwo people editing at the same time: the last one to save wins
Calculate totals and percentagesNo one knows if row 318 is the same person as 1,204
Filter and sort by columnThere is no history: you don't know when the status changed or who changed it
Share a file by emailThree versions of the same file circulating via email
Save anything in any cell"Status" with different values in each row: Active, active, OK, Yes
Copy and paste from anywhereNew entries from the web still come in manually, one by one

The clear signal that it has become insufficient is not the number of rows: it's when someone asks "Have we called this one yet?" and the answer is in someone else's email.


What exactly is lost in a poorly done migration?

What is lost is not the contacts: it is the part of the data that the CSV cannot represent. The six usual wrecks are accents, phone numbers, silent duplicates, the status column, empty rows, and labels stuck in a single cell. All of these are fixed before uploading the file.

Symptom at the endReal causeHow to avoid it
Jos� Mart�nez in all recordsThe CSV was saved in ANSI/Windows-1252, not in UTF-8Save as "CSV UTF-8 (comma delimited)"
The entire file in a single columnThe CSV uses ; and the importer expected , (or vice versa)Use an importer that detects the separator
34600112233 where there was +34 600 11 22 33Excel treats the cell as a number and eats the + and spacesFormat the column as Text before entering data
1,180 contacts where there were 1,050The same customer was in two sheets with the email in uppercase in oneNormalize emails to lowercase and check for duplicates
Everyone as "lead"The "Status" column had values that the CRM does not recognizeTranslate the values to those of the CRM before importing
A giant label per contactThe labels came in different columns (Tag1, Tag2, Tag3)Combine them into a single column separated by commas

The case of the phone number is the most treacherous because it doesn't give an error: the data enters, seems correct, and you only find out the day the duplicate detector does not recognize two records with the same number. Excel interprets a cell that starts with + as the beginning of a formula, so +34600112233 becomes the number 34600112233; and leading zeros disappear (0600… → 600…). Additionally, starting from 15 significant digits, Excel fills the rest with zeros (Microsoft, "Last digits are changed to zeros when you type long numbers in cells," consulted on October 25, 2026), which wrecks IBANs and long codes if you keep them in the same sheet.


Step 1: How do I prepare Excel before exporting?

Spend fifteen minutes putting the sheet in a form that an importer can understand: one row per person, a single header row in the first line, and one column per data point. Almost all import errors are decided here, not in the CRM.

The specific checklist:

  1. Delete decorative rows. Document title, logo, blank rows between blocks, totals at the end. The first row of the file must be the header: many importers —AiCrm among them— take row 1 as column names without asking.
  2. Unmerge the merged cells. A merged cell is exported as one cell with a value and another empty: part of your data will go to the wrong row.
  3. One column per data point. If you have "First and Last Name" in a single cell, that's fine: most CRMs store a single name field. What doesn't work is "Contact" with Maria (mobile 600...) inside.
  4. Normalize the phone numbers to a single format. Choose +34600112233 or 0034600112233, but don't mix: AiCrm compares phone numbers only by their digits and removes the initial +, so +34600112233 and 0034600112233 are not the same number for the duplicate detector (34600112233 versus 0034600112233). Format the column as Text before touching it.
  5. Convert emails to lowercase. Formula =LOWER(B2), copy, paste special as values. A [email protected] and a [email protected] are duplicates that no importer will catch if compared literally.
  6. Combine tags into one column. If you had Tag1, Tag2, and Tag3, concatenate them with commas: =TEXTJOIN(",";TRUE;E2:G2). AiCrm accepts ,, ;, and | as separators in the tags column, and normalizes each tag to lowercase, without double spaces and with a maximum of 40 characters.
  7. Decide on the status column. Further down you have the equivalence table; any value that the CRM does not recognize is silently converted to "lead".
  8. Remove obvious duplicates in the sheet itself. Data → Remove Duplicates, by the email column. It's free and saves you from having to merge records later.

This is how the file should look before exporting:

name,email,phone,company,country,status,tags,notes
Laura Méndez,[email protected],+34600112233,Example SL,ES,active,"client,recurring",Pays in 30 days
Carlos Ruiz,[email protected],+34611223344,Another Company,ES,lead,"web,quote",Requested quote in September
Sofía Torres,[email protected],,Online Store,ES,active,"store",

Note two details: the tags are in quotes because they contain commas, and Sofía's row has the phone empty. Both things are correct and a decent CSV parser can handle them.


Step 2: How do I export the CSV without breaking accents or separators?

In Excel, use File → Save As → "CSV UTF-8 (comma delimited) (*.csv)", available since Excel 2016. The old option "CSV (comma delimited)" saves in the local system encoding and is responsible for 90% of the Jos�.

There is a second detail that surprises many people: in a Windows system with Spanish regional settings, the list separator is the semicolon, so a "comma delimited" file actually comes out separated by ;. It's not your error or the CRM's.

SourceWhat to exportSeparator that usually comes outEncoding
Excel (Windows, Spanish)CSV UTF-8 (comma delimited);UTF-8 with BOM
Excel (macOS)CSV UTF-8,UTF-8
Google SheetsFile → Download → Comma-separated values (.csv),UTF-8
LibreOffice CalcCSV, marking "Edit filter settings"whichever you choosewhichever you choose (set to UTF-8)
Export from another CRMStandard CSV,usually UTF-8

What AiCrm does upon receiving the file, so you don't have to guess:

  • Detects the separator only: looks at the first line and keeps the most frequent candidate among comma, semicolon, and tab.
  • Removes the BOM (that invisible mark that Excel puts at the beginning of the UTF-8 file and that, if not removed, turns the first header into something unpronounceable).
  • Reads in UTF-8 and, if it detects the replacement character �, retries the entire file in latin1. It's the safety net for CSVs saved in ANSI; still, it exports in UTF-8: the safety net does not guess, it only succeeds in the common case.
  • Respects double quotes, including line breaks within a field and escaped quotes (""), which is what saves you with long notes.
  • Discards completely empty rows before counting anything.

Step 3: How do I map the columns of my sheet to the fields of the CRM?

Mapping means telling the CRM which column of your file is the name, which one is the email, and which one to ignore. In AiCrm, the mapping is proposed only from the header, and if the header says nothing, by looking at the content of the first row of data; you correct it before starting.

Mapping diagram: the columns of the CSV on the left connected with lines to the fields of the customer record on the right, with two unassigned columns
Mapping diagram: the columns of the CSV on the left connected with lines to the fields of the customer record on the right, with two unassigned columns

These are the eight fields to which a column can be mapped, and what each one accepts:

FieldHeaders it recognizes onlyWhat it acceptsIf missing
namenombre, name, contactoFree textUses the part of the email before the @
emailmail, correoFormat [email protected]Empty is accepted (but you lose duplicate detection)
phonetel, phone, movil, móvil, celularText; compared only by digitsEmpty is accepted
companyempresa, compan…, organiz…, firmaFree textEmpty
countrypais, país, countryText; stored in uppercase (uses ISO-2: ES, MX, AR)Empty
statusestado, statuslead, active, inactive, churnedlead
tagstag, etiquetaSeparated by comma, semicolon, or vertical barNo tags
notesnota, note, coment…Free text, allows line breaksEmpty

If a header does not match any of these patterns, AiCrm looks at the first row of data: if the value contains an @, it proposes it as an email; if it looks like a phone number (+ or digit followed by at least six characters of digits, spaces, periods, dashes, or parentheses), it proposes it as a phone number. What doesn't fit remains unmapped, and an unmapped column simply is not imported: this is the correct way to exclude the column "Assigned Salesperson" or "Amount 2024" that your CRM does not have.

Hard rule: you must map at least name or email. If you do not map either, the import does not even start.

The equivalence table of statuses, for you to translate in your sheet before uploading it:

What your Excel usually putsValue that needs to be written
Nuevo, Pendiente, Contacto, Prospecto, Interesadolead
Cliente, Activo, OK, Sí, En cursoactive
Dormido, Antiguo, Sin actividad, Fríoinactive
Baja, Perdido, Cancelado, No interesadochurned

Any other value —including Activo with a capital letter if your CRM expects active— falls to lead without notice. It is the most common silent error of the entire migration: the file enters "fine" and the next day you have 900 leads and no customers.


Step 4: How do I import and what do I do with the rows that fail?

Always import with the error file in front and in two batches: first a test of 20 rows, check the result in the CRM, and then the complete file. The test of 20 rows takes two minutes and detects 100% of mapping errors.

How it works internally in AiCrm, with real numbers:

ParameterValue
Maximum rows per file10,000 (if you have more, split the CSV)
Rows processed per batch100
ResumableYes: the job saves the cursor in the database
Errors saved per importUp to 500, downloadable in CSV
Automatic tag addedimportacion-AAAA-MM-DD (the date of the day)
Extra tagsThe ones you add when launching the import, to all rows
Is it necessary to keep the tab open?Yes. The processing is directed by the browser batch by batch; if you close it, the job is left halfway and resumes from the history

The rows that fail do not block the import: they are set aside, counted, and returned to you with the real line number of the file, so you can correct them in your Excel and upload only those again. The three reasons for rejection:

ReasonWhen it occursHow to fix it
Empty nameThe row has no name or email to extract it fromFill in the name in the CSV
Invalid emailDoes not comply with [email protected]Correct the email or leave it empty
Duplicate (email already exists)That email is already in your CRM and you did not check "update existing"Decide: ignore the row or reimport with the option activated

That third option, "update existing ones", is what makes a second import secure. With the box checked, when the email already exists, AiCrm does not create a new record: it updates the existing one, and does so conservatively —only fills in the fields that were empty and adds the new tags to the ones it already had, without deleting any—. With the box unchecked, the row is rejected and you find it in the error CSV.

An important detail before starting: there is no "undo import" button. Deleting the import from the history removes the record of the job, not the contacts it created. What you do have is the tag importacion-AAAA-MM-DD on all records of that batch, which allows you to filter and review them; but deletion in AiCrm is contact by contact, there is no bulk deletion. That is why the test of 20 rows is not optional.


What do I do with the duplicates that the sheet already had?

Duplicates are not resolved during import: they are resolved later, one by one, deciding which data from each record you keep. What the CRM can do is find them and let you merge them without losing the history of either.

Two almost identical customer records that overlap and merge into one, with the labels of both combined into a single row
Two almost identical customer records that overlap and merge into one, with the labels of both combined into a single row

AiCrm detects them with three rules, in this order:

  1. Exact email (compared in lowercase). This is the strong rule: if it matches, there is definitely a duplicate.
  2. Exact phone number, only if the record does not have an email. It compares only the digits, ignoring spaces, dashes, dots, and parentheses, and removing the initial +.
  3. Similar name, with a similarity greater than or equal to 0.8. The formula is 1 − (Levenshtein distance ÷ length of the longest name), all in lowercase.

That third rule deserves two concrete examples, because it explains both what it catches and what it does not:

  • jose luis ramirez versus josé luís ramírez: three substitutions (three accents) over 17 characters → 1 − 3/17 = 0.82. Exceeds 0.8: detected.
  • marta gomez versus marta gomez ruiz: five characters need to be inserted over 16 → 1 − 5/16 = 0.69. Does not reach 0.8: not detected.

Practical conclusion: automatic detection catches typos and accents, but does not catch the same person registered once with the first name and another with the first name and both last names. For those cases, the search by name, email, phone, or company is your tool.

Detection also runs while you are typing a new contact: half a second after typing the email or name, the form alerts if there is already someone similar, with a link to view the existing record and the option to create the contact as well.

What exactly happens when merging two records

Merging in AiCrm is field by field: you choose, for each piece of data, whether to keep the one from the primary record or the secondary one.

DataHow it is resolved
Name, email, phone, company, country, state, notesYou choose, field by field
TagsUnion of both: none is lost
Registration dateThe oldest of the two is kept
Last activityThe most recent is kept
Activity timelineAll from the secondary record are reassigned to the primary
RemindersSame: reassigned to the primary record
Secondary recordIt is deleted, and a merge entry with the name and email of the absorbed record is logged in the timeline

That is to say: after merging you can still reconstruct what happened, because the merge itself is noted in the history.


How do I avoid going back to the spreadsheet in six months?

Almost everyone who migrates goes back to the spreadsheet for the same reason: new entries keep coming in manually. As long as the web form arrives by email and someone has to copy and paste, the parallel sheet reappears. The solution is not discipline; it is removing the manual step.

Integration scheme: web form, chat, and store send contacts to the CRM database, which in turn alerts two external servers via webhooks
Integration scheme: web form, chat, and store send contacts to the CRM database, which in turn alerts two external servers via webhooks

AiCrm provides a public REST API for that. You create a key from the app, send it in the X-Api-Key header, and your website registers the contact at the moment:

curl -X POST https://aistohire.com/api/apps/public/crm/customers \
  -H "X-Api-Key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"email":"[email protected]","name":"Laura Méndez","phone":"+34600112233","company":"Example SL","tags":["web","quote"],"subId":"contact-form"}'

What you need to know about that endpoint:

BehaviorDetail
New entryResponds 201 with action: "created" and the contact id
Email (or phone) already existsResponds 200 with action: "existing": does not duplicate; fills in empty fields, adds new tags, and notes a new submission in the timeline
Minimum requiredEmail or phone. If you do not send a name, the part of the email before the @ is used
Automatic tagapi-lead, in addition to the ones you send
Source dataSaves the IP and the country of the request
subIdFree text to know which form or campaign it came from; appears in the contact's timeline
Usage limit100 requests per minute and key; exceeding it returns 429

And conversely: when something changes in the CRM, the webhooks notify your systems. Available events: customer.created, customer.updated, customer.deleted, and activity.created. Each delivery carries the header X-AiCrm-Event with the event name and X-AiCrm-Signature with an HMAC-SHA256 of the body, calculated with the webhook secret, so you can verify that the call is ours and not from a third party.

Honesty about webhooks, because it matters for designing your integration: the delivery is best effort, with one immediate retry at 1.5 seconds and a timeout of 5 seconds. There is no persistent retry queue: if your server is down for two minutes, that event is lost. If you need delivery guarantees, use the webhook as a notification and the API GET /crm/customers as the source of truth for reconciliation.

The rest of the API, in case your integration needs more than just registration:

MethodPathPurpose
POST/api/apps/public/crm/customersCreate (or retrieve the existing one)
GET/api/apps/public/crm/customersList with page, perPage, search, status, tags, country
GET/api/apps/public/crm/customers/:idRecord with its activity
PUT/api/apps/public/crm/customers/:idUpdate
DELETE/api/apps/public/crm/customers/:idDelete
POST/api/apps/public/crm/customers/:id/activitiesAdd an activity to the timeline

What AiCrm does NOT do? The limits, in writing

An article on migration that does not tell you the limits of the tool is wasting your time twice: during migration and when migrating back. These are AiCrm's limits, as is.

LimitValue / behavior
Contacts traversed when filtering or exporting5,000 per creation. Above that number, this is not the tool
Rows per import10,000 per file
Contacts in the PDF list500
Push notifications for remindersNone. Reminders are seen in "Today's Pending" and in the Reminders section, with overdue, complete, and postpone marks
Sending emails or messages to clientsNo. AiCrm organizes and records; it does not write to anyone on your behalf
Mass deletion of contactsNone. It is deleted one by one
Import with the tab closedNo. It can be resumed from the history, but it does not advance on its own
Webhook retriesOne immediate, in memory. No persistent queue
Saved views and @ mentions in notesPending: there is no interface yet
Artificial intelligenceThe app does not use any AI engine. Statistics, trends, duplicate detection, and reports are deterministic calculations on the server

That last row is deliberate: there is no magic scoring of leads or generated summaries. Duplicate detection is a formula that you can calculate manually, and that is why we have shown it to you above.

What it does do, summarized: records with tags, statuses, and timelines; search by name, email, phone, or company; filters by status, tags, country, and dates; reminders with date and time; detection and merging of duplicates; CSV import and export; reports in PDF or CSV; API keys, webhooks, and a dashboard with totals by status, new entries from the last 30 days with their variation, countries and browsers of origin, and trend graphs for 7, 30, 90 days, and one year.


And what about data protection when moving a customer database?

Moving contacts from a spreadsheet to a CRM does not change your obligations, but it does change how easy it is to fulfill them. Regulation (EU) 2016/679 (GDPR) requires in its Article 5.1(d) that personal data be "accurate and, where necessary, kept up to date," and that inaccurate data be erased or rectified without delay: three different records of the same person with different phone numbers are, literally, inaccurate data.

A migration is also the natural moment for the cleanup you have been postponing: contacts that no longer have a basis for being there, do not migrate them.

What AiCrm brings to the table in that regard, without exaggeration: each contact is a unique and deletable record, the history notes when it was created and by what means, and you can export the data of one person or the entire database to CSV whenever you want, with the fields you choose —useful for responding to a request for access or portability (arts. 15 and 20 of the GDPR)—.

Where the limit is: AiCrm is a contact database, not a compliance tool. It does not manage consents, does not keep the activity log of processing under Article 30, nor does it evaluate your legitimizing basis. This article is informational and is not legal advice: for your specific case, consult a professional or your data protection officer.

Migration checklist, from start to finish

Print it or copy it. For a database of 1,000-2,000 contacts, the entire process can fit in a morning.

  1. Gather all sources: spreadsheets, old exports, your mobile's contact book, emails from the form.
  2. Paste everything into a single sheet with the same header and add a "source" column (this will serve as a label).
  3. Clean: remove decorative rows, unmerge combined cells, emails in lowercase, phones in a single format.
  4. Translate the status column to lead / active / inactive / churned.
  5. Merge the labels into a single column separated by commas.
  6. Data → Remove duplicates by email.
  7. Save as CSV UTF-8.
  8. Cut a test file with 20 rows and upload it.
  9. Review the proposed mapping, correct what is necessary and launch the test.
  10. Check in the CRM: accents, phones, statuses, and labels of those 20 records.
  11. Upload the complete file (maximum 10,000 rows per batch) and keep the tab open until the end.
  12. Download the CSV of errors, correct those rows in your Excel and re-upload only them, this time with "update existing ones" checked.
  13. Check for duplicates in records that seem repeated and merge them.
  14. Schedule reminders for the five contacts you need to call this week. If you don't do this, the CRM becomes a dead file.
  15. Connect your website form to the API so that future entries do not go through any spreadsheet.

Frequently Asked Questions

How many contacts can I import at once?

Up to 10,000 rows per file in AiCrm, processed in batches of 100. If your database is larger, split the CSV into several files and upload them one after the other. Keep in mind the other limit, which is the one that really matters: filters and exports go through a maximum of 5,000 contacts per creation, so if you're handling tens of thousands of records, you need a tool from another category.

Why does my CSV have broken accents?

Because it was saved in the local system encoding (ANSI / Windows-1252) instead of UTF-8. Go back to Excel and use Save As → "CSV UTF-8 (comma delimited)", available from Excel 2016. AiCrm retries the file in latin1 if it detects the character �, but this is a safety net, not a guarantee: export correctly from the start and save yourself the scare.

Will my contacts duplicate if I import the same file twice?

No, if you use the option correctly. With "update existing ones" checked, rows whose email is already in the CRM update the existing record instead of creating another: they fill in the fields that were empty and add new tags without deleting the old ones. If you don't check that option, those rows are rejected and appear in the error CSV as "Duplicate (email already exists)". Note: the protection is by email; a row without an email can still create a new record.

Can I undo an import if I made a mistake?

There is no undo button. Deleting the import from history removes the record of the work, not the contacts it created, and in AiCrm, deleting contacts is done one by one, without bulk deletion. What you do have is the automatic tag importacion-AAAA-MM-DD on all records from that batch, which allows you to filter and review them quickly. That's why it's advisable to always do a test with 20 rows before the full file.

What happens with the "Status" column in my Excel?

Only four values are recognized: lead, active, inactive, and churned. Anything else —"Client", "Active" with a capital letter, "OK", "Pending"— will be converted to lead without any warning. Translate that column in your sheet before exporting, or you'll end up with your entire database marked as lead.

Can I import from Google Sheets or from the export of another CRM?

Yes: any CSV works. In Google Sheets, File → Download → Comma-separated values (.csv), which already comes out in UTF-8 and comma-separated. If you're coming from another CRM, export to CSV and map its columns to the eight fields of AiCrm; columns without an equivalent can be left unmapped and simply will not be imported. What is not migrated in any case via CSV is the conversation history from the other system: the timeline starts from scratch, with the import registration as the first entry.

Do I need to know how to code to use AiCrm?

No. All management —creating records, tagging, searching, filtering, importing, exporting, merging duplicates, and scheduling reminders— is done from the app without touching a line of code. The API and webhooks are optional and are there for when you want your website, your form, or your chatbot to register contacts automatically.

What if I want to switch to another tool tomorrow?

You export. AiCrm generates a CSV with the fields you choose from thirteen available (id, name, email, phone, company, country, status, tags, registration date, last activity, notes, IP, and browser), over the entire database or only over what you have filtered at that moment. You can also generate summary reports, contact reports, or activity reports in PDF or CSV. Your data comes out in the same format it went in.


In a nutshell

AiCrm is the lightweight CRM from AIsToHire that gathers your contacts into records with tags, statuses, notes, reminders, and timeline, imports and exports them in CSV, detects and merges duplicates, and allows your website to register clients automatically through its REST API with webhooks; it is at AiCrm — Lightweight CRM with API to create customers from any system It does not use artificial intelligence, does not write to your clients for you, and is designed for databases of up to a few thousand contacts: if that describes your case, the migration you just read can fit into a morning.

Try AiCrm for free and import your CSV

Try AiCrm for free and import your CSV

↑ Back to top