Pick the method from four inputs: the objects and fields you need, sync direction, data volume and frequency, and who maintains it. Digitalegy is a HubSpot partner. Use HubSpot Data Sync when a supported app covers your objects and a sync that is not instant is fast enough (HubSpot checks for changes in HubSpot every five minutes), a Marketplace app when its ready-made features do the job, and a custom API build or middleware otherwise.
Key takeaways
- Pick the least custom method that passes all four questions: Data Sync or a Marketplace app first, then middleware, then a custom API build.
- HubSpot Data Sync runs one-way or two-way syncs for supported apps, and HubSpot checks for changes in HubSpot every five minutes, so updates are not instant.
- Data Sync does not maintain custom object associations and cannot sync data for a sales process without deal or opportunity stages.
- A publicly distributed OAuth app installed from the HubSpot Marketplace is limited to 110 requests every 10 seconds for each HubSpot account that installs it, so plan high volume data around summaries, batches or caching.
- A custom build means owning API changes, such as HubSpot's end of support for v1 to v3 APIs with a September 2027 enforcement date.
- Decide the source of truth, direction, conflict rule and match key for every shared field before any sync runs, because a failed match can create duplicates.
What is the difference between a Marketplace app, HubSpot Data Sync and a custom API integration?
A Marketplace app is a ready-made integration you install, HubSpot Data Sync is HubSpot's built-in record sync for supported apps, and a custom API integration is code your team writes and owns. Middleware, or iPaaS (integration platform as a service), is a fourth option where you assemble integration flows on a third-party platform.
The four options differ in who writes the logic and who keeps it working.
- Marketplace app: a ready-made integration you install from the HubSpot App Marketplace. Its builder decides which objects, fields and directions it handles.
- HubSpot Data Sync: after you connect a supported app, HubSpot's guide to connecting and using data sync explains that you can configure a one-way or two-way sync that updates records in both systems automatically. Field mappings determine which data syncs, in which direction, and how conflicts are handled.
- Custom API integration: code your developers write against HubSpot's APIs, so you control every object, field and rule, and you own every fix.
- Middleware (iPaaS): a third-party platform where you assemble flows between HubSpot and other systems instead of writing all the code yourself.
On the review of Marketplace apps, HubSpot's page on installing apps from the HubSpot Marketplace states:
"All apps in the HubSpot Marketplace are reviewed and certified by HubSpot's team of developers."
The choices inside any app (which fields map, which system wins a conflict) are still yours to make and test. With an app or Data Sync, someone else wrote the logic and you configure it. With middleware or a custom build, you assemble or write the logic, so your team maintains it.
Which objects does HubSpot Data Sync support, and what does it not cover?
Data Sync keeps records in step with a supported app in one or two directions and can sync custom objects, but HubSpot documents clear gaps: custom object associations are not maintained, some dropdown fields map only one way, and deals need a stage-based sales process.
First confirm in your HubSpot account and the app's Marketplace listing that your tool is a supported data sync app, and confirm which HubSpot subscription the sync and any custom field mappings need before you plan around them. If you need a refresher on how HubSpot organizes customer, sales and marketing data, read An Overview of the HubSpot Data Model before you decide which objects to sync.
What Data Sync covers
- Custom objects: HubSpot's article on syncing custom objects with data sync apps gives the example of custom objects that track subscriptions, assets or locations, synced so updates are reflected in both systems.
- Field mappings: the guide to data sync field mappings explains that mappings determine which data syncs, in which direction, and how conflicts are handled.
What it does not cover
- When syncing with a custom object, associations won't be maintained, so links to companies or deals need another method.
- In some cases a dropdown select property can only be mapped one way, as text, to the other app's text field.
- HubSpot will not automatically update the properties when new dropdown options are added or a field type is changed.
- If a user exists in one app and not the other, data will not sync, so record owners must exist in both.
- If you have a non-standard sales process that doesn't rely on deal or opportunity stages, the data sync guide says you won't be able to sync data using Data Sync.
- Invoices must be manually associated to company records in HubSpot.
- Syncs with millions of records may take several days to complete.
How do the integration methods compare on cost, setup time, flexibility, maintenance and data volume?
In our assessment, Marketplace apps and Data Sync are the cheapest and fastest to set up but the least flexible, a custom API build is the most flexible and the most expensive to own, and middleware sits between them.
The table scores the four methods on six criteria. Linked cells come from HubSpot's public documentation; every other cell is our assessment.
| Criterion | Marketplace app | HubSpot Data Sync | Custom API integration | Middleware (iPaaS) |
|---|---|---|---|---|
| Cost (our assessment) | Low: no code; check the listing for any app fee | Low: configuration only; confirm the subscription your app needs | Highest: developer time to build and to keep running | Medium: a platform subscription plus time to build flows |
| Setup time | Short (our assessment): install, connect, configure | Short to configure; syncs with millions of records may take several days to complete | Longest (our assessment): design, build, test, document | Medium (our assessment): build and test flows |
| Flexibility (our assessment) | Limited to what the app's builder offers | Mappings set which data syncs, direction and conflict handling; some dropdown properties map one way only, as text | Highest: any object, field or logic the APIs expose | High, within the platform's HubSpot connector |
| Maintenance burden (our assessment) | Low: the builder maintains the code; you watch for disconnected apps | Low to medium: HubSpot will not automatically update properties when dropdown options are added or a field type changes | Highest: you own API changes, such as the September 2027 end of support for v1 to v3 APIs | Medium: you maintain the flows; ask the vendor how it maintains its connector |
| Data volume | For a publicly distributed OAuth app installed from the HubSpot Marketplace, 110 requests every 10 seconds for each HubSpot account that installs it | HubSpot checks for changes every five minutes | Bound by your account's API limits; for private or custom integrations, you can purchase a maximum of two API limit increases | Bound by the same HubSpot API limits on the calls it makes (our assessment) |
| Sync direction | Whatever the app supports | One-way or two-way | Any direction you code | Any direction your flows define |
Read it row by row. The row where your requirement is strictest usually decides the method, which is why the checklist below turns these rows into four questions.
What do HubSpot API rate limits and sync frequency mean for high volume or near real time data?
A publicly distributed OAuth app installed from the HubSpot Marketplace is limited to 110 requests every 10 seconds for each HubSpot account that installs it, and Data Sync checks for changes every five minutes, so high volume or near real time data has to be designed around those ceilings.
HubSpot's API usage guidelines and limits set these rules:
- For a publicly distributed OAuth app installed from the HubSpot Marketplace, each HubSpot account that installs the app is limited to 110 requests every 10 seconds.
- For your account's own API limits, which private or custom integrations use, you can purchase a maximum of two API limit increases.
- Error responses should not exceed 5% of your total daily requests.
- Data your site or app uses on each page load should be cached, not requested from the APIs every time.
- Webhook calls made via workflows do not count towards the API rate limit.
The connected apps guide adds that multiple private apps (apps built for your own account) share the same daily API limit pool rather than each receiving its own allowance.
The arithmetic (a Marketplace app example)
For a publicly distributed OAuth app installed from the HubSpot Marketplace, 110 requests every 10 seconds is 11 a second, 660 a minute and 39,600 an hour. Updating 50,000 records at one request each takes at least 50,000 divided by 660, about 76 minutes, before retries. An integration sending 20,000 requests a day should stay under 1,000 error responses.
Data Sync runs on a different clock: HubSpot checks if there are changes every five minutes, according to the data sync guide. For high volume event data, our recommendation is to send HubSpot summaries, such as last login date or weekly active users per account, rather than every event.
When is a custom API integration or middleware worth the build and maintenance cost?
Build custom when the data or logic you need is outside what any supported app maps, or you need control over freshness, batching or summaries, and a named person will own the code for years. Choose middleware when you need multi-step logic across systems but no one will own code.
Signs the packaged options have run out:
- No supported Data Sync app or Marketplace app exists for the system, as with an in-house database.
- You need associations on synced custom objects, which Data Sync does not maintain.
- Your logic spans systems, such as creating a record only when two other tools agree.
- You need control over batching, caching or summaries to fit the rate limits.
The cost continues after launch. HubSpot's changelog on legacy APIs and apps going unsupported announces the end of support for v1 to v3 APIs, legacy public apps and legacy private apps, all with a September 2027 enforcement date. Integrations using any of these have a 12-month window to migrate, and HubSpot will no longer release bug fixes, reliability improvements or security updates for these versions. A custom build carries that migration work; with middleware, ask the vendor when its HubSpot connector moves to current APIs.
For small internal builds, the same changelog says: "Service Keys are the recommended path for lightweight integrations, internal tools, and AI agent integrations, with no full OAuth backend required."
Our recommendation: if nobody can name who will fix the integration a year from now, choose middleware or a packaged app over custom code.
What should you decide before you connect anything?
Before any sync runs, decide the source of truth for each shared field, the sync direction, the conflict rule, the property used to match records, and a cleanup baseline that records how clean the data is today.
- Source of truth per field: one system owns each field. Two writers for one field is how good data gets overwritten.
- Sync direction: one-way for fields one system owns, two-way only where both teams edit the field.
- Conflict rule: decide which app wins. In Data Sync, note that "If there’s no field value in the default app, no data will be changed in the third-party app" (data sync guide).
- Match key: HubSpot's guide to matching records in data sync recommends reviewing both apps to decide which shared properties to use to pair records, because a failed match can cause a duplicate record.
- Cleanup baseline: count duplicates, blank owners and blank match keys in both systems before launch, so you can tell later whether the integration made things worse.
Record the decisions in a field decision record. Copy this example and replace the rows with your own fields:
| Field | Source of truth | Direction | Conflict rule | Match key |
|---|---|---|---|---|
| Company domain | HubSpot | Two-way | HubSpot wins | Domain (companies) |
| Billing address | Accounting tool | One-way into HubSpot | Accounting tool wins | Domain (companies) |
| Contact email | HubSpot | Two-way | HubSpot wins | Email (contacts) |
| Record owner | HubSpot | One-way out of HubSpot | HubSpot wins; owner must be a user in both apps | Not a match key |
| Lifecycle stage | HubSpot | Not synced | None | Not a match key |
If you are about to map legacy fields into HubSpot, How to Use Claude to Map Fields and Plan Data Cleanup Before a HubSpot Migration helps you draft a field map and ranked cleanup rules before any sync goes live.
Which integration method does the four-question rule pick for common stacks?
Answer four questions (objects and fields, direction, volume and frequency, owner) and pick the least custom method that passes all four: Data Sync or a Marketplace app first, then middleware, then a custom API build.
The four-question checklist
- Objects and fields: does a supported Data Sync app or a Marketplace app handle every object, field and association you need? Custom object associations or a sales process without deal stages fail Data Sync here.
- Sync direction: can the method sync each field in the direction and with the conflict rule your field decision record sets?
- Volume and frequency: is a five-minute check fresh enough, and, if you use a publicly distributed OAuth app installed from the HubSpot Marketplace, does peak volume fit within its 110 requests every 10 seconds?
- Owner: who maintains it? No developer rules out a custom build; no one to own flows rules out middleware.
Stack 1: CRM plus accounting tool
Objects: companies, contacts and invoices. Direction: billing fields one way into HubSpot, contact details two ways. Volume: a few thousand records; finance data that is not instant is fine. Owner: a marketing operations manager, no developer.
Result: Data Sync, if the accounting tool is a supported app. Plan one manual task: invoices must be manually associated to company records in HubSpot, so name who does it and how often. If the tool is not supported, look for its Marketplace app.
Stack 2: CRM plus support and billing
Objects: tickets, companies, and subscriptions in a custom object. Direction: ticket status two ways, subscriptions one way from billing. Volume: moderate. Owner: a RevOps admin who can maintain flows but not code.
Result: Data Sync for the support tool (if supported), middleware for billing. Data Sync can sync the subscription custom object, but associations won't be maintained, and a renewal report needs each subscription linked to its company. A middleware flow that writes the subscription and sets the association closes that gap without custom code.
Stack 3: CRM plus in-house product database
Objects: product usage per account, written to company properties. Direction: one way into HubSpot. Volume: millions of raw events a month, summarized to one update per company per day. Owner: the product engineering team.
Result: custom API integration. No ready-made app knows your schema, and engineering can own the code. If the integration runs as a private app, it draws on the account's daily API limit pool, which multiple private apps share, so check your account's limits before you size the nightly job. Do not build on v1 to v3 APIs or legacy private apps, which reach end of support in September 2027, and evaluate Service Keys, which HubSpot names as the recommended path for internal tools.
If your team wants help scoping and building HubSpot integrations, start with our HubSpot implementation service.
What risks come with each integration method?
Every method can create duplicates, overwrite good data or break after a change; what differs is where the logic lives and who fixes it, which gets hardest when the person who built a custom integration leaves.
- Duplicate records (all methods): HubSpot's guide to matching records warns that matching on only the selected property "can result in records not matching as expected", which can cause a duplicate record. With matching turned off, duplicates sync and no merging of records will occur.
- Overwritten fields (two-way syncs): when both systems can write a field and the conflict rule is wrong, good data is replaced. Data Sync mappings set how conflicts are handled; in custom code or middleware, your logic does.
- Broken syncs after changes: Data Sync does not pick up new dropdown options or field type changes automatically, and unmigrated custom code on v1 to v3 APIs loses bug fixes and security updates after the September 2027 enforcement date.
- Shared limits: private apps in one account share one daily API limit pool, so another team's new app can eat into your allowance.
- Ownership (custom and middleware): when the builder leaves, undocumented logic becomes a black box. Our recommendation: keep a one-page runbook with the field decision record, the credentials owner, who gets error alerts and how to pause the sync.
If duplicate or incomplete records after connecting systems are your main worry, read How to Improve Data Quality in RevOps with HubSpot's AI Tools for deduplication and enrichment with HubSpot's AI tools.
How do you test and monitor a HubSpot integration after launch?
Test on a small, known set of records before the full sync, then review HubSpot's connected apps insights on a fixed schedule for errors, disconnected apps and unexpected creates, updates or deletes.
Before launch
- Pick about 50 test records that include the hard cases: missing owners, unusual dropdown values, near-duplicates and blank match keys.
- Confirm every record owner exists as a user in both apps, because data will not sync for a user who exists in only one (field mappings guide).
- Run the sync, then compare record counts and key fields against the source.
- Edit the same field in both systems and check that the documented winner wins.
- For large syncs, plan the calendar: syncs with millions of records may take several days to complete.
After launch
- HubSpot's guide to managing connected apps describes insights that "help you monitor app behavior, identify and resolve errors, and prevent issues like expired connections or reaching API call limits." Use them to spot disconnected apps and see how each app creates, updates or deletes CRM records.
- Keep error responses under 5% of total daily requests, the ceiling in HubSpot's API usage guidelines.
- Recheck mappings after any new dropdown option or field type change.
- Name one owner who reviews all of this weekly and before each quarter-end report.
If your integration problem is really a process and reporting problem across teams, look at our revenue operations service.
Frequently asked questions
Is HubSpot Data Sync real time?
No. HubSpot checks for changes every five minutes, and the data sync guide notes that syncs with millions of records may take several days to complete. If sales needs fresher data, a custom build or middleware can push changes when they happen, within HubSpot's API limits.
Can one stack use more than one integration method?
Yes. Apply the four questions per system, not per stack; Stack 2 above combines Data Sync and middleware. Give every field one writer so the methods never overwrite each other.
Who should own an integration if we have no developer?
Pick Data Sync or a Marketplace app first, or middleware if someone such as a RevOps admin can maintain flows, and name that person as owner of the runbook and the weekly connected apps review. Skip a custom build, because its owner has to handle API changes such as the September 2027 end of support for v1 to v3 APIs.
Can I buy more API capacity?
Within a cap: for your account's own API limits, which private or custom integrations use, you can purchase a maximum of two API limit increases. Reduce calls first with caching and daily summaries, then buy capacity if your private or custom integrations still exceed the limit.
Should we clean data before or after the first sync?
Before: count and fix duplicates, blank owners and blank match keys in both systems, and record the counts as your baseline. A failed match can cause a duplicate record, and data will not sync for an owner who is a user in only one app, so cleaning first keeps those problems out of the sync.
Why would records fail to sync between HubSpot and another app?
Causes HubSpot documents include: if a user exists in one app and not the other, data will not sync, and a sales process that doesn't rely on deal or opportunity stages can't sync through Data Sync. Check owners and pipelines first.
Sources
- Connect and use HubSpot data sync, HubSpot Knowledge Base (October 2, 2026)
- Understand your data sync field mappings, HubSpot Knowledge Base (January 21, 2026)
- Install apps in the HubSpot Marketplace, HubSpot Knowledge Base (September 4, 2026)
- Sync custom objects between HubSpot and data sync apps, HubSpot Knowledge Base (June 29, 2026)
- Manage your connected apps, HubSpot Knowledge Base (October 6, 2026)
- Legacy APIs and Apps: What's Going Unsupported and When, HubSpot Developers (September 21, 2026)
- API usage guidelines and limits - HubSpot docs, HubSpot Developers (September 4, 2026)
- Match records in data sync, HubSpot Knowledge Base (April 17, 2026)
Want help putting this into practice?