Skip to main content

Why does my HubSpot to QuickBooks sync create duplicate customers and invoices?

Your sync creates duplicates because it doesn’t remember what it already created: it retries a create without an idempotency key, matches customers on email or name instead of a stored ID, or leaves half a record behind after a failure. Fixing that in an existing sync usually takes 10–30 hours, about $850–$2,550 at our $85 hourly rate.

The four causes, and how to tell which one you have

Most duplicate problems come from one of four causes. You can usually tell which one by looking at the duplicates themselves: when they were created, and what is different between the two copies.

Cause What the duplicates look like Fix
Retry with no idempotency key Identical copies created seconds or minutes apart Send a unique request ID on every create, and check a mapping table before creating
Matching on email or name Copies days apart, with a slightly different name or a changed email Store the QuickBooks ID on the HubSpot record and match on that first
Partial failure Customer exists but the invoice is missing, or the invoice was created twice when the run was restarted Make each step resumable and record progress per step
Child before parent Invoice attached to a new, bare customer record instead of the real one Queue the child record until its parent exists

Check the timestamps first. Copies seconds apart are almost always retries. Copies days apart are almost always a matching problem.

Retries without an idempotency key

HubSpot retries webhook deliveries that don’t get a quick success response, and no-code tools retry failed steps. If QuickBooks created the customer but the response timed out on the way back, the sync thinks the create failed and sends it again. QuickBooks receives two valid create requests and makes two customers.

The fix is an idempotency key: a unique ID for each logical operation, sent with the request so the receiving side treats a repeat as the same request. The QuickBooks Online API accepts a requestid parameter for this. Build it from something stable, such as the HubSpot deal ID plus the operation name (deal-48213-create-invoice), not a random value generated on each attempt.

Also keep your own mapping table: HubSpot object ID, QuickBooks object ID, and status. Before any create, the sync looks up the HubSpot ID. If there is a row, it updates instead of creating. This one table prevents most duplicates.

Matching on email or name instead of an ID

Many syncs use a "find or create" step that searches QuickBooks by email or display name. That fails in predictable ways:

  • A contact changes their email in HubSpot, the search finds nothing, and a second customer is created.
  • Several contacts at one company share a billing address like ap@company.com, so the search matches the wrong customer or several at once.
  • The name differs slightly ("Acme Inc" vs "Acme, Inc."), so the search misses. QuickBooks requires display names to be unique across customers, vendors and employees, so some tools add a suffix and create "Acme Inc (1)".
  • The QuickBooks customer is the company, but the sync is keyed on the HubSpot contact. Every new contact at that company creates another customer.

Match on IDs. When the sync first creates or links a customer, it writes the QuickBooks customer ID to a custom property on the HubSpot company, such as quickbooks_customer_id. Every later run matches on that property. Email and name matching are only a fallback for records that have never been linked, and an ambiguous match should go to a person to decide, not create a new record.

Partial failures and data that arrives early

A typical flow: a HubSpot deal moves to closed-won, the sync creates or finds the QuickBooks customer, creates a draft invoice with line items, then a Stripe payment webhook marks the invoice paid and updates the deal. That is four writes across three systems, and any one of them can fail.

Partial failures

If the customer write succeeds and the invoice write fails, a sync that reruns from the beginning will create the customer again unless it checks first. Record each step’s result in the mapping table (customer: done, invoice: failed) so a rerun picks up at the step that failed. QuickBooks Online limits API calls to about 500 per minute per company. A bulk backfill that hits that limit will fail partway, which is exactly when a restart creates duplicates.

Child before parent

Webhooks do not arrive in order. The Stripe payment can arrive before the invoice exists, or the invoice event can be processed before the customer. A sync that handles a missing parent by creating one will make a bare duplicate customer with no address or terms. Instead, put the event in a queue, retry it on a backoff (after 1 minute, then 5, then 30), and alert someone if the parent still hasn’t appeared after an hour.

Cleaning up the duplicates you already have

  1. Pause the sync. If you clean up while it is running, it will recreate what you remove.
  2. Export customers and invoices from QuickBooks and group them by HubSpot ID, then by email and normalized name.
  3. Merge duplicate customers in QuickBooks. You merge two customers by renaming one to exactly match the other’s display name, and QuickBooks moves the transactions across. Merges cannot be undone, so have your accountant sign off first.
  4. Void duplicate invoices instead of deleting them, especially if a payment was applied or the period has been reconciled.
  5. Write the surviving QuickBooks IDs back to HubSpot, then turn the sync back on with ID matching in place.

For a few hundred customers this is a day of careful work. It is the step people skip, and it is why the duplicates keep coming back.

Patch it, rebuild it, or leave it

If you see fewer than a handful of duplicates a month and someone already reconciles monthly, leave it. Merging five records by hand costs less than any fix.

If you’re on Zapier or Make and the cause is matching, you can often patch it there. Store the QuickBooks ID in a HubSpot property, add a filter that skips the create step when the property is filled, and turn off automatic replay for steps that create records. That handles one-way, low-volume syncs.

Rebuild in code when you need a mapping table, ordered retries, resumable steps or payment data from Stripe in the flow. No-code tools don’t do those well. We bill hourly at $85, and a full two-system integration is a fixed scope of $1,500 to $6,000 depending on field mapping and error handling. We built deduplication, incremental sync and rate-limit handling into Fundz’s HubSpot, Salesforce and Gmail connector framework, and the same patterns apply here.

Frequently asked questions

Why does Zapier create duplicate QuickBooks customers from HubSpot?
Usually the Zap uses a find-or-create step that searches by email or name, and the search misses when either one has changed or differs slightly. Retries and Zap replays can also run a create step twice. Storing the QuickBooks customer ID on the HubSpot record and skipping the create step when it is filled fixes most cases.

What is an idempotency key in a HubSpot to QuickBooks sync?
It is a unique ID for one logical operation, such as creating the invoice for a specific deal, sent with the request so a retry is recognized as the same request and not a new one. The QuickBooks Online API accepts a requestid parameter for this. The key must be built from stable data like the deal ID, not generated fresh on each attempt.

Should I match HubSpot contacts to QuickBooks customers by email?
Only as a fallback for records that have never been linked. Emails change, and shared billing addresses match several people at once. Match on a stored QuickBooks customer ID first, and send ambiguous email matches to a person to decide.

How do I merge duplicate customers in QuickBooks Online?
Rename one customer’s display name to exactly match the other’s, and QuickBooks offers to merge them, moving all transactions to the surviving record. Merges cannot be undone, so pause your sync and check with your accountant first. Afterwards, write the surviving customer ID back to HubSpot so the sync doesn’t recreate the duplicate.

How long does it take to fix a sync that creates duplicates?
In our experience, adding ID matching, idempotency keys and resumable steps to an existing sync takes 10 to 30 engineering hours, plus about a day to clean up a few hundred existing records. A full rebuild of a two-system integration is a fixed scope of $1,500 to $6,000 at Smplfy.

Related answers

Want us to find where your duplicates come from?

Send us a short description of your sync and we’ll reply with a free written diagnosis, or book a free 20-minute call. Have ready the tool you sync with, roughly how many customers and invoices you have, and a few example duplicates with their creation dates. Start a conversation.