A CRM implementation checklist is an ordered list of decisions and pass/fail tests that takes a business from "we bought a CRM" to "the team works from it and the spreadsheet is retired". The order is the point: pipeline stages come before fields, fields before import, import before automation. The buying side is covered in our guide to CRM software in South Africa; this post covers what happens after you sign.
Most rollouts go wrong in the same place: someone imports a messy list into a half-designed system, and the first month is spent explaining to the team why the data cannot be trusted. The checklist below puts a gate in front of each step so that does not happen, and marks where South African law and platform behaviour change what you should do.
Quick Answer
A working CRM implementation checklist has seven gates: name one owner and one problem, define pipeline stages, clean the data, choose the duplicate key, connect lead sources, set permissions and retention, then pilot before cutting over. Each gate has an exit test you can answer yes or no. Do not move to the next gate until the current one passes, because errors in stages and duplicates get more expensive with every record loaded.
Jump to
Not sure which gate your rollout is stuck at?
Send us a description of where leads enter and where they go today, and we will point to the first gate that is failing.
Request a flow reviewCRM Implementation Checklist: The Seven Gates in Order
A gate is a point in the rollout where you stop and test one condition before continuing, and a CRM implementation plan built from gates is easier to run than one built from dates. We deliberately print no standard timeline: how long each gate takes depends on how much data you hold, how many people touch a lead and how messy the current records are.
| Gate | What you decide | Exit test | Best for catching |
|---|---|---|---|
| 1. Owner and problem | One named owner; one problem the CRM must fix | Can the owner state the problem in one sentence? | Tool bought with no job to do |
| 2. Stages and fields | Pipeline stages, required fields, naming rules | Can two staff place the same lead in the same stage? | Stages that mean different things to different people |
| 3. Clean the data | Which records are active, accurate and lawful to hold | Does every removed or archived record have a recorded reason? | Importing years of dead contacts |
| 4. Duplicate key | Which field matches an incoming row to an existing record | Does a test re-import update records instead of doubling them? | Silent duplicate creation |
| 5. Lead sources | How forms, ads, calls and WhatsApp enquiries land in the CRM | Does a test enquiry from each source arrive with its source tagged? | Leads that bypass the system |
| 6. Access and retention | Who sees what; how long records are kept | Can you list who can export the full database? | Over-broad access, indefinite retention |
| 7. Pilot and cut over | Pilot group, success measure, date the spreadsheet goes read-only | Did the pilot group finish a full lead cycle in the CRM only? | Parallel systems that never converge |
Gate 1: Name One Owner and One Problem Before Configuring Anything
A named owner and a one-sentence problem statement give every later configuration choice a tiebreaker, because "does this help fix the problem?" settles most arguments about fields and stages. Without them, the CRM collects whatever each department asks for and nobody is accountable for whether it works.
The problem should be a business problem, not a feature: "enquiries from our website are followed up by whoever notices them" is a problem; "we need pipeline reporting" is a feature wish. The owner does not have to be technical. They need the authority to say no to a field request and to tell the team that the CRM is now the only place a lead exists.
Budget belongs here too. Licence fees are the visible part; migration, integration and training time are the rest, and we break them down in our post on CRM implementation cost in South Africa.
Gate 2: Define Pipeline Stages and Fields on Paper First
A pipeline stage is a checkpoint a lead passes through, and it only works as a measurement if each stage has an entry condition a second person could verify. "Qualified" is a stage nobody can audit; "Budget confirmed and decision-maker named in the notes" is one they can.
Draft the stages and the field list in a document, not in the software. For each field, write who fills it, when, and whether it is required to move a lead forward. As a working judgement rather than a measured rule, keep required fields few at early stages so staff are not tempted to skip logging a lead, and put the stricter requirements at the stage where a deal is won or lost.
Agree a naming convention at this point, before any import. Inconsistent labels such as "Source: Google", "google ads" and "Adwords" split reporting three ways, and our post on CRM field naming conventions gives a workable scheme.
Takeaway: The Two-Person Stage Test
Give two staff members the same five real leads and ask each to assign a pipeline stage without talking to each other. If their answers differ, the stage definitions are not ready, and importing data into them will bake the disagreement into every report.
Gate 3: Clean the Data Before It Is Loaded, Not After
Cleaning before import means deciding which contacts deserve a place in the new system and fixing or dropping the rest, so the CRM starts small and trusted instead of large and doubted. This is the stage a CRM data migration checklist spends most time on, and the one most often skipped.
Work through the source files in this order:
- Merge obvious duplicates in the source file, matching on the field you provisionally expect to use as the key, and confirm that choice at Gate 4.
- Remove records with no usable contact route: no valid email, no mobile number.
- Standardise phone numbers to one format and put names in consistent case.
- Tag each record with where it came from, even if the answer is "legacy spreadsheet".
- Decide a cut-off for inactive contacts and archive, rather than import, the ones beyond it.
The legal reason to be strict is in POPIA.
Section 16(1) requires a responsible party to take "reasonably practicable steps to ensure that the personal information is complete, accurate, not misleading and updated where necessary" (section 16 of POPIA). Section 14 adds that records must not be kept longer than necessary for the purpose they were collected for, unless retention is required or authorised, and must be destroyed, deleted or de-identified as soon as reasonably practicable once the authority to keep them ends (section 14 of POPIA).
A database migration is therefore a natural moment to apply a retention rule, since importing a decade of dead contacts moves the problem into a new system. The mechanics of moving the cleaned file are in our CRM data migration guide. Whether a particular record is still needed for a lawful purpose is a question for your own attorney.
Gate 4: Choose the Duplicate Key Before the First Import
The duplicate key is the field a CRM uses to decide whether an incoming row is a new record or an update to an existing one, and picking it wrongly is the quietest way to corrupt a migration. Platforms differ on what they do by default, so check the behaviour of your tool rather than assuming.
| Platform | How it matches on import | Identifier options | What to do |
|---|---|---|---|
| HubSpot | Record ID, if included, supersedes any other unique identifier in the import | Record ID; a contact's email address; custom properties set to require unique values | Export with Record ID before re-importing edits |
| Zoho CRM | By default, a file record whose unique field (such as email or phone) matches an existing record is skipped | Unique field, or the Lead/Contact ID when updating (check your edition's features) | Choose update on purpose, and match on ID for corrections |
These come from the vendors' own documentation: HubSpot's import guide and Zoho's duplicate-handling article. HubSpot's guide also notes that in a single-file import covering several objects, identical data across rows is treated as one record when no unique identifier is provided for that object. Zoho's article says you can choose to skip or update existing records for duplicates.
The practical risk, as our own judgement rather than a vendor finding: B2B lists often hold several people behind one shared mailbox such as info@ or accounts@, and some people have two mobile numbers. Matching on email alone can then merge distinct people into one record or skip rows you wanted loaded. Decide what a "person" is in your business, whether that is an individual, a company or both, and pick the key to match.
Takeaway: Test the Key With a Small File
Before the full import, load a small test file twice.
Edit a few rows in the file between the two loads. A tool set to create new records will double them; a tool set to skip duplicates, such as Zoho CRM by default, will quietly leave your edits out. Either result means the duplicate key or the update setting is wrong. Fix it while the database is small enough to delete and start again. Cleaning up after a full load is a separate job, covered in our duplicate contact cleanup guide.
Worried a messy list will break your import?
Tell us what columns your current lead list holds and we will say which one should be your duplicate key and what to clean first.
Get a data fit assessmentGate 5: Connect Every Lead Source and Test Each With a Dummy Enquiry
Connecting lead sources means every route by which a prospect can reach you, whether a web form, a Meta lead ad, a Google Ads call, a WhatsApp message or a walk-in, ends as a tagged record in the CRM. A source that is not connected keeps living in an inbox, and the pipeline report will under-count it.
List the sources, then submit a dummy enquiry through each and check three things: the record arrived, the source field is filled with a value from your naming scheme, and a follow-up task or owner was assigned.
Meta lead forms are submitted on Meta rather than on your website, so they reach the CRM by a different route from a web form and need testing separately; our post on Meta lead ads CRM integration covers it, and the WhatsApp equivalent is in WhatsApp CRM integration.
For paid traffic, capture the campaign tags (UTM parameters) in hidden form fields so that the CRM, not a spreadsheet, answers which campaign produced paying customers. GA4 and Google Ads reports show clicks and form submissions; the CRM is where you see which of those became revenue. Agree on that division before launch so that nobody argues about whose number is right afterwards.
Gate 6: Set Permissions and Retention Before Staff Log In
Permissions decide who can view, edit and export records, and setting them before launch avoids retrofitting them after a salesperson has already downloaded the whole database. A CRM holds names, contact details and conversation history, which are personal information under POPIA.
Section 19(1) requires a responsible party to secure the integrity and confidentiality of personal information in its possession by taking appropriate, reasonable technical and organisational measures.
Section 19(2) lists what that involves: identifying foreseeable internal and external risks, maintaining safeguards against them, regularly verifying the safeguards work, and updating them as risks change (section 19 of POPIA). For a CRM, a reasonable reading of that is a short list you can show an auditor:
- Each user has their own login; no shared passwords.
- Export rights are limited to named people.
- Departing staff are removed on their last day.
- A written retention rule exists, including who deletes what, and when.
That reading is our practical interpretation rather than prescribed wording from the Act. Until this gate is done, limit access during the earlier import and test loads to the owner and one administrator.
A CRM vendor that stores your contacts may count as an operator under POPIA, and if it hosts them outside South Africa the transfer rules can apply; our posts on POPIA operator agreements and POPIA cross-border transfers cover both, and your attorney should confirm how they apply to your vendor. For marketing messages sent from the CRM, the consent rules are separate; see POPIA for marketing.
Gate 7: Run a Pilot, Then Cut Over and Retire the Spreadsheet
A pilot is a small group using only the CRM for a complete lead cycle, from first enquiry to won or lost, and it exists to find the failures that a configuration review cannot. The pilot group should include the person most sceptical of the system, since their objections are the ones the wider team will repeat.
Use this CRM go-live checklist as the final pass before the cut-over date:
- The pilot group completed at least one full lead cycle inside the CRM only.
- Every issue the pilot raised is fixed or consciously deferred in writing.
- A date is set after which the old spreadsheet is read-only.
- Each person knows where to log a call, a quote and a lost reason.
- One dashboard exists that the owner actually opens each week.
Adoption is a people problem more than a software one, and the specific tactics for getting a reluctant team to log activity are in CRM adoption and team buy-in. Whichever route you take, the read-only date matters most: while two systems stay live, staff will update the one that is easier, and the CRM loses.
Planning the cut-over and want it checked?
Share your pilot plan and the date you intend to retire the spreadsheet, and we will review it for gaps before launch.
Book a free auditAfter Go-Live: What to Review and When
The last item on a CRM implementation checklist is a post-launch review, which checks whether the CRM is solving the Gate 1 problem, and it should happen on a schedule you set in advance rather than when somebody complains. Look at four things: whether all lead sources still arrive tagged, whether duplicates are creeping back, whether stage definitions still match how the team sells, and whether the retention rule has actually been run.
If you are still choosing between tools, this CRM rollout checklist applies to any of them, and our comparisons of HubSpot vs Pipedrive and Zoho CRM vs Bigin cover the platform choice for South African budgets.
Why South African Businesses Choose Growth Pulse Media
South African businesses choose Growth Pulse Media when they want the data a CRM receives from advertising and enquiries planned before it is loaded, rather than repaired afterwards.
Dirk van Greuning built and scaled a large South African ecommerce business before founding Growth Pulse Media, and that operator background shapes how we advise. Our work sits on the acquisition and measurement side, namely Google Ads, Meta Ads and GA4, and our digital strategy work decides what the CRM has to record before money goes into traffic. All work is executed in-house, with a limited client load so senior attention stays on your account.
Who This Is Not For
This checklist is not for businesses that want a CRM configured for them, have no single owner for the rollout, or need legal sign-off on their records.
Businesses wanting someone to configure their CRM. We advise on what the system must record and how enquiries and campaign tags should reach it. We do not list CRM platform configuration among our services.
Teams without a single owner. If nobody will own the system and enforce the cut-over date, no checklist will rescue the rollout. Settle Gate 1 first.
Anyone after legal sign-off. The POPIA points here are practical reading of the Act, not legal advice. Retention periods and lawful basis for your records are for your attorney.
Frequently Asked Questions
What should a CRM implementation checklist include?
A CRM implementation checklist should cover a named owner and problem, pipeline stages and fields, data cleaning, the duplicate key, lead-source connections, permissions and retention, and a pilot before cut-over. Each item needs a pass/fail test so you know when to move on.
How long does a CRM implementation take?
It depends on how much data you hold, how many lead sources feed the system and how many people use it, so no single duration fits. Plan by gates rather than dates: move on when each exit test passes, not when a calendar says so.
Should I clean my data before importing it into a CRM?
Yes. Clean before import, because fixing duplicates and dead contacts in a live CRM is slower than fixing them in a file. POPIA section 16 also expects reasonably practicable steps to keep personal information complete, accurate and updated where necessary.
How do HubSpot and Zoho CRM handle duplicates on import?
HubSpot matches on Record ID if included; otherwise it uses a unique identifier such as a contact's email address, or a custom property set to require unique values. Zoho CRM skips a file record by default when its unique field, such as email or phone, matches an existing record, and you can choose to update instead. Features can vary by edition, so check each vendor's current documentation before a large import.
How do I get my team to use the CRM after go-live?
Set a date after which the old spreadsheet is read-only, make the CRM the only place a lead exists, and include a sceptical user in the pilot. Keeping required fields few at early stages is a practical way to make logging easier, though that is our judgement rather than a measured result.
Get your CRM rollout checked before launch
We work with Google Ads, Meta Ads and GA4 to make sure the right lead data reaches your CRM tagged by source. No obligation — we'll get back to you within 24 hours.
Talk to Growth Pulse Media

