The Post-Close CRM Merge Playbook: Two Books to One Cleared Database
Run a post-close CRM merge in signed waves: suppression union first, provenance kept per contact, shared accounts owned by rule, and zero unexplained deltas.
Key takeaways
- Load the opt-out lists from both companies as one suppression union before anyone sends from either system, because CAN-SPAM requires opt-outs to be honoured within 10 business days.
- Under CAN-SPAM, opted-out addresses can only be transferred to a company hired to help you comply, so the acquired suppression file has to move as its own compliance artifact, kept apart from the marketing list.
- CASL treats the buyer of a business as having the seller's existing business relationship, but the 2-year and 6-month implied-consent windows keep running and the buyer now carries the burden of proving consent.
- HubSpot stamps merged property values with the merge date and keeps pre-merge exports for only 90 days, so copy consent date, basis, and source into fields of their own before any merge runs.
- Salesforce blocks activation of a matching rule when sets of 100 or more identical match keys add up to more than 2% of all keys, which two combined books with similar account names can easily trigger.
- A wave passes only when the target count equals the source count minus documented exclusions, contacts and opportunities per account reconcile, and the count of child records with a null parent is 0.
This playbook is for the recruiting or revenue operations lead who has to fold an acquired company's accounts and contacts into the surviving CRM after a close or a PE add-on, and who will be asked later how each record got there. It sets out a numbered wave plan with gates: consent classified on every contact, both suppression lists combined before anyone sends, shared customers matched in stated tiers and owned under a rule signed in advance, and a reconciliation for every wave that leaves no delta unexplained.
What does a post-close CRM merge have to protect?
A post-close CRM merge has to protect three things: the history on every record, single ownership of customers both companies already sold to, and every opt-out recorded by either company. Moving objects between systems is the easy part. The failures that land on operations happen in the records themselves.
Most consolidation advice treats this as a systems migration: pick the surviving CRM, map the fields, move the objects. That work matters, but it rarely goes wrong in ways that reach a regulator or a customer. These are the three failures operations owns:
- Shared customers matched wrongly. Two records for one customer end up with two owners, or two genuinely separate customers get merged into one, along with their contracts and history.
- Consent that does not survive the move. An opt-out recorded in the acquired system never reaches the system you send from. Or a merge overwrites the date the consent evidence depends on.
- History that cannot be traced. Records arrive without their source system, legacy ID, or original dates, so nobody can later show where a contact came from.
An acquired database brings its past with it. The Marriott and Starwood breach involved a database that was added to Marriott's IT estate when it acquired the Starwood business in 2016. Whatever was true of the acquired book before the close becomes your problem once it sits in your estate.
Out of scope here: choosing the surviving platform, and field mapping in general. Both are well covered elsewhere. This guide starts once you know which CRM survives.
Can you email the acquired company's contacts?
Only contact by contact. Whether you can email someone depends on the regime that covers them, whether they are a corporate or individual subscriber, which entity collected their details, and whether they opted out on either side. The acquired list does not inherit the acquirer's permissions, and the acquirer's list does not inherit the target's.
The three regimes treat a change of ownership differently:
| Regime | Opt-out deadline | Implied or soft basis window | Transfers on sale? |
|---|---|---|---|
| CAN-SPAM | 10 business days | n/a (opt-out regime) | Opted-out addresses only to a compliance provider |
| CASL | Not captured in sources | 2 yrs from purchase, 6 months from inquiry | EBR deemed to transfer (s.10(12)) |
| PECR | Not captured in sources | No fixed period | Soft opt-in does not extend to other group entities |
What each regime means for the merge:
- CAN-SPAM (US). The FTC says that once people have told you they don't want more messages from you, you can't sell or transfer their addresses, "even in the form of a mailing list". The one exception is transferring them to a company you've hired to help you comply. You must honour an opt-out within 10 business days, and your opt-out mechanism must keep working for at least 30 days after you send. I found no FTC guidance that deals specifically with asset sales or M&A, so treat the M&A position as not publicly established and ask counsel.
- CASL (Canada). Under section 10(12), the buyer of a business "is considered to have, in respect of that business, an existing business relationship" (EBR) with the seller's contacts. An EBR is the implied consent that comes from a purchase, contract, or inquiry. The clock does not reset at the close. Implied consent lasts up to 2 years, or 6 months for inquiries. For an ongoing subscription or account, the period starts on the day the relationship ends. The CRTC says the burden of proving consent "would transfer to the new owner".
- UK GDPR and PECR (UK). The ICO says that when an M&A deal means data moves to a different or additional controller, you must consider data sharing as part of due diligence. That includes the purposes the data was originally collected for, your lawful basis for sharing it, and whether either has changed. Soft opt-in is the PECR exception that allows marketing email to existing customers without fresh consent. It "doesn't apply if someone else obtains the contact details for you, even if it's another organisation within your own group structure." It also requires that you offered an opt-out both when you collected the details and in every message since. The ICO also says organisations "cannot infer consent just because consent was given to a similar organisation, or an organisation in the same group". The exception that matters most: for marketing email to corporate subscribers, the PECR rules on consent or soft opt-in don't apply.
The practical result is that subscriber type decides most of your UK exposure. Corporate contacts may be mailable. Individual subscribers who bought from the target probably are not mailable by the acquirer brand without fresh consent. Named counsel goes further: if you are acquiring an organisation rather than merging with it, you need to reconfirm consent where appropriate, because the change of ownership makes you the new data controller. Some advisers also include a phase to notify customers about the change in controller and any new processing purposes. No regulator has published a template for that notice, so draft it with counsel.
What a cleared contact record carries
- Suppression statusOpted out on either side? Then nothing below matters
- Regime and subscriber typeCAN-SPAM, CASL, or PECR; corporate or individual
- Consent basisExpress, implied or EBR, soft opt-in, or corporate subscriber
- Date and methodWhen and how consent was obtained, e.g. acquired through purchase of another business
- Source entity and systemWhich company and which CRM collected the details
The wave plan, step by step
The merge runs as eight gated steps. Freeze and profile, classify consent, build the suppression union, write provenance fields, dedupe and match, apply the owner rule, load in waves, and reconcile. No step starts until the one before it is signed off.
Post-close CRM merge in gated waves
- Freeze and profileCount every object in both systems before anything moves
- Consent and suppressionClassify each contact and union both opt-out lists
- Provenance fieldsCopy original dates, basis, and legacy IDs before any merge
- Dedupe and matchWithin each side first, then across sides in tiers
- Owner ruleApply the pre-signed rule to every shared account
- Load and reconcileAccounts, contacts, opportunities, activities, each signed off
Post-close CRM consolidation procedure
- Freeze and profile both systemsPause imports and bulk edits in both CRMs and record the count for every object in each source. Done means a complete source inventory with per-object record counts.
- Classify consent on every acquired contactRecord each contact's regime, subscriber type, and basis: express, implied or EBR, soft opt-in, or corporate subscriber. Done means basis, date, and method are recorded on every contact.
- Build and apply the suppression unionLoad both opt-out lists into one union and apply it to every system that can send. Done means the union is live everywhere within 10 business days and every change is logged with a timestamp.
- Write provenance fields before any mergeCopy original consent date, basis, method, source entity, and legacy ID into dedicated fields. Done means the fields are filled on both sides and a pre-merge export is stored outside the CRM.
- Dedupe within each side, then match acrossRun duplicate jobs inside each CRM, then match the two books in stated tiers. Done means exact-tier merges are logged and the fuzzy tier is queued for a named reviewer.
- Apply the signed shared-account owner ruleSales leadership signs the rule before the match list exists, and the steward applies it. Done means every shared account has one owner and records which clause decided it.
- Load accounts, then contacts, then opportunities, then activitiesLoad in dependency order, one object per wave. Done means each wave finishes with no load errors left unexplained.
- Reconcile and sign off each waveCheck record counts, relationship counts, field spot checks, and business totals against the source. Done means variance is within threshold, orphans are 0, exceptions are resolved, and sign-off is recorded.
Who owns each step, and the evidence that closes it:
| Step | Owner | Closing evidence |
|---|---|---|
| Freeze and profile | RevOps admin | Per-object counts, both sources |
| Consent and suppression | Privacy counsel, marketing ops | Basis, date, method on every contact |
| Dedupe and match | Data steward | Merge log plus review queue |
| Owner rule | Sales leadership | Signed rule, applied per account |
| Load and reconcile | Admin, QA | Signed wave sheet, 0 orphans |
How do you match shared customers across the two books?
Dedupe inside each CRM first, then match the two books against each other in two stated tiers. Exact matches merge automatically and get logged. Fuzzy matches go to a human, who checks the contracting entity before merging anything. If you match across sides before each side is clean, a duplicate inside the acquired book shows up as two separate cross-book pairs.
The match tiers
In Salesforce, match keys narrow the candidates "to the most likely duplicates before the rule applies the comprehensive matching equation". The standard account rule's keys combine company, website, city, and ZIP. Standard matching rules can't be edited, so build custom ones. A tiering that works:
- Accounts, exact tier: an exact Account Name rule set to Block.
- Accounts, fuzzy tier: a fuzzy Account Name rule set to Allow, with an alert.
- Contacts: Email exact OR (First + Last fuzzy AND Phone normalized).
What each tier proves: an exact email match proves the same mailbox, though a shared inbox can still sit behind it. An exact account name proves the same string, not the same legal entity. A franchise and its parent, or two brand subsidiaries, can share a name. A fuzzy match proves only that two records look alike. Salesforce Ben describes it as "a potential duplicate that could also genuinely be a separate account."
No public rule says which confidence score should merge automatically and which should go to review. Set your own threshold, write it into the runbook, and sample the auto-merges at sign-off to see whether it held.
Platform limits that bite on a combined book
| Limit | Value |
|---|---|
| Active matching rules per object | 5 |
| Matching rules per duplicate rule | 3 |
| Activation block threshold | >2% of keys in sets of 100+ |
| Skew threshold | 10,000 records on one account |
The activation block catches people out. Salesforce refuses to activate a matching rule when sets of 100 or more identical match keys add up to more than 2% of all keys. Two books full of similar names can cross that line when neither book did alone. The documented fix is to add Site or Billing Zip Code to the account matching rule. A second limit: fuzzy methods exist only for common standard fields, so custom fields match exactly. Normalise the acquired side's custom fields (case, punctuation, country codes) before you match on them.
Cross-book account match decisions
HubSpot's default survivorship keeps the primary record's property values. Legal basis is the exception: HubSpot keeps the most recent values from both records. That default helps, but it does not replace your own provenance fields. These are the fields I'd add on both sides before matching:
legacy_record_id Source system record ID, never overwritten consent_source_system acquirer_crm | acquired_crm consent_source_entity Legal entity that collected the details subscriber_type individual | corporate consent_regime CAN-SPAM | CASL | PECR | other consent_basis express | implied_ebr | soft_opt_in | corporate_subscriber | none consent_date_original Date consent was obtained, copied before any merge consent_method How obtained; for transferred CASL EBR: acquired through purchase of another business ebr_clock_start Purchase or inquiry date, or end date of an ongoing subscription suppressed true | false suppression_source acquirer | acquired | both suppression_logged_at Timestamp of the last suppression-list change
Create these on contacts in both systems, fill them from source exports, and lock them against merge overwrites.
The CRTC says consent records should include the electronic address, the date, and the method consent was received, with "acquired through purchase of another business" as its own example of a method. The template above covers all three.
Who should run the merge, and where to find them
You need five roles: a RevOps admin to freeze, profile, and load, a data steward to own matching, marketing ops to own the suppression union, privacy counsel to sign the consent classification, and a sales leader to sign the owner rule. If nobody on staff has merged two CRMs before, hire someone who has. Plenty of people list the skills, but the label alone tells you little.
In Refolk's index of professional profiles, the supply of people listing CRM migration skills looks like this:
| Market | Skill pair | Profiles |
|---|---|---|
| US | Salesforce + Data Migration | 852,006 |
| UK | Salesforce + Data Migration | 104,681 |
| US | HubSpot + Data Migration | 143,245 |
Two ratios follow from those counts. US profiles outnumber UK profiles 8.1x for Salesforce, and in the US, Salesforce outnumbers HubSpot 5.9x. The caveat matters more than the count, though. Matching on skills is broad, and the samples were dominated by executives who list the skills, not by people who did the work. The figures show how big the pool is, not how deep the expertise runs. What you actually want is someone who has merged two orgs after a deal, and a skills filter cannot find that.
This is where Refolk removes the friction. You describe the experience you need in plain English, and it searches public LinkedIn and Crunchbase records, the public GitHub graph, and the open web for people whose history matches, rather than people who ticked a skill box.
For Canadian contacts, the same approach finds privacy lawyers who advise on CASL consent in acquisitions. That classification is a decision for counsel, not something operations should settle on its own.
The shared-account owner rule
The owner rule is a short precedence order that decides who owns a customer both companies sold to. Sales leadership signs it before the match list exists. No public standard prescribes one, so the value comes from agreeing it in advance and applying it the same way every time, not from any particular clause.
Write it before matching for a simple reason. Once reps can see which named accounts are shared, every decision becomes a negotiation. A rule signed in advance turns those decisions into lookups. The common candidates are below. None of them is established publicly as best practice, so treat them as options to choose from:
- Open-opportunity precedence. Whoever has an open opportunity on the account keeps it, at least until the opportunity closes.
- Trailing-12-month activity. Whoever logged the most recent meaningful activity in the last 12 months owns the account.
- Acquirer wins. The surviving organisation's owner takes every shared account.
My recommendation is to use them in that order, as tie-breakers. Open opportunities come first because moving a live deal mid-cycle costs revenue. Recent activity comes second because it reflects a real relationship. Acquirer wins comes last as the default, so no account is ever left without a decision. Record on each account which clause decided it, so you can audit the rule later.
Then decide one harder question in writing. When a customer has separate contracts with each brand (the bottom-right quadrant of the matrix above), do you keep two accounts linked to a common parent, or merge them into one? Keeping brand-separate shared customers as two accounts is not established practice either way. What matters is that the steward knows the answer before the fuzzy queue reaches that customer.
Write the owner rule before anyone sees the match list, or every shared account becomes a negotiation.
Load order and wave reconciliation
Load parents before children, one object per wave, and don't start the next wave until the current one reconciles with zero unexplained deltas. A delta is any difference between source and target counts. It is explained only when the exclusion behind it is documented and signed.
Load order
One practitioner's order, which I use as the default:
- Users and record types
- Products and price books
- Accounts
- Contacts
- Opportunities and cases
- Opportunity line items
- Tasks, events, notes, and files
Sources disagree on leads. One puts leads after cases, another puts them before opportunities, and a third lists them after products. Pick a position, write it down, and keep it the same for every add-on. Before the accounts wave, check for data skew. A combined book can end up with more than 10,000 records looking up to the same account, which Salesforce Ben flags as a skew threshold worth catching before the load.
What passing a wave means
A wave passes when record counts, relationship counts, field spot checks, and business totals all reconcile against the source. Here is what each check proves:
- Record count. Source count equals target count minus documented exclusions. This proves nothing went missing without a reason. It lies when exclusions are recorded as a total instead of a list of record IDs.
- Relationship counts. Contacts per account and opportunities per account match the source. This proves the records attached to the right parents. It lies when a wrong merge moves children to a different but valid parent, so the totals still add up.
- Orphans. The count of child records with a null parent is 0.
- Field spot checks and business totals. Sampled fields match the source, and pipeline or revenue totals agree. This catches mapping errors that counts cannot see.
Wave: [number and object, e.g. Wave 3 - Contacts] Source system(s) and freeze date: Source count (acquirer): Source count (acquired): Records excluded (documented): Exclusion list file: Target count: Check: target = source - excluded [pass / fail] Contacts per account reconciled [pass / fail] Variance: Opportunities per account reconciled [pass / fail] Variance: Child records with null parent: [must be 0] Field spot checks (n sampled): Mismatches: Business totals compared: Variance: Suppression union re-run on merged records, incl. secondary emails [pass / fail] Unexplained deltas: [must be 0] Variance threshold agreed: Exceptions resolved: Signed (QA): Date: Signed (RevOps owner): Date:
One sheet per wave; attach the exclusion list as record IDs, not totals.
How much of the acquired book should you expect to drop?
No published consolidation reports what share of the acquired book was dropped, so I can't give you a benchmark. The closest proxy is a study of business cards.
Expect a meaningful part of the acquired contacts to be stale. Exclude them deliberately, with reasons, rather than loading them and cleaning up later. If a valuable contact has moved on, Refolk can find where they work now. But a re-found person carries none of the old consent. Treat them as a new prospect under your own rules, not as a migrated contact.
How this goes wrong: failure modes and false positives
Most merge failures look like success on the screen. The merged record shows as subscribed, the counts add up, or the rule activates. The checks below catch the failures that a quick look at the dashboard misses.
- A merge loses an opt-out. HubSpot opt-outs follow the email address. A community answer says that if you merge into the contact that unsubscribed from all, the primary email's status should persist. That is unverified, so test it with dummy contacts before the real merge. False positive: the merged record shows as subscribed when one of its source records had opted out. Check: re-run the suppression union against every email address on every merged record, secondary emails included.
- The consent date gets overwritten. The merge timestamp replaces the original date, and with it your provenance. Check: confirm
consent_date_originalwas filled before the merge, then compare it against a sample of the pre-merge export. - Same-name accounts block rule activation. Franchise and brand-subsidiary names stack up in shared match keys and cross the 2% threshold. Check: if activation fails, add Site or Billing Zip Code to the account matching rule rather than loosening the rule.
- Custom fields fail to fuzzy match. Fuzzy methods only exist for common standard fields, so "Acme Ltd" and "ACME LIMITED" in a custom field never match. Check: normalise the acquired side's custom fields before matching, and compare match rates before and after.
- A fuzzy match that is not a duplicate. Brand-separate contracts look like duplicates in the fuzzy tier. Check: the reviewer confirms the contracting entity on both records before merging, and the decision goes in the merge log.
- Soft opt-in assumed to carry over. A UK individual subscriber who bought from the target gets email from the acquirer brand. Soft opt-in doesn't extend to another entity, even inside the group. Check: filter on subscriber type, collecting entity, and basis before any acquirer-brand send to UK contacts.
- The opt-out sync gap. In a HubSpot and Salesforce setup, a contact that isn't syncing won't sync its email opt-out status from either system until it syncs for the first time. Check: count the records outside the sync inclusion segment, and apply the suppression union to them directly.
- Orphans. Child records load with no parent to attach to, usually because a wave ran out of order or a parent was excluded without its children. Check: the count of child records with a null parent is 0 before sign-off.
Two of these (1 and 7) fail silently in the direction that causes harm: someone who opted out gets mailed. Run both checks on every wave that touches contacts, not just the contacts wave, because later merges and syncs can undo earlier fixes.
Keeping the runbook ready for the next add-on
A PE platform that has done one add-on usually does another, so treat the first merge as the draft of a runbook. Keep the field template, the signed owner rule, the wave sheets, and your measured durations, and re-check every regulatory value at its source before you reuse it.
Things to keep after this merge:
- The provenance field set, already deployed, so the next acquired book loads into fields that exist.
- The signed owner rule, with a log of how many shared accounts each clause decided.
- Every wave sign-off sheet and its exclusion lists, stored outside the CRM.
- The suppression union and its timestamped change log, as an ongoing compliance artifact.
- Actual durations for each step, since no public source supplies them.
- Your matching thresholds, plus what the auto-merge sample at sign-off showed about them.
Regulatory values change, and some of them are adjusted over time. Rather than copying figures from this guide or any secondary source, check penalty amounts and deadlines against the FTC's CAN-SPAM compliance guide, the ICO's data-sharing and PECR guidance, and the CRTC's CASL guidance each time you run a merge. Also check your CRM vendor's help pages for current matching-rule limits.
Before you call the merge done
- Per-object record counts for both source systems are dated and saved.
- Every acquired contact has a regime, subscriber type, consent basis, date, and method recorded.
- The suppression union from both companies is live in every system that can send, with timestamped change logs.
- Provenance fields were filled before the first merge, and a pre-merge export is stored outside the CRM.
- Exact-tier merges are logged, and every fuzzy-tier pair has a recorded reviewer decision.
- Every shared account has one owner and names the clause of the signed owner rule that decided it.
- Each wave has a signed sheet showing target equals source minus documented exclusions.
- Child records with a null parent total 0 across all waves.
- The suppression union was re-run on merged records, secondary emails included, after the final wave.
- Records outside the CRM sync scope were counted and suppressed directly.
- Actual step durations and exclusion counts are written into the runbook for the next add-on.
Questions practitioners ask
Can we email the acquired company's contacts after the deal closes?
Only after you classify each contact, and only if that contact is not on either company's suppression list. Under CASL, the existing business relationship transfers to the buyer, but the consent clock keeps running. Under PECR, soft opt-in does not carry over to another entity, even one in the same group. Corporate subscribers fall outside the PECR consent rules. CAN-SPAM is an opt-out regime, so opt-outs from both sides have to be honoured. Have privacy counsel sign off the classification before the first send.
How do we combine suppression lists after a merger?
Take the union of both lists, meaning any address opted out on either side, and apply it to every sending system before anyone sends. Keep the acquired suppression file apart from marketing lists. CAN-SPAM allows opted-out addresses to be transferred only to a company hired to help you comply. Log every suppression change with a timestamp. Check merged records against the union on secondary emails too, not just the primary address.
Which CRM's record wins when a customer exists in both books?
No public standard says. HubSpot's default keeps the primary record's property values, except legal basis, where it keeps the most recent values from both records. Owner rules such as acquirer-wins, trailing-12-month activity, or open-opportunity precedence are not established publicly. Pick one, write it as a precedence order with tie-breakers, and get it signed by sales leadership before anyone sees the match list.
What order should objects load in during a CRM consolidation?
Load parents before children. One practitioner's order is users and record types, then products and price books, then accounts, then contacts, then opportunities and cases, then opportunity line items, then tasks, events, notes, and files. Sources disagree on where leads belong. Whatever you choose, write it into the runbook and load each object as its own wave, so you can reconcile each one.
What does a passing wave reconciliation look like?
The target count equals the source count minus excluded records, and every exclusion is documented. Contacts per account and opportunities per account match the source. Field spot checks and business totals agree. The count of child records with a null parent is 0. Any variance has a written reason and a named person who signed it. A delta nobody can explain fails the wave, however small it is.
Does CASL consent transfer when you buy a company?
The relationship transfers. The clock does not reset. CASL section 10(12) treats the buyer of a business as having the seller's existing business relationship with that person. Implied consent is valid for up to 2 years, or 6 months for inquiries. For ongoing subscriptions or accounts, the period starts when the relationship ends. The CRTC says the burden of proving consent passes to the new owner. Record the method as acquired through purchase of another business.
Try it on the search you came here for
Stop building boolean strings. Just describe the person.
Type one sentence. I plan the search, read GitHub, public LinkedIn and Crunchbase records, and the open web as it is right now, and hand back a ranked list with the reason next to every name.
01Describe them
One plain sentence. Role, city, stack, stage, whatever matters to you.
02I read the web live
GitHub, public LinkedIn and Crunchbase records, the open web. Not a database that went stale last quarter.
03You read the shortlist
Ranked, with the reasoning under every name. Open a profile, ask a follow-up, narrow it down.
- Staff backend engineers in NYC who shipped Rust in production
- Series A fintechs in SF under 50 people, growing headcount this year
- Maintainers of fast-growing Rust web frameworks on GitHub
- No boolean, no filters, no seat to buy. One box.
- Read at search time, so a profile updated yesterday counts today.
- Every step visible as it runs, every name with its reason.
500 free credits on sign-up. No card, no demo call. See real searches.