Why fields need an audit
Fields are cheap to add and expensive to keep. Every field is something a form can write to wrongly, a report can be built on by mistake, a workflow can depend on without anyone knowing, and an import can fill with values nobody checked. Most CRMs that have run for a few years show the same symptoms. Several fields hold the same thing under different names, created by different teams or integrations. Free-text fields hold values that reports need, typed differently by every user, so the report groups dozens of spellings of the same answer. Fields that a retired form tool or a former integration wrote are still on every record, empty or frozen at the day the tool was switched off. And some fields hold Personal DataAny information relating to an identified or identifiable living person, such as a name, email address, location data or an online identifier. that no current purpose needs. Long records full of unread fields are also one of the reasons sellers stop keeping the CRM current, which the diagnosis for sales not updating the CRM separates from stages without exit criteria and reviews run from spreadsheets.
The data protection principles make the last point more than tidiness. The GDPR entry sets out data minimisation, holding only what a purpose needs, accuracy and storage limitation, keeping data no longer than necessary, and the guide to personal data in analytics applies the same principles to reporting tools. A field holding personal data with no purpose is a candidate for removal on those grounds, but the decision belongs with whoever owns data protection in the business, not with marketing alone.
The difference from other audits is worth stating. The data operations guide audits the campaign taxonomy for unused values and drift; the automation governance guide inventories workflows; the metrics layer guide defines metrics. This tutorial audits the fields themselves, which all three of those depend on.
Before you start
Gather four things. Administrator access that can see every object and field, including hidden and system fields, and export them. The workflow inventory from the governance guide, or the time to build a rough one, because workflows are the readers and writers most often forgotten. A list of every integration that writes to or reads from the CRM: forms, the email or automation platform, ad platform lead forms, calendar and calling tools, the warehouse. And a named person who owns data protection decisions, who will be asked about specific fields.
Agree the scope. Contacts and companies first is a sensible start for marketing, with deals added if marketing reports on them; trying to audit every object at once tends to stall. Agree who decides on each field: marketing can propose, but fields that sales relies on are decided with sales, and fields holding personal data with the data protection owner. Write those rules at the top of the audit sheet before the first row.
The audit sheet is one row per field, and its columns are filled in the order of the steps below. The same sheet, trimmed to the fields that survive, becomes the field dictionary at the end.
| Column | What it records | Filled in at |
|---|---|---|
| Object and field name | Where the field lives and what it is called | Export every field |
| Type and permitted values | Text, picklist, number, date and so on; the allowed options | Record type and fill |
| How often filled | Share of records with a value, recent and older records separately | Record type and fill |
| Writers | Forms, people, integrations, workflows and imports that set it | Find who writes it |
| Readers | Reports, lists, workflows, integrations and scoring rules that use it | Find what reads it |
| Personal data and purpose | Whether it holds personal data and the purpose it serves | Flag personal data |
| Decision | Keep, merge, convert or retire, with the reason and who agreed | Decide |
| Owner | The person who answers for the field from now on | Write the dictionary |
The steps
Export every field for each object
Export the full field list for each object in scope: contacts, companies and, where relevant, deals. Include system fields, hidden fields and fields created by integrations, because those are the ones nobody remembers. Put them in one sheet with a row per field and columns for everything the next steps will record. Note when each field was created and by whom, where the CRM shows it; a cluster of fields created on the same day often traces back to one integration or project.
Record type, permitted values and how often each is filled
For each field, record its type (short text, long text, picklistA CRM field type that accepts one value from a defined list of permitted values, so that records can be reported on instead of typed in many spellings., multi-select, number, date, true or false, lookup), its permitted values where it has them, and how often it is filled across the records in scope, ideally split between records created recently and older ones. A field filled on old records and empty on new ones has lost its writer. A picklist with values nobody chooses needs pruning. A free-text field with a short list of recurring answers is a picklist waiting to be built.
Find who or what writes each field
For each field, find every writer: a form (and which one), a person editing by hand, an integration, a workflow, an import. The CRM's change history on a sample of records, the form mappings, the integration settings and the workflow inventory together usually answer it. Source and campaign fields are often written by a form's hidden fieldA form field the visitor never sees that records values such as source, campaign and landing page, so each lead arrives in the CRM with its origin., so the form mappings name those writers too. A field with no current writer is frozen; a field with several writers needs a written rule about which one wins, recorded in the dictionary at the end. Note fields written by a tool that has since been switched off. A field whose value keeps reverting is usually one of these undocumented writers at work, and the diagnosis for CRM values changing back after a sync reads the record's history to find it.
Find what reads it: reports, lists, workflows and integrations
For each field, find every reader: reports and dashboards, saved lists and segments used for SegmentationDividing a market or customer list into groups that need different offers or messages., workflow triggers and conditions, integrations that send the field elsewhere, and scoring rules. Use the governance inventory for workflows and ask report owners for the reports they actually open. A field read by nothing is a candidate for retirement; a field read by a workflow nobody listed is the reason the retirement step goes in stages. The Chesterton's fence framework states the general rule behind this step: find why a field exists and who relies on it before removing it.
Flag fields that hold personal data and confirm their purpose
Mark every field that holds Personal DataAny information relating to an identified or identifiable living person, such as a name, email address, location data or an online identifier.: names, contact details, job and employer details, anything typed into free-text notes about a person, and identifiers that can be linked to someone. For each, write the purpose it serves. Take the fields without a clear purpose, and any that look excessive for the business's relationship with these people, to the data protection owner with the writers and readers already found, so the decision is made with the facts in front of them. This is educational, not legal advice.
Decide keep, merge, convert or retire
Give each field one decision. Keep: it has a writer, a reader and a purpose. Merge: it duplicates another field, and its values move into the one that stays. Convert: it holds values reports need in a form they cannot use, usually free text that should be a picklist. Retire: it has no reader, no purpose or no lawful reason to hold its data. With a long list, sort the work with the impact-effort matrix: fields that break important reports or hold unneeded personal data and are easy to fix go first; large conversions are planned as projects; fields that cost nothing to leave for now wait.
Convert free-text fields that reports need into picklists
For each field to convert, list the distinct values it holds, group them into a short set of permitted values agreed with the people who report on it, and map every old value to a new one. Create the picklist, move the mapped values into it, and send values that fit nowhere to a review list rather than inventing an option for each. Change every form, import template and integration that writes the field, or the free text comes straight back. Merges follow the same pattern: map, move, change the writers, then retire the old field.
Retire fields in stages: hide, stop writing, archive
Never delete first. Hide the field from forms, layouts and pick lists so nobody adds new values. Then stop every writer the audit found, and watch the readers it found, plus anything that complains, for a period long enough to cover the reports and workflows that run least often. If something breaks, the field is still there to restore. Then archive the values the retention rules say to keep, outside the CRM where that is the decision, and only then delete the field. Record each retirement, with its date, where the team keeps its change logA dated record of changes made to tracking, accounts, reports, lists or rules, with who made each and why, so later movements in the figures can be traced..
Write the field dictionary and the rule for adding a field
Turn the audit sheet into the field dictionary: one row per field that remains, with its name, object, type, permitted values, writers, readers, owner, purpose and whether it holds personal data. Name fields as the fundamentals guide recommends, saying what they hold and, where an integration writes them, which system; the naming conventions guide supplies the case and separator rules. Then write the rule for adding a field: a request that states the purpose, the writer, the reader, the owner and the type, checked against the dictionary for an existing field that already does the job. Where the CRM feeds metrics, link each KPIA metric tied directly to an objective. in the KPI dictionary to the fields it is built from.
A packaging firm's field audit (illustrative)
Drovetta Packaging audits its contact fields
Drovetta Packaging, an invented maker of corrugated and moulded packaging for food producers, exports its contact fields and finds 412 of them. Of those, 188 are empty on every record created in the last year, and 74 were written only by a form tool the company stopped using three years earlier. There are three industry fields, created by marketing, sales and a former integration, holding 140 distinct values between them. The source field is free text, with 96 spellings of what turn out to be 11 sources. The personal data review finds a date-of-birth field filled on 2,300 business contacts from an old event registration form; with the data protection owner, it is retired and its values deleted rather than archived. The impact-effort matrix puts the industry merge and the source conversion first, because the pipeline report depends on both. The three industry fields become one picklist with 14 values, mapped value by value. The source field becomes a picklist with 11 values, and its free-text history is kept in an archived field. 291 fields are hidden, then retired over the following quarter; one hidden field turns out to be read by a quote workflow and is restored the same day. The team publishes a field dictionary of the 121 fields that remain and a one-page form for requesting a new one.
The order matters more than the numbers: readers and writers are found before anything is hidden, personal data is decided with its owner, and nothing is deleted until it has been hidden and silent for long enough to prove it is not needed. Fields whose values arrive in a mess every day need fixing where they enter, which is a job for workflows and form design rather than for the audit, and the tutorial on standardising CRM field values builds the workflows that do it. Where a separate marketing platform writes to the same fields, the guide to syncing the CRM and the marketing platform decides which system owns each one. Lifecycle stage and the other fields that drive stages are audited here like any other field, but their definitions belong to the stages themselves, set out in the lead stages and pipeline guide alongside the lifecycle stageA CRM field recording where a contact stands in the relationship, such as subscriber, lead, opportunity or customer. rules.
Common mistakes
- Deleting first. A field that looks unused is read by a quarterly report or a forgotten workflow. Hide, stop writing, watch, archive, then delete.
- Judging fields on fill rate alone. A rarely filled field can be essential, and a well-filled one can be written by an integration nobody reads. Find the writers and readers.
- Converting a field without changing its writers. The picklist is created and the form keeps sending free text. Change every form, import and integration in the same step.
- Marketing deciding on personal data alone. Fields with no clear purpose go to whoever owns data protection, with the facts the audit found.
- Inventing a picklist option for every odd value. The report fragments again. Map to the agreed set and review what does not fit.
- Auditing everything at once. The audit stalls halfway. Start with contacts and companies, finish them, then move on.
- No rule for new fields. The dictionary is out of date within months. Every new field needs a stated purpose, writer, reader and owner before it is created.
