
Why This Migration Is Harder Than the Others
Moving from one CRM to another is a known problem. The old system has objects, fields and relationships; the new one has objects, fields and relationships; you map one to the other. Difficult in places, but structured on both sides.
Moving from spreadsheets is a different and harder problem, and it is routinely scoped as though it were easier.
A spreadsheet has no enforced structure. It has whatever its owners chose to put in it over the years: a company name in one column, a contact name in the next, a phone number that is sometimes a mobile and sometimes a switchboard, a βnotesβ column holding the entire history of the relationship in one cell, colour coding that meant something to someone in 2021, and rows that are customers, prospects, suppliers and former employees with nothing to distinguish them.
The data is genuinely valuable. It is also built for a person to read, not for a system to understand. That gap is the whole migration, and it is invisible until somebody tries to import the file.
This article covers the structural decision that has to be made before anything is imported, the eight problems present in almost every spreadsheet, how to decide what not to bring across, how to run the migration, and the ninety days after go-live that decide whether the CRM gets used. It is written for sales directors, founders and operations managers adopting their first CRM.
The Business Problem: Data Built for People, Not Systems
- One row means several things. A single line holds a company, a person, an opportunity and a history. A CRM separates those, and the separation is not automatic.
- There are multiple files. Each salesperson has their own, maintained their own way, with overlapping customers recorded differently.
- Ownership is implicit. Who looks after an account is known because it is in their file. Move everything into one place and that knowledge disappears unless it is captured deliberately.
- History is in free text. The notes column carries years of relationship context in prose. Valuable, unstructured and almost impossible to use once imported as a block.
- Formatting carries meaning. Bold for important clients, red for problem accounts, a colour for a region. None of it survives an import.
- Nobody owns the data quality. The spreadsheet was always good enough for the person using it. Nobody was responsible for it being good enough for anyone else.
- The current process is invisible. How deals actually progress, what stages exist, what triggers a follow-up: none of it is written down, because the spreadsheet never required it to be.
What a Bad Migration Costs
- A CRM nobody trusts from day one. The most damaging outcome. If the first thing a salesperson sees is a duplicate of their best client, or a contact attached to the wrong company, they go back to their spreadsheet, and a CRM that loses its users in the first month rarely gets them back.
- Duplicates that compound. Import three overlapping files without deduplication and every subsequent report is wrong. Removing duplicates after go-live is far harder than preventing them, because by then activity has been logged against both copies.
- Lost relationship context. Notes imported as a single text block are technically present and practically gone.
- Ownership disputes. Accounts imported without owners, or with the wrong owners, produce arguments that land on the sales director within a week.
- Automation that misfires. Workflows and email sequences triggered by imported data act on whatever was imported, including the records that should never have come across.
- A second migration. The most expensive outcome of all: a business that imports badly, abandons the CRM within a year, and then pays to do it again properly.
Structure Before Import: The Three-Object Problem
This is the decision that determines whether the migration works, and it has to be made before anything is imported.
A spreadsheet row usually conflates three distinct things that a CRM keeps separate:
- The organisation: the company you sell to. In Zoho CRM terms, broadly the account.
- The person: the individual you deal with at that organisation. Broadly the contact.
- The opportunity: a specific piece of business you are pursuing. Broadly the deal.
One organisation has several people and many opportunities over time. A spreadsheet that holds one row per relationship cannot express that, so every row has to be split.
The practical consequence: you need to decide, for each column in each spreadsheet, which of the three it belongs to. Company name and address belong to the organisation. Name, phone and email belong to the person. Value, stage and expected close belong to the opportunity.
The harder consequence: where the same organisation appears in several rows (because you have several contacts there, or several opportunities), those rows must collapse into one organisation with several people and several opportunities attached. That is the deduplication problem, and it is fundamentally a structural decision rather than a data-cleaning task.
There is also a fourth question: leads. Is a record somebody you have qualified, or somebody you have merely heard of? A first CRM is a good moment to decide whether unqualified records come in as leads or not at all, and that decision is easier made now than unpicked later.
Confirm module and field behaviour for your edition before finalising the mapping, since what is available varies, and design against what you actually have.
The Eight Things Wrong With Every Spreadsheet
Almost every spreadsheet migration encounters all eight. Planning for them is most of the work.
- Duplicates, inexact. βAcme Ltdβ, βACME Limitedβ, βAcme Ltd.β and βAcme (Pune)β are the same company. Exact-match deduplication catches none of them. You need normalised matching (case, punctuation, legal suffixes, location qualifiers) and a human review of the borderline cases.
- Merged cells. Visually tidy, structurally disastrous. A merged cell spanning five rows imports into one row and leaves four blank. Unmerge and fill down before anything else.
- Several values in one cell. Two phone numbers separated by a slash, three email addresses in one field, a contact name and title together. Each needs splitting into the right fields.
- Inconsistent formats. Dates written five different ways, phone numbers with and without country codes, currency with and without symbols. Standardise before import, because a system will either reject them or store them as text.
- Free-text where a list belongs. A βstatusβ column with forty slightly different values that mean six things. Map each variant to a controlled value before import, or the pipeline reports will be meaningless.
- Meaning carried by formatting. Colours, bold and font size encode categories. Convert each into an actual field value, or the information is lost on import.
- Notes as history. Years of context in one cell. Decide whether to import it as a single historical note, split it into dated entries where the structure allows, or summarise it, but decide deliberately rather than discovering the result.
- Records that should not exist. Former employees, suppliers mixed in with customers, test rows, people who asked not to be contacted, and businesses that closed years ago. Every one of these imported becomes noise and some become a compliance problem.
Deciding What Not to Migrate
The most valuable decision in the project, and the one teams find hardest, because it feels like losing information.
- Set an inactivity threshold. A record with no contact for several years is rarely worth migrating as a live record. Archive it: export it, keep it, but do not load it into the system people work from.
- Exclude anyone who has asked not to be contacted, and handle consent deliberately. Importing records without a lawful basis to contact them, and then pointing marketing automation at them, is a compliance exposure that is entirely avoidable at this stage. Data protection obligations are jurisdiction-specific, so confirm your position with whoever advises you on them before deciding what comes across.
- Separate suppliers, partners and staff from customers. They may belong in the CRM, but not mixed indistinguishably with prospects.
- Do not migrate fields nobody uses. A column filled in for six months three years ago and abandoned is not worth a CRM field. Every field you create is one more thing to maintain and one more thing users ignore.
- Archive rather than delete. Everything excluded should still exist somewhere readable. The aim is a clean working system, not a lost history.
The reasoning to give people: a smaller, accurate CRM gets used. A complete, inaccurate one gets abandoned. Every record you leave out is a record nobody has to wade past.
Running the Migration
- Consolidate the files first. Every spreadsheet into one working file, with a column recording which file each row came from. That source column is what lets you trace a problem back later.
- Clean in the working file, not in the CRM. Every correction is cheaper before import than after. Unmerge, split, standardise, map, deduplicate: all of it in the spreadsheet.
- Assign an owner to every record before import. Not after. Unowned records produce disputes in the first week, and resolving ownership in a spreadsheet is far easier than in a live system.
- Load in order: organisations, then people, then opportunities. Each depends on the one before. Loading people before their organisations exist leaves orphaned contacts.
- Test with a small batch first. Fifty records from across the data, including the awkward ones. Check them in the CRM properly, record by record, before loading the rest. This is where you discover the field mapping was wrong, at a cost of fifty records rather than five thousand.
- Do not switch on automation until the data is verified. Workflows and sequences act on what is there. Turn them on once you trust what is there.
- Keep the original files untouched. Read-only, retained. If something goes wrong you need to know what the source said.
- Freeze the spreadsheets on go-live day. Not deleted, but made read-only. If people can still edit them, they will, and you will have two sources of truth within a week.
The First Ninety Days After Go-Live
A clean migration is necessary and not sufficient. Most first CRMs that fail, fail here.
- Week one: fix what users find, fast. Users will find errors. The response time to those reports determines whether they keep using the system. A visible, fast correction process builds trust; a slow one ends adoption.
- Define what βusing itβ means. Every call logged? Every opportunity updated weekly? Be specific, because βuse the CRMβ is not a standard anyone can meet or fail.
- Make the sales director use it in meetings. If the pipeline review happens from the CRM, the CRM gets updated. If it happens from somebodyβs spreadsheet, the CRM is optional. This single habit does more for adoption than any training.
- Resist adding fields and automation early. The temptation after go-live is to configure everything people ask for. Let usage stabilise first; most early requests turn out to be unnecessary once people are working in the system.
- Measure adoption, not just data. Logins, activities logged, opportunities updated. A declining trend in the second month is the early warning, and it is recoverable if caught.
- Review data quality at ninety days. New duplicates, blank required fields, stale opportunities. The migration cleaned the past; the ninety-day review tells you whether the present is staying clean.
Approaches Compared
| Consideration | Import the files as they are | Clean lightly and import | Restructure, cleanse and migrate in stages |
|---|---|---|---|
| Effort before go-live | Minimal | Moderate | Substantial |
| Duplicates | Imported, compounding | Exact matches removed only | Normalised matching with human review |
| Organisation, person and opportunity separated | No | Partially | Yes |
| Ownership assigned | Rarely | Sometimes | Every record, before import |
| Free-text statuses mapped to controlled values | No | Partially | Yes |
| Records excluded deliberately | No | Some | Inactive, opted-out and non-customer records handled |
| User trust on day one | Low | Moderate | High |
| Likely outcome | Abandoned within a year, then a second migration | Usable with persistent quality problems | Adopted |
Best Practices
- Decide the organisation-person-opportunity structure first, mapping every column of every file to one of the three.
- Consolidate all files into one working file with a source column before cleaning anything.
- Clean in the spreadsheet, not in the CRM. Every fix is cheaper before import.
- Deduplicate with normalised matching, and review the borderline cases by hand.
- Map free-text statuses to controlled values before import.
- Convert formatting into fields so colours and bold do not take their meaning with them.
- Exclude inactive, opted-out and non-customer records deliberately, archiving rather than deleting.
- Assign an owner to every record before import.
- Test with fifty awkward records before loading the rest, and load organisations before people before opportunities.
- Freeze the spreadsheets on go-live day, and run the pipeline meeting from the CRM from week one.
Common Mistakes
- Importing the files as they are. The fastest route to a CRM nobody trusts.
- Exact-match deduplication only. Catches almost none of the real duplicates.
- Loading one row as one record. The three-object problem, unresolved, produces a CRM that cannot represent a company with several contacts.
- Leaving ownership until after go-live. Guarantees disputes in the first week.
- Migrating everything because deleting feels risky. Archive the old records; do not load them.
- Importing records with no lawful basis for contact, then pointing automation at them.
- Switching on automation before the data is verified.
- Skipping the test batch. The mapping error found at fifty records costs nothing; found at five thousand it costs the project.
- Leaving the spreadsheets editable. Two sources of truth within a week.
- Running the pipeline meeting from a spreadsheet after go-live. Signals that the CRM is optional, and everyone reads the signal correctly.
An Illustrative Scenario: A 25-Person Sales Team
The following is a composite illustration drawn from patterns common to first CRM implementations. It is not an account of a specific Techvaria client, and no figures below come from any real engagement.
Consider a business with a twenty-five-person sales team selling to other businesses, adopting a CRM for the first time after years of individual spreadsheets.
The Starting Position
Eleven spreadsheets: one per senior salesperson, plus a shared βmasterβ file nobody fully trusted. Several thousand rows in total. The sales director believed the business had a few thousand customers and prospects.
What Consolidation Found
Once the files were merged into one working file with a source column, normalised matching found that a large share of rows were duplicates of each other: the same organisations recorded by different salespeople, under slightly different names. The number of distinct organisations was considerably smaller than the row count suggested, which surprised the sales director and explained a long-running argument about who owned which account.
What the Structure Exercise Found
Every file used one row per relationship. Companies with three contacts appeared three times. Companies with two live opportunities appeared twice. Resolving that into organisations, people and opportunities was the largest single task, and it was a structural decision rather than a data-cleaning one.
The status column contained around thirty distinct values meaning roughly six things. Two salespeople used colour coding for priority; one used bold for accounts at risk.
What Was Excluded
Records with no contact in several years were archived rather than migrated, after a conversation in which the team was initially reluctant and then relieved. A set of supplier contacts mixed into one file were separated. Records marked as not wanting contact were excluded, after confirming the position with the businessβs own adviser on data protection.
What Was Implemented
Ownership was assigned to every organisation in the working file before import, which surfaced and settled most of the ownership disputes in a single meeting, rather than one at a time after go-live.
Fifty awkward records went in first. The test found that one phone field was mapped to the wrong place and that some notes were truncating. Both were fixed before the main load.
Organisations, then people, then opportunities. Automation stayed off for the first fortnight. The spreadsheets were made read-only on go-live day.
What Made It Work
The sales director ran the next pipeline meeting from the CRM, and every one after. Two salespeople who had not updated their opportunities found that their deals were not in the review, and updated them before the following week.
After Ninety Days
The data quality review found a modest number of new duplicates, almost all from one web form feeding leads in without a duplicate check, which was fixed at source. Adoption held. The sales directorβs own summary was that the migration had been the visible work and the pipeline meeting had been the thing that decided whether it mattered.
Industry Use Cases
- Trading and distribution. Large customer lists maintained by branch, where cross-branch duplication is the dominant problem. See trading and distribution.
- Manufacturing. Distributor and end-customer records mixed together, where separating the two relationships is a structural decision rather than a cleaning task. See manufacturing solutions.
- Professional services. Relationship-led selling where the notes column carries most of the value, making the history decision unusually important. See professional services.
- IT services and software. Several lead sources feeding spreadsheets in different formats, where standardisation before import is the main effort. See IT services and ERP.
- Healthcare. Referrer and provider records where consent and data protection obligations make the exclusion decisions a compliance matter. See healthcare solutions.
Frequently Asked Questions
Technically yes, and it is the most common reason first CRMs fail. A spreadsheet row usually holds an organisation, a person and an opportunity at once, along with duplicates, merged cells and free-text statuses. Importing it as it is produces a CRM full of duplicates and broken relationships that users stop trusting within weeks.
A spreadsheet typically has one row per relationship, while a CRM separates the organisation, the person and the opportunity. One organisation can have several people and many opportunities, so rows have to be split and duplicates collapsed. It is a structural decision to make before import, not a cleaning task to do afterwards.
Normalise before matching (case, punctuation, legal suffixes such as Ltd and Limited, and location qualifiers), then review the borderline cases by hand. Exact-match deduplication catches very few real duplicates, because the same company is almost never recorded identically across files.
No. Archive records with no contact in several years, separate suppliers and staff from customers, exclude anyone who has asked not to be contacted, and drop fields nobody uses. Archive rather than delete, so the history still exists. A smaller, accurate CRM gets used; a complete, inaccurate one gets abandoned.
Organisations first, then people, then opportunities, because each depends on the one before. Loading people before their organisations exist leaves contacts orphaned. And load a test batch of around fifty awkward records first, checking each one properly before loading the rest.
Handle it deliberately at this stage, because importing records without a lawful basis to contact them and then pointing automation at them is an avoidable exposure. Obligations are jurisdiction-specific, so confirm your position with whoever advises you on data protection before deciding what comes across.
The sales director running the pipeline meeting from it, from the first week. If the review happens from the CRM, opportunities get updated; if it happens from a spreadsheet, the CRM is optional and everyone knows it. Alongside that: fast fixes for errors users report, a specific definition of what using it means, and freezing the old spreadsheets on go-live day.
Conclusion
A spreadsheet migration gets scoped as the easy case because there is no old system to map from. It is the hard case for exactly that reason: the data was built for a person to read, and nothing in it enforces structure.
The work is mostly decided before anything is imported. Split every row into the organisation, the person and the opportunity it actually describes, and collapse the duplicates into single organisations with their contacts and deals attached. Deal with the eight problems present in nearly every file: inexact duplicates, merged cells, several values per cell, inconsistent formats, free-text where a list belongs, meaning carried by formatting, history buried in notes, and records that should not be there at all.
Then decide what not to bring across. Archive the inactive records rather than loading them, separate the non-customers, exclude anyone without a lawful basis for contact, and drop the fields nobody uses. Assign an owner to every record before import. Test fifty awkward records first. Load in dependency order, keep automation off until the data is verified, and freeze the spreadsheets on go-live day.
And then run the pipeline meeting from the CRM. Every migration in this article is preparation for that single habit, because a clean CRM that nobody reviews from is still a CRM nobody uses.
Get Your First CRM Right
Techvaria is a Zoho Premium Partner with offices in Bangalore, Ahmedabad and Dubai.
A CRM migration assessment works from your actual spreadsheets and produces a count of distinct organisations once duplicates are resolved (usually smaller than you expect), a column-by-column mapping into organisations, people and opportunities, a list of what should be archived rather than migrated, and an ownership map settled before go-live rather than argued about after it.
If you are about to adopt your first CRM, this is the cheapest moment to get it right, because every problem costs less to fix in a spreadsheet than in a live system.
Talk to our Zoho consultants, or see how we approach Zoho CRM implementations.
Get Your First CRM Right

Director @ Techvaria | Solutions Architect | Low-Code & AI Automation for Growth | Proven Expertise in Digital Transformation Across Industries