TL;DR
- HubSpot native: A starting point for straightforward duplicates; access to particular duplicate-management features depends on your subscription.
- Insycle: Consider for configurable matching, survivor rules and a preview report before applying changes.
- Dedupely: Consider for dedicated duplicate searches and recurring merge rules; verify the objects and safeguards in your plan.
- Koalify: Consider for duplicate handling close to HubSpot record and workflow processes; verify support and limits for your account.
- Before any bulk action: Decide which record survives, how fields are selected and which associations must remain. HubSpot says merged records cannot be unmerged.
A duplicate in HubSpot is not always a second copy of the same person. Two people may share a support inbox; a parent company and a subsidiary may use the same domain. A merge rule that treats either situation as a match can quietly change ownership, reporting and sales history.
For a small, well-understood queue, start with HubSpot's own record review and merge tools. For more complicated matching and repeatable field rules, evaluate Insycle or Dedupely. Koalify is worth a look when your team wants duplicate handling within its HubSpot workflow. The right choice depends less on how many records a tool can merge than on whether it preserves the right record, values and associations for your business.
What should a safe HubSpot merge preserve?
A safe merge has two separate decisions: whether records describe the same entity and which information should remain. Matching identifies candidates; it does not establish that every candidate should be merged. Define the expected outcome before asking a tool to apply it.
For contacts, consider the person, current verified work email, consent and subscription status, owner, lifecycle stage and links to companies, deals and tickets. For companies, consider legal identity, domain, account owner, parent-child relationships and open opportunities. If the same domain serves several legal entities, it is a useful signal, not proof they are the same account.
Specify a survivor rule for choosing the primary record, followed by field rules for values that should come from another record. The oldest record might have the desired identity and associations but an outdated title; the most recently updated record may contain a current title and an incorrect owner. A single “latest record wins” rule does not settle both questions.
HubSpot states that the remaining record combines activities, associations and most property values, but details vary by object and property. Its merge documentation explains that merged records cannot be unmerged. Creating a replacement contact or company is not the same as restoring the original record and all its relationships. Keep an export and a recovery plan, and test the actual outcome in a safe environment.
Record IDs need particular care in connected systems. HubSpot says a merge normally creates a new record ID and retains the previous IDs in a merged-record property; its public beta for primary-ID preservation changes that behaviour for enrolled accounts. Check the actual portal and update any integrations that depend on IDs. The same documentation notes that a secondary primary-company association can remain while losing its Primary label, so inspect association labels as well as counts.
The impact extends beyond the visible record. HubSpot's merge guidance says the secondary record is removed from static segments and active segment membership can change. Outside the primary-ID-preservation beta, it says both records are unenrolled from active workflows at merge; in that beta, the primary remains enrolled while the secondary is removed. Check the portal's current behaviour and document any necessary follow-up before merging. A pre-merge export of properties is useful evidence, but it is not by itself a restoration of workflow state or associations.
When a merge changes a property value, HubSpot says its displayed timestamp becomes the merge time rather than the source value's original time. A later “most recently updated” rule can therefore favour an older fact merely because it was merged recently. Preserve the underlying evidence and apply a field-specific freshness rule before using that timestamp as a proxy for accuracy.
HubSpot also documents exceptions to the primary-value rule: for contacts, the most marketable marketing-contact status is retained, and the most recent legal-basis values are kept. A merge can therefore affect both marketing-contact counts and compliance workflows in ways a general survivor rule may not predict. Review subscription and consent evidence separately; marketing-contact status is not itself consent to a particular communication.
How do the four options compare?
| Option | Useful starting point | What to examine before buying or applying |
|---|---|---|
| HubSpot native | Individual review and straightforward records | Subscription and object eligibility; detection and bulk options; merge behaviour |
| Insycle | Configurable matching, survivor and field rules with preview | Objects, sync setup, report and chosen plan |
| Dedupely | Dedicated searches, merge rules and recurring cleanup | Matching logic, objects, preview, audit detail and pricing |
| Koalify | HubSpot-centred handling and workflow actions | Permissions, object and workflow coverage, Salesforce restrictions and plan |
This is a shortlist for evaluation, not a claim that each tool has the same controls. Confirm current pricing, record allowances, billing term and any additional HubSpot subscription requirements with each supplier. A preview is only as safe as the rules and data used to produce it.
1. When are HubSpot's native tools enough?
Start here when you have a manageable backlog, can review records individually and do not need a separate rules engine. HubSpot documents automatic matching based on contact email and company domain in certain creation scenarios, plus record-ID and unique-property options for imports. Its deduplication guide explains the specific cases and exceptions: do not assume every integration or manual creation path behaves identically.
The duplicate manager can surface potential contact and company duplicates. HubSpot says individual duplicate management requires a qualifying Professional or Enterprise subscription; bulk management and custom rules require Data Hub Professional or Enterprise. It documents different display limits by tier, so verify your portal and permissions. When reviewing a pair, choose the primary record and inspect the fields and relationships that matter. HubSpot also permits choosing property values during supported merges.
For audit planning, HubSpot says its duplicate-management audit log covers recent auto-merges and rejected suggestions but does not surface manually merged records. Keep an appropriate separate record of manual decisions if the team needs a complete review trail; confirm what your portal can export before the cleanup begins.
HubSpot's merge guide also describes an export with pre-merge properties for up to 90 days, available for merges performed in the Data Quality tool. That is useful evidence, but it is not a complete backup of every merge path or a guarantee that associations and workflow state can be restored. Agree what to retain and where before the retention window closes.
Native handling is less attractive when a large queue needs consistent field-level policies, a detailed pre-change report or routine cross-system coordination beyond the capabilities available in your edition. Count review time and the cost of a mistake before deciding that an included feature is necessarily the cheapest process.
2. When should you consider Insycle?
Insycle is a candidate when your matching and field selection rules need to be repeatable. Its documentation describes identifying duplicate groups, selecting a master record and setting field rules. It also describes a preview mode that produces a CSV report before an update mode applies changes. Review that report against known cases rather than treating a successful preview as proof that the policy is correct.
For example, a team might select the contact already linked to the active account as the primary, retain an approved account owner and take a verified current job title from a secondary record. The configuration must be designed for the actual data. Insycle documents its defaults and field-level overrides in its merge module guide; defaults are not a substitute for that design.
Ask for a quote for your record volume and objects, whether an audit export is included and how a Salesforce-connected portal changes the process. A scheduled rule is useful only after manual cases establish that its match and survivor logic work for low-risk records.
3. When should you consider Dedupely?
Dedupely focuses on finding and merging CRM duplicates. Its product information describes duplicate searches, merge rules and automatic merging. That makes it a candidate for a team that wants recurring cleanup instead of a broader data-operations platform.
Use your difficult pairs to evaluate it: can a reviewer see why records matched, inspect the values that would survive and exclude ambiguous shared inboxes? Ask how the app handles specific HubSpot objects, record limits, sync connections and audit exports in your account. The HubSpot marketplace listing is a starting point for checking current installation and plan details, not a substitute for a scoped trial.
If billing depends on account size or another tier, calculate cost on the relevant total rather than the number of duplicates you hope to merge. Confirm minimum term and implementation costs before comparing it with internal review time.
Dedupely's product page lists plans from $40 per month, or $32 per month paid annually ($384 per year), for up to 30,000 records in selected synced objects. Its stated undo feature applies to most merges within 30 days and adds records back to the CRM; it does not establish that HubSpot's native merge can be reversed or that every original association and workflow state is restored. Ask for the object-specific limitations and test recovery before relying on it. A 60,000-record scope exceeds the published starting tier, so request the applicable tier rather than doubling the entry price without confirmation.
4. When should you consider Koalify?
Koalify is worth evaluating if operations staff want duplicate handling in the context of HubSpot records and workflows. Its documentation describes CRM-card handling and workflow actions for bulk merging. Ask which objects and actions your subscription supports and who can trigger them.
The practical question is not just whether an action can be automated. Can a reviewer see its inputs, understand which record wins and pause it when a case is ambiguous? Can the team export an audit of intended and completed changes? Check whether the required HubSpot workflow features and Koalify plan are available before adding their combined cost to a comparison.
Koalify's pricing page bases its price on all records in the objects selected for deduplication, not on the number of merges. Its displayed 50,000-record example is $25 per month billed annually; the 60,000-record example below needs the actual tier or checkout price for both selected objects. If only contacts are selected, companies are excluded from that count and from the selected cleanup scope. Confirm renewal and the required HubSpot workflow subscription before budgeting.
HubSpot's own restrictions still matter. Koalify's troubleshooting guide describes a Salesforce-connected company-merge restriction. HubSpot's current merge documentation distinguishes the older Salesforce integration, where company merges are unavailable, from the newer version, where they are supported. Check your integration version and each tool's current behaviour before choosing a method.
For company records, Koalify documents a record-preserving merge, which updates the existing primary record and deletes the secondary rather than performing a native merge. Its documentation says leads associated only with the secondary are not preserved in that mode, and read-only analytics and system properties cannot be updated. Inventory those dependencies and test the proposed recovery process before permitting an irreversible bulk action.
What do 20 difficult duplicate pairs reveal?
Build a test set before you turn on any rule. The following 20 synthetic pairs are a design exercise, not tests we ran in these products. For each, record the proposed action, primary ID, expected field values and expected associations; compare those with each tool's preview and, if appropriate, a controlled test portal. “Hold” means do not automatically merge until someone resolves the identity.
| Pairs | Fixture and expected decision | Survivor and association check |
|---|---|---|
| 1–2 | Same person and verified email, with an older and newer title; repeat with the newer title unverified. Merge only with identity confirmed. | Keep approved primary; current verified title in pair 1, retain existing title in pair 2. Check company and open deal. |
| 3–4 | Same person appears with work and former work email; repeat with personal email. Confirm both belong to that person. | Keep current work email and intended owner; check activities and company links. Do not infer consent from identity. |
| 5–6 | Two people share info@; repeat with a shared support phone. Hold both cases. | No survivor; retain separate people and their ticket associations. |
| 7–8 | Same name at the same company, first two different employees then the same employee with a spelling variation. Hold pair 7; merge pair 8 after verification. | For pair 8 preserve the verified identity, owner and deal links; pair 7 stays separate. |
| 9–10 | Same contact, conflicting opt-out values; repeat with different regional consent records. Hold pending consent review. | Do not replace an opt-out with an opt-in through a general latest-value rule. Check subscriptions separately. |
| 11–12 | Same legal company under old and new names; repeat where a parent and subsidiary share a domain. Merge first only after verification; hold second. | Keep the approved account and owner; inspect open deals, contacts and hierarchy. |
| 13–14 | Same company imported twice with two domains; repeat with one domain shared by unrelated franchises. Verify the first, hold the second. | If merged, retain approved primary domain and appropriate alternate information; inspect location associations. |
| 15–16 | Same contact attached to different open deals; repeat with two separate people at one deal. Merge first if identity is certain; hold second. | Check both deal associations on the first, including labels where used; do not collapse the people in the second. |
| 17–18 | Duplicate contacts have different owners; repeat with one record owned by an inactive user. Confirm ownership policy first. | Keep the approved active owner, check queue assignment and preserve relevant activities. |
| 19–20 | One contact pair is linked to a Salesforce record on each side; repeat with only one side synced. Hold for integration-owner review. | Record the intended surviving ID in each system and verify sync, associations and deletion effects before action. |
A passing match test is not enough. For a confirmed pair, compare the predicted primary and properties with the expected answer, then inspect associations and activity after a controlled merge. For a hold case, the expected result is no automatic merge. Capture false positives and false negatives separately; one incorrect merge can cost more than several missed candidates.
What does cleanup cost for 60,000 records?
Suppose a portal contains 50,000 contacts and 10,000 companies. For planning only, assume 5% of each total represents excess duplicate records: 2,500 contacts and 500 companies. If every excess record belongs to a separate two-record group, there are 3,000 candidate pairs. Triples, false matches and duplicates spread across objects change that count. Measure a sample before using the estimate for a budget.
At an assumed five minutes per pair, individual review of 3,000 pairs takes 250 hours. At an illustrative loaded labour rate of $60 per hour, that is $15,000 before software, investigation and correction. A 20-pair sample at the same rate takes 100 minutes, or about $100 of labour. These are arithmetic examples, not supplier prices or a prediction of your team's speed.
For a possible alternative, assume a month of software costs $500 and a scoped process requires 30 hours of setup and review at $60 per hour. The modeled direct cost is $2,300; add subscription minimums, HubSpot edition costs, specialist support and post-merge remediation where relevant. Automation has value only if the approved review policy and the quality of resulting records are comparable. Include recurring work, not just the initial backlog, when making the purchase decision.
How do you stop duplicates returning?
Trace the creation source for a representative sample: imports, forms, enrichment tools, integrations and manual entry can each create a different pattern. Fix the source rule before increasing merge frequency. Require a durable identifier where one exists, check import mappings and make ownership of exceptions explicit. An email address or domain is not a universal identity key.
For connected CRMs, document the primary ID and merge direction in both systems. Coordinate with the integration owner before changing sync or deletion settings. Insycle describes one approach for its own active-sync process; it is not a universal instruction for every connector. Our CRM activity sync guide addresses broader ownership and reconciliation decisions.
Assign someone to monitor the number of new candidates, unresolved exceptions, match errors and failed syncs. Store the policy, owners and escalation path alongside the HubSpot GTM setup template. For a wider tool comparison, see CRM data cleaning tools; for recovery preparation, see CRM backup and recovery.
FAQ
Can you undo a HubSpot merge?
HubSpot says merged records cannot be unmerged. For contacts and companies it describes creating a new record with merged email addresses or domains, but that is not a complete restoration of the original record and relationships. Prepare an export, document original IDs and test recovery procedures before making changes.
Should you merge every contact with the same email?
No. An address can be shared by several people or a team, and a field may have been entered incorrectly. Treat the email as a signal, investigate ambiguous matches and avoid making a shared mailbox the identity of an individual without an explicit policy.
Does HubSpot automatically prevent all duplicates?
No. HubSpot documents automatic matching in particular contact and company creation scenarios, as well as exceptions and other ways to identify records. Different import and integration paths can behave differently. Review your actual sources and periodically measure new duplicates.
Is an external deduplication tool required?
Not necessarily. Native tools may be sufficient for a small or straightforward queue on an eligible plan. An external tool is easier to justify when it demonstrably reduces the time or risk of complex, repeated work. Compare the cost and the expected outcome on the same representative cases.





