What goes wrong in an import
Imports fail in ways that raise no error. The file loads, the count looks right, and the damage is found weeks later by a seller who calls a customer about a product they already own, or by a complaint from someone who unsubscribed a year ago. The causes are few and predictable.
| What goes wrong | How it shows up | The step that prevents it |
|---|---|---|
| Nobody checked what people were told | Marketing emails to people who only agreed to a call back, or to a bought-in list | Confirm where the list came from |
| Records get no source or campaign | Reports credit "import" or show a large blank segment | Decide source, campaign and lifecycle values |
| Several people per row, mixed formats | Contacts named after companies, countries spelled three ways | Clean the file to one person per row |
| Suppressed people re-created | People who opted out are emailed again | Check against the suppression list |
| Columns mapped loosely | Values land in the wrong field or outside the picklist | Map columns and picklist values |
| Existing records overwritten | Customers reset to new leads; original source erased | Choose the matching key and the rule on a match |
| Workflows fire for the whole file | A welcome or nurture sequence goes to everyone at once | Check which workflows the import will trigger |
| No way to find the import later | A bad load cannot be removed cleanly | Import the rest with a batch tag |
The data operations guide lists an import check that validates format, source, consent basis and duplicates before a list enters the CRM. This tutorial is that check written out as a procedure, from the moment someone hands over a file to the moment its records belong to someone.
Before you start
Gather four things. The file itself, with a note from whoever supplied it saying where it came from: an event, a partner, an old system, a seller's spreadsheet. The CRM's field list for contacts and companies with the permitted values of every picklist, so that mapping is done against what exists rather than what people remember. The current suppression list in a form that can be matched against the file. And the list of active workflows from the automation governance guide's inventory, or access to see them, because the import will be read by every workflow whose trigger it meets.
Agree two things in advance. Who owns data protection decisions, because the first step may need one. And which parts of the import can be walked back. The one-way and two-way doors framework is useful here: a tagged import of new records that can be found and removed is a two-way door, decided quickly; an overwrite of existing fields, a merge, or a workflow that sends messages is a one-way door, decided slowly and by a named person. The rest of the steps are arranged so that the one-way doors are opened last, after the sample has shown what they will do.
The steps
Confirm where the list came from and what people were told
Ask the person who supplied the file how each name on it was gathered and what the people were told at the time. That answer decides what follow-up is possible. The GDPR entry sets out that each purpose for processing Personal DataAny information relating to an identified or identifiable living person, such as a name, email address, location data or an online identifier. needs a Lawful BasisOne of the six grounds the GDPR accepts for processing personal data, such as consent, contract or legitimate interests, chosen for each purpose. chosen before the processing starts. The list building and consent guide notes that exports from an old platform are usually fine when the consent recordsThe stored evidence of a marketing permission: the channel, the wording the person saw, the date and the source of each opt-in, kept with the contact record. travel with them, that files from sales or events depend entirely on what people were told, and that regulators treat bought-in lists as a risk area in their guidance; the UK regulator's direct marketing guidance, listed in sources, is one example. The trade show tutorial adds that a badge scan is not consent to marketing. Record the answer on the import, and where it is unclear, take the question to whoever owns data protection rather than deciding it in marketing. This is educational, not legal advice. For an event list, the in-person event readiness checklist settles what visitors are told and checks that the import will keep it before the event, so this answer exists when the file arrives.
Decide the source, campaign and lifecycle values the records get
Every new record needs an original source that says where it really came from, such as the event or partner by name, never a generic "import". Decide the campaign the people join and their member status, using a campaign record named from the taxonomy, so that the guide to measuring webinars and events can later report the list by attendance and pipeline. Decide the lifecycle stageA CRM field recording where a contact stands in the relationship, such as subscriber, lead, opportunity or customer. new records start at, and write down that existing records keep theirs. Record what people were told in a consent or purpose field, per channel, as the first step found it.
Clean the file to one person per row with consistent formats
Split rows that hold two people, remove rows that hold only a company or a role, and drop columns nobody will use; a column with no purpose is Personal DataAny information relating to an identified or identifiable living person, such as a name, email address, location data or an online identifier. collected for nothing. Make formats consistent: one way of writing countries, phone numbers with their country codes, names in separate fields, email addresses trimmed and in lower case. Remove duplicates inside the file itself so that the CRM never has to choose between two rows for the same person.
Check the file against the suppression list
Match the file against the suppression listA list of addresses or contacts excluded from every send, such as unsubscribes, hard bounces and spam complaints. before anything is loaded, and take every match out of the file or mark it so that the import cannot make that person reachable. The suppression list entry explains why suppressed people are kept on file rather than deleted: a deleted record can come back in the next import and be mailed as if nothing had happened. The email list hygiene SOP makes the same rule standing practice. Once a person objects to direct marketing, the GDPR entry notes, their data is no longer processed for that purpose, whatever file they reappear in.
Map columns to fields and picklist values
Build a map with one row per column: the CRM field it goes to and, for every picklist, a value-by-value map from what the file says to what the CRM permits. Values that fit no permitted option go to a review list, not into a new option created on the fly. Columns the CRM has no field for are dropped unless there is a reason to create a field, which is a decision for whoever owns the CRM's fields, not for the import.
Choose the matching key and what happens on a match
Decide how a row is recognised as a person already in the CRM. The CRM fundamentals guide matches contacts on email and companies on website domain, and sets merge rules in advance. Then decide, field by field, what a match is allowed to change. A safe default: the import may fill empty fields and add the person to the new campaign, but never overwrites original source, created date, lifecycle stage, owner or consent. Rows with no usable key, such as a name and a company but no email address, go to a review list rather than creating records that will become DeduplicationRemoving double counts when one conversion is reported by more than one source. work later. If an earlier import has already left pairs behind, the diagnosis for duplicate contacts in the CRM traces each pair back to the batch, form or sync that created it.
Import a small sample and inspect the records
Choose a sample that covers the awkward cases: rows that match existing customers, rows that match open leads, new people, rows with unusual picklist values and rows from each source in the file. Import it, then open the records one by one and compare them with the file and with what they looked like before. Check the source, the campaign membership and status, the lifecycle stage, the owner and the consent fields. Fix the map or the match rule and repeat on a fresh sample until every record looks right. Remove each sample by its tag before the next.
Check which workflows the import will trigger
List every active workflow whose trigger the imported records will meet: a contact created, a list joined, a campaign membership added, a lifecycle stage set. The workflows hub warns that a workflow can enrol every contact that meets its trigger, and an import creates many such contacts at once. For each one, decide whether these people should enter it. A welcome series written for people who subscribed is wrong for a conference list; an internal task for the owner is usually right. Exclude the batch from the workflows it should not enter, by its tag, before the full import.
Import the rest with a batch tag
Give every row a batch value, a field or a static list naming the import and its date, so that the records it created and changed can be found together later. Load the full file with the same map and match rule as the last clean sample. A tagged import of new records is reversible; changes to existing records are not, which is why the rules for them were settled on the sample.
Reconcile the counts and hand the records to their owners
Compare the rows in the file with what the CRM now holds: created, matched and updated, sent to review, suppressed and skipped. Every row should be in exactly one of those groups, and a gap means a rule did something unexpected. Work the review list. Then make sure every new record has an owner by the CRM's assignment rules, and tell the owners what arrived, from where and what the people were told, so the follow-up stays within it. Where consent for marketing is unclear, a single re-permission message is the route the list building guide describes for turning an unknown list into a known one.
A lab equipment supplier's event import (illustrative)
Skelmira Lab Equipment imports two event lists
Skelmira Lab Equipment, an invented supplier of benchtop instruments to university and hospital laboratories, comes back from a conference with 640 badge scans and receives 1,120 registrants from a partner's webinar. The supplier confirms what each group was told: visitors to the stand were told a specialist would follow up about the instrument they asked about; webinar registrants agreed to hear from the partner and from Skelmira about the session's subject. Both lists get their own campaign records and an original source naming the event. The suppression check removes 37 people. The first sample of 50 rows shows two problems. Eleven rows matched existing customers, and the match rule as first drafted would have reset their lifecycle stage to lead and overwritten their original source. And the welcome workflow, triggered by any new contact, was about to send its first email to every new record. The team changes the match rule so existing records keep their stage, source and owner, excludes the batch tag from the welcome workflow and gives owners a call task instead. A second sample loads cleanly; the full import reconciles to the row, and 28 rows with no email address go to the review list.
The pattern holds for any list, including bulk updates to records already in the CRM: decide the rules, prove them on a sample, tag the batch and reconcile. Moving a whole system is a larger version of the same discipline, and the tutorial on moving to a new CRM covers it.
Verify it worked
- Reconcile the file. Every row should sit in exactly one group: created, matched and updated, sent to review, suppressed or skipped. In the illustrative example above, the rows with no email address sit in review.
- Filter by the batch tag. It should list every record the import created or changed, and nothing else.
- Open sample records beside the file. Source, campaign membership, lifecycle stage, owner and consent fields should match the plan. Existing customers should keep their stage, source and owner.
- Check workflow enrolments for the batch. Only the workflows you chose should have picked it up.
- Run the source report. It should name the event or partner, never "import".
If it goes wrong
- The counts do not reconcile. A rule did something unexpected. Find the rows in no group and check the match key before loading more.
- Customers show as new leads or have lost their original source. The match rule overwrote existing fields. Stop loading and fix the rule on a fresh sample. Deleting the import will not undo it.
- A welcome or nurture message went to the batch. A workflow trigger met the import. Pause that workflow, exclude the batch by tag and tell whoever owns data protection.
- Opted-out people are reachable again. The suppression check was skipped or incomplete. Match the file against the suppression list again and keep those records marked suppressed.
Next steps
- Read the CRM fundamentals guide to review the matching and merge rules your import relied on.
- Read the list building and consent guide to turn a list with unclear consent into a known one.
- Read the trade show tutorial to see why a badge scan limits what you may send an event list.
Common mistakes
- Importing before asking what people were told. The answer decides what follow-up is allowed; finding out afterwards means some messages have already gone.
- Letting the source read "import". Every record needs the real event, partner or system it came from, or the source report loses the list entirely.
- Skipping the suppression check. Opted-out people come back as new contacts. Match before loading, every time.
- Overwriting on match. Customers become new leads and their original source disappears. Fill empty fields only, unless a field-by-field decision says otherwise.
- Creating picklist values during the import. Every new spelling splits a report. Map to existing values or send the row to review.
- Loading the whole file first. A sample of awkward rows finds in minutes what a full load hides until someone complains.
- Forgetting the workflows. The import is a trigger. Check every workflow that will read it and exclude the batch where it does not belong.
- No batch tag. Without it, a bad import cannot be found, let alone removed.
