Get in Touch
About Us Insights Newsroom Gallery Why AtomicNorth Get in Touch

Salesforce
Migration

A migration is judged on one thing: whether the people who used the old system trust the new one in week one. Record counts do not establish that. Reconciliation the business has signed does.

Rehearsed in a full-copy sandbox, reconciled field by field, with a rollback position that exists.

RehearsedIn a full sandbox

A migration that has not been run end to end is a plan, not a rehearsal.

ReconciledField level

Counts, values and exceptions, signed by the business rather than the project.

RollbackDefined

A stated position for going back, agreed before cutover rather than improvised.

LegacyKept readable

Read-only access to the old system for as long as the business needs it.

What we move

Four migrations, four different problems

They share a method and almost nothing else. Treating a consolidation like a CRM move is how a merger loses its pipeline.

Legacy CRM to Salesforce

From an on-premises or hosted CRM, usually with years of local conventions, free-text where a picklist should have been, and reports nobody can reproduce.

The hard part is rarely extraction. It is deciding what the fields were actually being used for, which only the users can tell you.

A model that matches how the work is done.

Org consolidation

Two or more Salesforce orgs into one after an acquisition, where both sides have a working data model and neither wants to be the one that changes.

The decisions are organisational before they are technical. We make that explicit early instead of letting it surface at cutover.

One org with a model both sides recognise.

Classic to Lightning

Moving the interface and, more importantly, the automation and customisation behind it that was written against the old one.

The interface switch is the visible part and the small part. Overlapping legacy automation is the work.

A move that does not re-break what worked.

Data and deduplication

Extraction, cleansing, matching and merge rules agreed with the business, with the survivorship decisions made by the people who own the records.

Merging accounts on rules the business has not approved is the quickest way to lose their trust permanently.

Duplicates resolved the way the owners wanted.

Attachments and history

Files, notes, activity history and audit trail, which is usually the largest volume and the part that gets deferred until it is a problem.

A migration that moved the records but not the correspondence is one users work around rather than adopt.

The context, not only the record.

Legacy decommission

A read-only period on the old system, a retention position agreed with whoever is responsible for it, and a decommission date that is a decision rather than a drift.

Systems kept alive because nobody decided to switch them off are a cost and a liability at once.

An ending that was planned.

Migration, rehearsed and reconciled

StageOneProfile

The source data profiled rather than described: actual field usage, value distributions, duplicate rate, orphaned records and the conventions nobody documented.

What the old system was used for is almost never what its documentation says it was for.

You receive:

A data profile and a mapping

Field-level usage and quality, the duplicate position, and a target mapping with transformation rules and a survivorship policy.

This is yours whether or not the migration continues.

StageTwoRehearse

A full migration run into a full-copy sandbox, timed, with exceptions captured and the mapping corrected. Then run again, until the result is boring.

The first rehearsal always finds something. The purpose of the second is to establish that the first fix worked.

You receive:

A timed run and an exception list

Elapsed time, record counts, exception categories with volumes, and a corrected mapping for the next pass.

Cutover timing is estimated from this, not guessed.

StageThreeReconcile

Field-level reconciliation the business reviews and signs: counts, totals that must match the finance system, spot checks on records the users pick themselves.

Users should choose some of the records checked. The ones they pick are the ones they do not trust.

You receive:

A signed reconciliation

Counts, control totals, exception dispositions and the spot checks, signed by the data owner rather than the project manager.

Go or no-go is decided on this document.

StageFourCut over

A rehearsed cutover with a freeze window, the migration run, reconciliation repeated in production, and the legacy system placed in read-only rather than switched off.

The rollback position is agreed before the window opens. Deciding during it is how a bad hour becomes a bad month.

You receive:

A live system and a readable legacy

Production reconciliation, the open-exception list with owners, and legacy read-only access for as long as the business needs it.

Decommission is a later, separate decision.

01
Risk

Where migrations actually go wrong

None of these are exotic. They are the four that recur, and each has a cheap mitigation if it is decided early.

Salesforce Migration: risk, risk, how it shows up, mitigation, decided by
RiskHow it shows upMitigationDecided by
Dirty source dataDuplicates and inconsistency arrive intact and users stop trusting the systemProfile first, cleanse before load, matching rules the business approvedData owner
Unrehearsed cutoverElapsed time is discovered on the night and the window is missedFull-copy rehearsal, timed, run more than onceProgramme
Deferred attachmentsRecords arrive without the correspondence and users keep the old system openFiles and history scoped in from the start, not as a phase twoSponsor
No rollback positionA problem during cutover becomes an improvised decision at 2amRollback agreed and documented before the freeze window opensIT and sponsor jointly
Survivorship by defaultMerge rules chosen by the tool rather than by the businessRules written, reviewed and signed before any merge runsData owner
Open-ended legacyThe old system runs for years because nobody decided to stop itA read-only period and a decommission date agreed at cutoverSponsor
Method

Boring on the night, by arrangement

A cutover should be the least interesting part of a migration, because everything that could be found has already been found in rehearsal.

That costs more in week three and considerably less in the weekend the business cannot afford to lose.

Talk to an expert
  • Rehearsed more than onceThe first run finds problems. The second establishes that the fixes worked and gives you a reliable elapsed time.
  • Signed by the businessReconciliation is reviewed and signed by the data owner, including spot checks on records the users chose themselves.
  • Reversible until it is notA documented rollback position, agreed before the freeze window, with the point of no return stated plainly.

Scope and targets are set after discovery, against the org as it stands.

Questions we get asked

Before you send an RFP

How long will cutover take?

We will tell you after the rehearsal, because that is the only honest source for the number. Elapsed time depends on volume, attachment size and how much transformation sits in the path, and estimating it from record counts alone has a poor record.

Can we keep the old system running?

Yes, read-only, for as long as the business genuinely needs it. What we would push back on is leaving it writable, because two live systems diverge immediately and the migration you just paid for starts decaying the same week.

What about attachments and activity history?

In scope from the start. They are usually the largest volume and the first thing cut for time, and a migration that moved the records but not the correspondence is one users work around rather than adopt.

We are merging two Salesforce orgs. Is that the same thing?

It shares a method and little else. The technical work is comparable; the hard part is that both sides have a working data model and neither wants to be the one that changes. We surface that as an explicit decision early rather than letting it arrive at cutover.

Rehearse ittwice.

Send us the source system and we will profile it first. What that turns up usually changes the plan, which is the point of doing it before the plan is fixed.

Get in touch +91-120-4849200