May 21, 2025
Most best-practice advice for admins is a list of topics, which is why it is hard to act on. A better frame is a cadence: what a Salesforce administrator checks daily, reviews weekly, works through at each release, audits quarterly, and tests once a year. Same practices, attached to a calendar. That is what makes them actually happen.
Here is what the topic-list approach produces. An admin reads a good article, agrees with all of it, and changes nothing, because "maintain data quality" is not a task you can schedule. Meanwhile the org drifts. Nine months later a report shows the wrong number, and the investigation finds two automations writing the same field, a permission set nobody remembers assigning, and a custom object near its field limit. Nothing failed loudly. It just decayed.
This post gives you the cadence, the exact checks at each interval, and the numbers to pull so you can prove the org is getting healthier and faster rather than hoping it is. Start with the first table and turn it into recurring calendar entries.
Five intervals. Each catches a different class of problem, and none substitutes for another.
Table 1 — The Admin Cadence at a Glance
| Cadence | What you do | Budget | What it catches |
|---|---|---|---|
| Daily | Failed logins, failed automations, integration errors | 15 minutes | Breakage before a user reports it |
| Weekly | Permission changes, user status, orphaned records, failed subscriptions | 45 minutes | Access drift and abandoned data |
| Every release, three times a year | Review release notes, test in a sandbox, brief users, keep your credential current | 1–2 days | Features changing under you |
| Quarterly | Debt metrics, automation audit, storage, page and report performance | Half a day | Slow decay nobody notices |
| Annually | Documentation test, credential access, restore test | 1 day | Single-admin risk |
The intervals are the point. A quarterly audit will never catch a failed integration, and a daily check will never catch a record page that has grown four seconds slower.
Three things, and you are done.
Failed logins. Setup → Login History, filtered to failures. A handful is normal. A pattern from one IP, or a spike against one user, is not.
Failed automations. Your Flow error emails go somewhere. If that somewhere is an inbox nobody opens, fix that first. A silently failing Flow writes nothing and tells no one.
Integration errors. Every connected system has an error queue. Check it. A middleware job that failed at 2am has been failing all morning, and the sales team is working from yesterday's data without knowing it.
The value of the daily check is not thoroughness. It is that you find out before a user does. Practical walkthroughs of these monitoring screens are posted regularly on Salesforce Geek, and our Salesforce administration practice runs this as a standing daily deliverable rather than a favor.
Weekly work is about drift — small changes that are individually fine and collectively a problem.
Permission changes. Setup → View Setup Audit Trail, filtered to the last seven days. Every permission change should map to a request you recognize. One that does not is either an undocumented favor or something worse. Review connected apps on the same pass, since an over-scoped integration is the same risk with a longer reach — our post on Data Loader security risks covers what to look for.
New and deactivated users. Confirm leavers are deactivated, not just unassigned. An active login for someone who left in March is the finding that ends an audit badly.
Orphaned records. Records owned by inactive users stop appearing in the right reports and stop being worked. Build one report on owner-is-inactive and run it weekly.
Failed report subscriptions. When a subscribed report stops delivering, executives do not chase it. They stop looking, then decide from memory instead.
A Salesforce system administrator who does this consistently spends far less time on emergencies, because most emergencies are drift that ran unnoticed for six months.
Salesforce ships three releases a year. Each one changes behavior in your org whether you engage with it or not, which makes the release the most important interval in the cadence.
Filter the release notes. Search for the objects and features your org actually uses rather than reading front to back.
Check Release Updates in Setup — formerly called Critical Updates. These are changes Salesforce will enforce on a stated deadline whether you act or not, and they are the ones that break things.
Test in a sandbox before the release reaches production. Then send users a short note covering only what changes for them, not the full notes. Release-by-release breakdowns are covered well on SFDCStop.
Your credential is not permanent, and letting it lapse is an avoidable self-inflicted problem.
Salesforce publishes maintenance content on Trailhead alongside its releases, and completing it is what keeps the credential active. Confirm the current requirement and deadline on Trailhead directly, because the modules and dates change with each release and any article quoting a specific date will be wrong within months — including this one.
Two naming notes worth carrying. Trailhead now lists the credential as Salesforce Certified Platform Administrator, with Platform Administrator II replacing what most people still call Advanced Administrator. Job postings and contracts will say certified Salesforce administrator for years, so recognize both. And the practice itself is simple: attach your Salesforce admin certification maintenance to the release you are already testing, rather than discovering an expiry notice in your inbox.
Everyone says avoid technical debt. Almost nobody says where to look or what number is bad. These eight are all pullable in an afternoon.
Table 2 — Quarterly Org Debt Metrics
| Metric | Where to find it | Look closer when | What it costs you |
|---|---|---|---|
| Custom fields per object against your limit | Object Manager field list; limits under Company Information | You are near the edition limit on a core object | You cannot add a field the business needs |
| Fields holding no data | A report per field, or field usage tooling | A field is empty on most records | Page clutter and slower data entry |
| Duplicate or inactive automations | Setup → Flows, and Process Automation | Two automations write the same field | Race conditions and values nobody can explain |
| Permission sets assigned to nobody | Setup → Permission Sets, assignment counts | Any set shows zero assignees | Audit noise and access confusion |
| Users holding the System Administrator profile | Setup → Users, filtered by profile | More than a couple of people hold it | One mistake with org-wide reach |
| Reports and dashboards by last run date | Reports tab, sorted by last run | Nobody has run it in a year | Storage, clutter, and false confidence |
| Data and file storage consumed | Setup → Storage Usage | Above roughly three quarters full | A hard stop on data loads at the worst moment |
| Audit trail entries you cannot explain | Setup → View Setup Audit Trail | Any change with no request behind it | Undocumented drift and no rollback story |
Record the numbers each quarter in the same place. One quarter's figures tell you nothing. Four quarters of the same figures tell you whether the org is getting healthier or quietly worse, and that trend is the only debt argument leadership responds to.
If Process Builder or Workflow Rules are still running, count them as debt. Both are now retired in favor of Flow, and every quarter you leave them in place adds migration work. Migration patterns are documented in detail on Salesforce Tutorial.
Run a duplicate and validation check on your core objects during the same pass. Our post on strengthening data management and security covers which parts of that to automate.
Hygiene metrics tell you the org is tidy. They do not tell you it feels fast. Add four performance checks to the same quarterly pass.
Page load times. Setup → Lightning Usage App reports your slowest pages and how long users wait on them. Start with whichever record page your busiest team lives on.
Record page complexity. Every extra component, related list, and dynamic-form condition on a page costs load time. If a page has grown for three years without anything being removed, that is where the seconds went.
Long-running automations. Check your debug logs for automations with unusually long run times. A Flow that takes four seconds on save is four seconds every user pays, on every save.
Reports and list views that time out. These almost always filter on a non-indexed or formula field, or scan more records than they need. Salesforce also documents thresholds for ownership and account data skew — confirm the current figures in its documentation and check your largest objects against them.
Performance decay is gradual, which is why it needs a fixed interval. Nobody files a ticket saying the page got half a second slower this quarter. They just start avoiding it.
This is the practice almost no admin guide covers, and the one your leadership should care about most.
Most orgs have one person who knows how the org actually works. Not what the documentation says — which Flow fires on save, which field the integration writes, why that validation rule has an exception for one profile.
Test it rather than assume it. Have someone else attempt one routine change using only your written material, while you say nothing. Where they get stuck is your documentation gap, and it will be somewhere you never thought to write down. Fix the three worst gaps and repeat next year. Approaches to documentation that people actually maintain are covered on SFDC Fanboy.
Confirm two more things annually. That more than one person can reach every admin credential and integration account — including any AI agent granted admin rights, which our post on AI in administration services covers in depth. And that your backup has actually been restored in a test, not merely scheduled. A backup nobody has restored is a hypothesis.
Attach them to intervals. Check failed logins, automations, and integration errors daily. Review permission changes and user status weekly. Test every release in a sandbox. Audit org debt and performance quarterly. Test documentation and restore readiness annually.
Daily takes 15 minutes, weekly around 45, each release one to two days, quarterly half a day, annual one day. The value comes from the interval being fixed, not from any single check being exhaustive.
Usually page complexity, a long-running automation on save, or a report scanning more records than it needs. Start with the Lightning Usage App to find the slowest pages, then strip components off the worst one and check what runs on save.
Access, data integrity, automation health, and change control. A Salesforce CRM administrator is accountable for the org staying trustworthy and usable — not for building every request that arrives.
Pull the eight metrics in Table 2 every quarter and record them in the same place. One quarter is a snapshot. Four quarters is a trend, and a trend is what gets budget approved.
Keep the parts that need business context: deciding what gets built, saying no, and knowing which stakeholder approves what. Salesforce admin services suit the recurring, evidence-based work — the daily and weekly checks, release testing, and the quarterly audit. Our comparison of in-house versus outsourced administration works through the split.
The difference between an org that scales and one that decays is rarely skill. It is whether the maintenance work has a schedule and an owner. Minuscule Technologies works as a Salesforce engineering partner, not a ticket queue. Our Salesforce administrator services are built around this cadence — the daily and weekly checks run as a standing deliverable, releases are tested before they land, and the quarterly debt and performance numbers arrive as a trend line rather than a status update.
Three things shape how we deliver. Our Accelerators and Starter Packs supply the pieces the cadence needs: prebuilt debt-metric reports, a page-performance baseline, a release-testing checklist mapped to your object model, permission-drift audit queries, and a documentation template built to survive a handover. Our Agentic DevOps practice runs configuration changes through AI-assisted CI/CD, so a change to a Flow or a permission set is a reviewable, reversible deployment rather than a click in production. And our re-engineering work clears the debt an org accumulated before anyone was measuring it, with the same numbers reported quarter over quarter.
Book a free org health review. We will pull the eight debt metrics against your org, baseline your slowest record pages, list the Release Updates you have pending, check whether anyone but your admin can reach every credential, and test your documentation against a real change. It takes about a week, there is no obligation, and the findings are yours either way. Schedule your strategic Salesforce call and start with numbers instead of a list of good intentions.
You've seen what's possible. Now, let's make it happen for your business. Whether you need an end-to-end Salesforce solution, a complex integration, or ongoing managed services, our team is ready to deliver.
Schedule a Free Strategic Call
