At a glance
Estimate integration costs across discovery, build, migration, operation and change, then add internal effort and recurring maintenance. A connector fee is only a small part of the bill.
The software has a native integration, so the project should be simple. Then someone discovers that the fields do not match, historical records need cleaning and nobody owns the errors that appear after launch.
Marketing integration costs include much more than access to a connector. The real bill covers making the data mean the same thing in both systems and keeping the relationship working as the business changes.
Count the work, not just the connection
MuleSoft’s 2025 connectivity research, reported by Salesforce, found that surveyed organisations used an average of 897 applications. That is an enterprise IT finding, not a typical marketing department’s stack size. It nevertheless illustrates why the number of tools alone tells only part of the story.
Each important connection creates decisions about identity, fields, timing, permissions and failure handling. Two systems can exchange data successfully while disagreeing about what it represents.
Break the estimate into five parts
| Cost area | What to include |
|---|---|
| Discovery | Workflow design, field definitions and ownership |
| Build | Configuration, custom work and data mapping |
| Migration | Cleaning, deduplication and historical transfer |
| Operation | Monitoring, retries, support and usage charges |
| Change | New fields, product updates and process revisions |
Ask the supplier which parts are included and which are your responsibility. A fixed implementation price is useful only when the assumptions and exclusions are clear.
Calculate internal effort explicitly
If a project needs a marketing operations manager for 30 hours, a CRM administrator for 20 and an analyst for 15, record those 65 hours. They are work the business cannot spend elsewhere, even if no additional invoice arrives.
Add the recurring effort. Two hours a week checking and repairing a connection becomes roughly 104 hours over a full year. A cheap connector can become an expensive operating habit.
These are illustrative inputs, not benchmark costs. Build your estimate from the actual workflow and the people who will support it.
Ask what happens when the happy path fails
What happens when the destination is unavailable? How are duplicate records handled? Can the team retry an event without creating a second purchase or lead? Where can someone see that a record failed?
Define which system is authoritative for each important field. If both systems can overwrite the same value, decide how conflicts are resolved. Otherwise the integration can become a fast way to circulate bad information.
Use a data quality audit before moving a large historical dataset. Fixing duplicates after they have spread across several tools is usually a more awkward job.
Compare the cost of not integrating
Manual work has a cost too: repeated exports, delayed follow-up, inconsistent reports and mistakes during transfer. Estimate the current effort and the consequences of delay before deciding whether automation is worth it.
But challenge the workflow itself. If a report is never used, automating its data feed is not a saving. If two tools duplicate the same job, removing one may be better than connecting them.
Our martech stack audit is a useful starting point for that conversation. The best integration project may be the one you no longer need.
Give the connection an owner
Record who monitors it, who can change it and who pays for maintenance. Keep the field mapping and operating notes somewhere the next person can find.
Review actual cost after the first few months. Compare the promised reduction in manual work with what the team now does. Include the time spent checking and correcting the automation.
A worthwhile integration reduces friction in a useful process. A connector badge on a product page is only the invitation to investigate whether it can do that.
