Guide

Salesforce technical debt audit: what it covers and how to run one

A Salesforce technical debt audit lists everything in an org’s setup that makes it slow, risky or expensive to change: retired automation, untested Apex, fields nobody can explain, access nobody reviewed, integrations about to break. A good one ranks what it finds and puts a dollar figure on each item, so the cleanup becomes a budget line instead of a wish list.

By VanguardOps, written by a Salesforce architect with 13+ years in enterprise orgs. Updated November 17, 2026.

What should a technical debt audit cover?

Six areas. Every one of them lives in metadata, so none of it needs access to customer records.

What a Salesforce technical debt audit covers
AutomationWhat to look for: Active Process Builder processes and workflow rules; several flows, triggers and rules firing on the same object and eventWhy it costs money: Support for Process Builder and workflow rules ended December 31, 2025, and stacked automation makes every change slower to test
CodeWhat to look for: Apex classes and triggers with low or no test coverage; old code nobody ownsWhy it costs money: Untested triggers can corrupt data quietly, and old code blocks deployments
Data modelWhat to look for: Custom fields and objects with no description; unused fields; validation rules that conflictWhy it costs money: Every project starts by working out what a field means
AccessWhat to look for: Permission sets nobody holds; powerful permissions held by many usersWhy it costs money: Excess access is the finding an auditor or a breach finds first
IntegrationsWhat to look for: Connected apps with broad access; named credentials and certificates near expiryWhy it costs money: An expired certificate stops an integration on a date you could have known
LimitsWhat to look for: Org limits close to their ceiling; objects with very large record countsWhy it costs money: Hitting a limit stops work in production, usually at the worst time

If the org runs Service Cloud, add case setup, entitlements, SLAs and routing.

How do you run one? An eight-step plan

  1. Agree the question first. “What is slowing us down and what does it cost?” gives a different audit from “are we ready for Agentforce?” Write it down.
  2. Take an inventory. Count objects, fields, flows, processes, workflow rules, Apex classes, triggers, permission sets, connected apps and named credentials. The count alone tells you how big the job is.
  3. Map automation by object. For each busy object, list everything that fires on create and update. Stacked automation is where most surprises come from.
  4. Check code health. Run all Apex tests and note every class and trigger with low coverage, and which ones nobody on the team can explain.
  5. Review access. For each permission set, how many people hold it, and who holds the powerful ones.
  6. Check integrations and limits. Expiry dates on certificates, what each connected app can reach, and which org limits are close to full.
  7. Price every finding. Hours to fix times an hourly rate, plus what it costs each year to leave it. See the formula below.
  8. Rank and plan. Sort by 3-year cost, not by count. Fix the expensive and risky first, and schedule the rest into normal releases.

How do you put a dollar figure on technical debt?

Two numbers per finding. The fix cost is the hours to fix it times the rate of the person who will do it. The carrying cost is what it costs each year to leave it: slower changes, more testing, more risk. VanguardOps prices every finding this way, with the formula and rates published.

A worked example: 12 active Process Builder processes, rated high severity, fixed by an architect at $175 an hour.

Worked example: pricing 12 active Process Builder processes
Value
Hours to fixAbout 17 hours
One-time fix cost$3,024
Yearly carrying cost$3,024
3-year cost of leaving it$9,072

These are directional estimates for prioritizing and budgeting, not a quote. Try your own numbers in the calculator.

Should you audit by hand or with a tool?

Steps 2 to 6 are counting and checking, and a metadata scan does them in minutes. Steps 1 and 8 need someone who knows the business. The usual split is a scan for the inventory and the prices, then a person for the plan. Five ways to do it, compared.

A VanguardOps scan covers steps 2 to 7 for Sales Cloud and Service Cloud orgs. It reads metadata only, changes nothing in the org, and returns each finding with its fix cost, yearly cost and 3-year cost.

Common questions

What is a Salesforce technical debt audit?
A review of how a Salesforce org is built (automation, code, data model, access, integrations and limits) that lists what makes the org slow, risky or expensive to change, ranks it, and ideally puts a cost on each item so the cleanup can be budgeted.
How long does a Salesforce org audit take?
A manual audit by an admin or consultant usually takes days to weeks, depending on the size of the org. An automated metadata scan takes minutes and gives you the inventory; deciding what to fix first still takes a person.
How often should you audit a Salesforce org?
At least once a year, and before any big change: a new cloud, an Agentforce rollout, a merger, or a new admin taking over. Seasonal releases are a natural checkpoint too, because each one can change how old automation and code behave.
What is the difference between a technical debt audit and Salesforce Health Check?
Salesforce Health Check in Setup compares your security settings with a baseline. A technical debt audit is wider: it looks at automation, code, the data model, access, integrations and limits, and asks what each problem costs to keep.
Can an audit read my customer data?
It doesn't need to. Technical debt lives in metadata: how the org is set up. VanguardOps reads metadata only and never the contents of records.