The complete guide
Salesforce integration: the complete guide
Updated November 18, 2026 · Published by Danny Rathod, Founder, StackLift AI. Written with AI assistance; how we make content.
Short answer
A Salesforce integration connects Salesforce to your other systems so leads, accounts, opportunities, and orders stay in sync without re-keying. It usually means an authorized connected app, API calls for records, event or webhook triggers for changes, and a mapping of your custom fields. Effort depends mostly on how customized your org is and whether sync runs one way or both.
What does a Salesforce integration do?
Salesforce usually sits at the center of sales and customer data, so it rarely works alone. An integration moves records between Salesforce and the product, website, billing system, or ERP that needs them, so each team works from the same facts.
- Send website, app, or form sign-ups into Salesforce as leads, with the source attached.
- Create accounts and contacts when a customer signs up in your product.
- Push closed-won opportunities to billing or an ERP so invoicing starts on its own.
- Show Salesforce account and plan data inside your own app or admin panel.
- Write product usage back to Salesforce so sales and support see real activity.
- Feed Salesforce data into a warehouse or dashboard for reporting.
How does a Salesforce integration work?
Salesforce exposes several APIs. The REST API handles reads and writes of single records. The Bulk API is built for large asynchronous jobs such as backfills and nightly reconciliation. For near-real-time change feeds, Salesforce offers event options such as Platform Events and Change Data Capture, delivered through its Pub/Sub API.
Authentication uses OAuth 2.0 through a connected app (or the newer External Client App) that you create in your org. You choose the scopes, and the integration holds an access token and a refresh token. Server-to-server jobs often use a dedicated integration user with the narrowest permissions that work.
Salesforce provides sandbox orgs for testing, so you can build against a copy of your configuration without touching live records. Limits on API usage depend on your edition and licenses, so check the current figures in Salesforce's own documentation before you design around them. See the glossary entry on APIs for the basics.
What data moves, and in which direction?
Decide the direction of each object before you build. One-way flows are simple and easy to debug. Two-way flows need rules for who wins when both sides change.
| Data | Direction | How often |
|---|---|---|
| Leads and contacts from web forms or your app | Into Salesforce | Real time |
| Accounts and plan details | Both ways, with one system as owner | Real time or every few minutes |
| Opportunities and closed-won status | Out of Salesforce | On change |
| Orders, invoices, payment status | Into Salesforce | On change |
| Product usage signals | Into Salesforce | Hourly or daily |
| Full history for reporting | Out of Salesforce | Nightly batch |
What drives the effort of a Salesforce integration?
A basic lead push is small. The effort grows with the following.
- Custom objects and fields: every custom field needs a mapping, a type check, and an owner.
- Validation rules, required fields, and automation in the org that can reject or change what you write.
- Data volume and the first backfill of existing records.
- Two-way sync, which needs conflict rules and loop prevention.
- Duplicate handling for leads, contacts, and accounts.
- Error handling, retries, and a queue so a failed call is not lost.
- Permissions, so the integration user sees only what it needs.
- Number of environments: sandbox, staging, and production.
The estimator adds a typical effort for Salesforce, shown above. Your own org's customization moves it up or down, which is why scope is reviewed before it is a commitment.
Common Salesforce integration pitfalls and how to avoid them
Most Salesforce integration trouble comes from the org, not the API.
- Hitting API usage limits: batch writes, use the Bulk API for volume, and cache reads.
- Duplicates: match on a stable key such as email or an external ID field before creating a record.
- Sync loops: tag records changed by the integration so it ignores its own updates.
- Silent rejections from validation rules: log every failed write with the reason.
- Over-permissioned integration users: give the narrowest profile that works.
- Expired or revoked tokens: monitor authentication failures and alert someone.
Should you build a custom Salesforce integration or use a connector?
A connector is quick for simple flows. A custom integration fits when logic, volume, or reliability matter. Zapier and Make both offer Salesforce connections.
| Question | Ready connector or iPaaS | Custom integration |
|---|---|---|
| Best for | Simple triggers, low volume, quick start | Custom objects, volume, business rules |
| Conflict and duplicate rules | Limited | Fully controlled |
| Error handling and retries | Basic | Queues, alerts, replay |
| Ongoing cost shape | Per task or per run | One-time build plus upkeep |
| Ownership of logic | Lives in the vendor's tool | Lives in your code |
Many teams start on a connector and move to custom code once volume or rules outgrow it. See the iPaaS glossary entry for how these tools work.
How do you test and launch a Salesforce integration?
Test against a safe copy of the system first, and move to live data only when the checks below pass.
- Build and test against a Salesforce sandbox org.
- Map every field and confirm types, picklist values, and required fields.
- Run the backfill on a copy and compare counts with the source.
- Test failures on purpose: bad data, expired token, a rejected write.
- Confirm the integration user's permissions and remove anything extra.
- Turn on logging and alerts before go-live.
- Release in a small slice, watch it, then widen.
- Keep a rollback step ready.
How does StackLift scope a Salesforce integration?
A Salesforce integration is scoped feature by feature in the free estimate, then reviewed by a senior architect before it becomes a commitment. Discovery covers which objects move, which direction, and how your org is customized. Pricing is by deliverable complexity, so you agree scope and budget before engineering begins. See the broader CRM and ERP cost guide.
Agents work on feature branches and never push to the default branch or touch production secrets. CI/CD, observability and rollback are in the plan from sprint one, so a bad release of the integration can be reversed. Start a free estimate and add Salesforce as a feature to see the cost, timeline and team.
Questions people ask
How long does a Salesforce integration take?
A simple one-way push can take days. A two-way sync with custom objects and a backfill takes longer. The typical effort above is a starting point, and a free estimate shows a timeline for your scope.
Do I need Salesforce Enterprise to integrate?
API access depends on your Salesforce edition and licenses. Confirm what your edition includes with Salesforce before you plan, because it affects both design and cost.
Can I integrate Salesforce with my own app?
Yes. Your app calls the Salesforce APIs after an OAuth authorization, and Salesforce can notify your app of changes through events. That is the usual pattern for a custom product.
Is two-way sync worth it?
Only when both systems must edit the same data. Otherwise pick one system as the owner of each object. It is simpler, safer, and cheaper to maintain.
Can AI be added to a Salesforce integration?
Yes. A language model can summarize account history or draft replies from Salesforce data. See LLM integration for how that is built safely.
Will the integration break when Salesforce changes?
Salesforce versions its APIs, so working integrations keep working. Plan a periodic check and monitoring so you catch a retired feature or a changed field early.