The test for any customisation

Who acts on this? Name the person and what they do differently because of it. Fields, automations and reports that fail this test are pure cost — they slow every interaction and degrade the data that does matter.

Over-customisation is not caused by bad decisions. It is caused by a hundred reasonable ones, none of which were ever revisited.

The warning signs

SignWhat it indicates
Nobody knows what happens when a record savesAutomation has passed comprehension
Users have spreadsheets alongside the CRMThe system does not fit the work
Simple changes take weeksEverything is entangled
Fields nobody can explainRequirements nobody revisited
Upgrades are fearedFragility
Only one person understands itKey-person risk
Training takes daysComplexity passed on to users
Bulk imports are avoidedAutomation is not bulk-safe

Users keeping spreadsheets is the clearest signal available. It means the system, as customised, does not support the work — so people rebuilt it somewhere they control. Adding more customisation in response usually makes it worse; asking them what the spreadsheet does is the productive move.

How it accumulates

  1. A field is added because someone asked. Reasonable in isolation.
  2. It becomes mandatory so it gets populated.
  3. Automation is added to derive it when users leave it blank.
  4. Exceptions are added for cases where the automation is wrong.
  5. The person who wanted it leaves. Nobody uses it.
  6. It stays, because removing it feels risky.
  7. Repeat forty times over four years.

Nothing in that sequence is a mistake. The mistake is that step six never triggers a review, so the cost compounds while the benefit disappears.

Auditing what you have

Ask of eachKeep ifRemove if
Custom fieldSomeone acts on itPopulated rarely, used never
Mandatory fieldA decision depends on itUsers enter anything valid
AutomationPurpose is stated and currentNobody knows why it exists
ReportSomeone opened it this quarterBuilt once, never revisited
Custom objectActively usedSuperseded, still present
Validation rulePrevents a real errorUsers work around it

Field usage data settles the argument. Report on how many records have each custom field populated in the last year. Fields at low single-digit percentages are not in use, whatever anyone says in the meeting. This turns a debate about opinions into a decision about evidence.

Simplifying safely

  1. Measure usage for every custom field, report and automation.
  2. Ask "who acts on this?" for anything that survives the usage cut.
  3. Remove from page layouts first — reversible in seconds, and it surfaces objections quickly.
  4. Deactivate automation rather than deleting it.
  5. Wait a full business cycle — a quarter catches most dependencies.
  6. Then delete, having exported the data first.
  7. Record what was removed and when, so the next person understands.

Preventing the next accumulation

  • Every new field needs a named person who acts on it. No name, no field.
  • Review annually against usage data, as a scheduled activity.
  • Someone has authority to say no to requests. Consensus never declines anything.
  • Prefer configuration to code where both work.
  • Document why, not just what — six months later the reason is what matters.
  • Set a removal expectation when adding something for a time-limited need.

What simplification typically returns

  • Faster record entry, which improves data quality on the fields that matter.
  • Changes that take days rather than weeks.
  • Training that takes hours rather than days.
  • Upgrades that stop being frightening.
  • An org more than one person understands.

None of that appears on a project plan as a benefit, which is why simplification rarely gets funded and almost always pays back.

Suspect your org has accumulated past the point of usefulness? Tell us what users complain about. See org health audits, why implementations fail, and platform automation.

Frequently asked questions

Necessary customisation serves a process someone can explain and a person who acts on the output. Over-customisation serves a request nobody remembers making. Asking "who acts on this?" for each field and automation separates them quickly.
Yes, carefully. Deactivate before deleting, watch for a full business cycle, then remove. The order matters — deletion is irreversible and something always turns out to depend on the field nobody used.
The process owner, with the data. Removal by consensus never happens, because someone always defends every field. Someone needs authority to decide on evidence of use.