Every engagement runs in four steps: map, prove, connect and hand over.
Most of the problems we fix come from two systems holding different versions of the same record, with nothing checking the difference. The steps are the same whichever systems are involved.
Follow one problem through all four steps
Renewal dates in Salesforce no longer match the signed contracts. Step through what we hand you at each stage.
Find where the numbers split
We list every system, integration, automation and report that touches the customer record, then compare them field by field. Every conflict goes into one register with the records it affects.
What you receive
- A map of your systems and integrations
- A register of every conflict, with the records affected
- Fixes ranked by effort, risk and impact
| Field | Systems that disagree | Accounts | What it costs |
|---|---|---|---|
| Renewal date | Salesforce and the signed contract | 312 | Notices sent on the wrong date |
| Billing contact | Salesforce and billing | 96 | Invoices bounce |
| Account owner | Salesforce and the support desk | 41 | Tickets reach former staff |
Reproduce it on real records
Before anything changes, we reproduce each conflict on real records and trace it to the rule that causes it. Each fix is sized in records and revenue, so the order of work is a business decision.
What you receive
- The root cause of each issue
- Its impact in records and revenue
- A go or no-go for each fix
- Contract signedRenewal date Apr 15, 2027
- Nightly job runsCopies the old renewal field over the new one
- Salesforce showsMar 1, 2027, six weeks early
Cause: an automation from an earlier rollout still writes the retired field. Impact: 312 accounts, nine notices already sent early.
Build it, rehearse it, release it
We build in a sandbox and rehearse at full production volume. Every bulk change keeps a copy of the data it changes and a file that reverses it, and the release sits behind a switch until you sign off.
What you receive
- The production build
- Counts checked after release
- A rollback plan for every change
- Records updated
- 312 of 312
- Failures
- 0
- Copy of the old values
- Saved before the run
- Rollback file
- Tested on the sandbox
- Old automation
- Switched off, kept for 30 days
- Release
- Behind a switch, waiting for sign-off
Leave it running
Each fix ships with a runbook, a named owner and a check that runs on its own. If the numbers start to drift again, the owner hears about it the same day.
What you receive
- Runbooks and a data dictionary
- A named owner for every component
- Monitoring that fires when a sync stops
- Source of truth
- The latest signed contract
- Daily check
- 9:00, every account with a contract
- Alert
- Email to the owner if any date differs
- Last check
- 0 differences
Example problem and artifacts with invented figures.
Every release passes five gates
Nothing reaches production without them. Pick a gate to see what it stops, and how we recover if something still goes wrong.
What it stops
A change that works on twenty test records and times out on two hundred thousand.
If something goes wrong
Fix it in the sandbox. Production never sees the failure.
What it stops
A bulk update nobody can undo.
If something goes wrong
Every changed field is restored from the saved copy.
What it stops
Criteria that match more records than anyone meant them to.
If something goes wrong
The list is reviewed record by record before anything is written.
What it stops
A release that can only be turned off with another deploy.
If something goes wrong
Turn the switch off. The old behaviour returns in seconds.
What it stops
A green deploy with the wrong number still on the report.
If something goes wrong
The release is not done until the report shows the right figure.
Ways to work with us
Most clients start with the diagnostic, build what it ranks highest, and keep us on for the systems we touched.
The lighter part of the build bar shows the range. Most builds land between six and twelve weeks.
Systems diagnostic
A map of your systems, every conflict with the records it affects, and fixes ranked by effort and risk. You keep it whether or not we do the build.
Best forKnowing what to fix first
Book a diagnosticBuild
One outcome in production, such as an integration, a migration, a contract workflow or a cleanup, rehearsed at full volume and handed over with runbooks.
Best forA problem the diagnostic has sized
Fractional platform lead
Senior ownership of Salesforce and the systems around it: architecture decisions, release governance and integration health.
Best forTeams without a full-time architect
What we are building next
The areas we are investing in for 2027.
Records AI agents can trust
Identity resolution, agreed definitions and per-user permissions, so AI agents act on records that agree.
Integration monitoring
Reconciliation counts and alerts on silence as a standard part of every pipeline.
Re-platform programs
Moving commerce, CRM or billing without losing lineage, history or revenue.
Contract data
Signed terms read, structured and kept current as contracts change.
Talk to us about your systems.
Tell us which systems are involved and what is going wrong. We reply within one business day, and new engagements begin in January 2027.