Salesforce is most valuable when it holds the whole customer picture: orders from the ERP, invoices from finance, usage from the product, tickets from support. Getting that data in and out reliably is the integration problem, and it is where many Salesforce projects quietly overspend.
The decision usually comes down to MuleSoft vs custom APIs: build your own point-to-point connections, or invest in an integration platform? This guide explains both, the middle-ground options and a practical framework for choosing.
What are your Salesforce integration options?
Native Salesforce integration capabilities
- Salesforce APIs: REST, SOAP, Bulk API 2.0 for large data volumes, Composite requests and the Pub/Sub API for event streams.
- Apex callouts to external REST or SOAP services, secured with Named Credentials.
- Flow HTTP callouts that let admins call simple REST APIs from Flow without code.
- Platform Events and Change Data Capture for event-driven integration.
- Salesforce Connect to view external data in Salesforce without copying it.
Custom-built integrations
Code written by your team that connects systems directly, whether in Apex, in a small service hosted on AWS, Azure or Google Cloud, or in scheduled scripts. You own the logic, hosting, monitoring and maintenance.
MuleSoft Anypoint Platform
Salesforce’s integration platform for designing, building, running and governing APIs and integrations across cloud and on-premises systems. It includes prebuilt connectors, a runtime, API management, security policies and monitoring.
Middle-ground tools
MuleSoft Composer for simpler, no-code flows between common applications, and other integration platforms that trade depth for speed of setup.
The six standard Salesforce integration patterns
Whichever tool you choose, most integrations fit one of these patterns described in Salesforce’s architecture guidance:
| Pattern | What happens | Example |
|---|---|---|
| Remote process invocation: request and reply | Salesforce calls another system and waits for the answer | Check real-time stock or credit limit while a rep builds a quote |
| Remote process invocation: fire and forget | Salesforce sends a message and continues | Send a closed-won order to the ERP for fulfilment |
| Batch data synchronisation | Data moves in scheduled bulk loads | Nightly sync of invoices and payment status |
| Remote call-in | Another system creates or updates Salesforce data | An e-commerce platform creates accounts and orders |
| UI update based on data changes | The Salesforce screen updates when data changes | An agent sees a live payment confirmation |
| Data virtualisation | External data is displayed without being stored in Salesforce | Viewing order history held in the ERP |
MuleSoft vs custom APIs: side-by-side comparison
| Factor | Custom APIs and point-to-point | MuleSoft |
|---|---|---|
| Upfront cost | Low: development time only | Higher: platform subscription plus implementation |
| Speed for the first integration | Fast for simple cases | Slower initially while foundations are set up |
| Speed for the tenth integration | Slows down as dependencies multiply | Faster, because system and process APIs are reused |
| Prebuilt connectors | You build each connection | Connectors for many common enterprise systems and protocols |
| Monitoring and alerting | You build it, often inconsistently | Centralised monitoring, logging and alerts |
| Security and governance | Depends on each developer | Consistent policies, API management and access control |
| On-premises and legacy systems | Possible but custom work each time | Designed for hybrid cloud and on-premises connectivity |
| Skills required | General developers | MuleSoft developers and an integration architect |
| Long-term maintenance | Grows with every connection | More predictable with a well-designed API layer |
When custom APIs are the right choice
- You have one to three integrations, and they are unlikely to multiply soon.
- Data volumes are modest and near-real-time is not critical.
- The connected systems have good, stable REST APIs.
- You have developers who will own and monitor the code long term.
- Budget is tight and the integrations are not reused by other projects.
A common example is connecting Salesforce to an accounting package to create invoices when an opportunity closes, using a Flow or Apex callout with Named Credentials and a simple error-logging object. Our walkthrough of connecting Salesforce with Twilio for SMS and WhatsApp shows this style of integration in practice.
When MuleSoft pays for itself
- Five or more systems exchange data, or several teams build integrations independently.
- The same data, such as customers, products or orders, is needed by Salesforce, a portal, a mobile app and analytics.
- You connect cloud applications with on-premises systems such as SAP, Oracle or legacy databases.
- Regulation or security policy demands consistent API security, auditing and monitoring.
- Integrations are business-critical and failures must be detected in minutes, not days.
- You are modernising legacy systems and want to hide their complexity behind stable APIs.
How API-led connectivity works
- System APIs expose a core system, such as SAP customer or order data, in a clean, secure way.
- Process APIs combine system APIs and apply business logic, such as "create order" across ERP, inventory and billing.
- Experience APIs shape data for a specific consumer, such as Salesforce, a customer portal or a mobile app.
The value appears on the second and third project: a new customer portal reuses the same process API Salesforce already uses, instead of building new connections to every back-end system.
A decision framework you can apply today
Score each statement from 0 (not true) to 2 (very true):
- We will connect more than five systems within two years.
- More than one application or channel needs the same core data.
- At least one important system is on-premises or legacy.
- Integration failures have a direct revenue, customer or compliance impact.
- We need central monitoring and consistent security across integrations.
- Several teams or partners will build integrations.
0–4: native Salesforce tools and well-built custom integrations are likely enough. 5–8: start with native tools, but design for a future integration layer and consider a lighter platform. 9–12: an integration platform such as MuleSoft is likely to reduce total cost and risk.
What does each option really cost?
Compare three-year total cost of ownership, not the first invoice:
- Custom: build effort, hosting, monitoring you have to create, maintenance for every API change in connected systems, and the cost of outages that nobody noticed.
- MuleSoft: platform subscription, which Salesforce packages based on usage and capacity and has changed over time, so request a current quote; implementation of the first APIs; and skilled MuleSoft resources.
Integration effort is a major line in any Salesforce implementation budget, so make this decision during discovery rather than mid-build.
Integration best practices, whatever tool you choose
- Use external IDs and upserts so retries never create duplicates.
- Design for failure: timeouts, retries with backoff and a queue or log for messages that still fail.
- Monitor actively: alerts that reach a person, not error emails to a shared inbox.
- Secure credentials with Named Credentials and OAuth, and give integration users least privilege.
- Respect limits: use Bulk API 2.0 for large volumes and keep an eye on daily API consumption.
- Version your APIs and document contracts so changes do not break consumers.
- Test with bad data and production-like volumes, not just the happy path.
Two example architectures
Growing business: Salesforce, accounting and e-commerce
A Flow triggers an Apex callout to create invoices in the accounting system when opportunities close; the e-commerce platform calls the Salesforce REST API to upsert accounts and orders using external IDs; an integration log object and a dashboard surface failures. Simple, affordable and owned by a small team.
Enterprise: Salesforce, SAP, customer portal and data platform
MuleSoft system APIs expose SAP customers, products and orders; process APIs handle order creation and credit checks; experience APIs serve Salesforce, the customer portal and a mobile app; Change Data Capture streams Salesforce changes to the data platform. Monitoring and security policies are managed centrally.
Plan your Salesforce integrations with Groviya
Groviya designs and builds Salesforce integrations with native tools, custom APIs and MuleSoft, and helps you choose the approach that fits your roadmap rather than the one with the biggest licence. See our application development and integration services and Salesforce services, or talk to an integration architect.
