What a custom software integration actually costs — and why estimates vary so much
Published 24 March 2026Updated 19 August 20268 min read
Integration cost is driven by six variables: number of objects and custom fields, sync direction, existing data quality, authentication and network constraints, throughput, and audit requirements. A one-way sync of two or three objects is typically two to four weeks of engineering; a bidirectional sync with conflict resolution, backfill, and reconciliation is usually six to twelve.
The six variables that set the price
- Object and field count, including tenant-specific custom fields that need mapping.
- Direction: one-way is roughly half the effort of bidirectional.
- Data quality: duplicate and malformed records in the existing estate often consume more time than the integration itself.
- Auth and network: OAuth with rotation is routine; private-network reach, VPNs, and jump hosts add procurement time, not just engineering time.
- Throughput: real-time paths, bulk backfills, and rate-limit budgeting are separate pieces of work.
- Audit and compliance: field-level logging and replay change the design, not just the documentation.
Why estimates from different suppliers differ by 3x
Usually because they are quoting different scopes. A quote that omits backfill, reconciliation, conflict rules, and monitoring will always undercut one that includes them — and the shortfall reappears as change requests once the integration is live and drifting.
Ask any supplier three questions: what happens on duplicate delivery, what happens when both sides edit the same field, and how drift is detected. The answers separate an integration from a script.
Discovery pays for itself
A short paid discovery — API access confirmed, real record volumes counted, sample data profiled, ownership decided per field — converts most of the uncertainty into fixed scope before the build begins. It is the cheapest way to avoid the change-request spiral.
What should be in the contract
- Milestones tied to demonstrable behaviour, not to elapsed time.
- Explicit acceptance criteria including duplicate-delivery and conflict behaviour.
- Handover pack: mapping documentation, runbook, monitoring dashboards, credential ownership.
- IP transfer on payment, with code and infrastructure in accounts you control.
Frequently asked questions
How much does a custom CRM integration cost?
Most one-way integrations of a few objects land in the two-to-four-week engineering range, and bidirectional builds with conflict resolution and reconciliation in the six-to-twelve-week range. Data quality in the existing estate is the most common reason a project exceeds its estimate.
Why do integration projects overrun?
Almost always scope that was never priced: backfill, duplicate records, conflict rules, rate-limit handling, and monitoring. Naming those in discovery removes most overrun risk.
What should an integration deliverable include besides code?
Mapping documentation, a runbook, monitoring and drift dashboards, replay tooling, and credentials held in accounts you own. Without those, the integration is a dependency on whoever built it.
iPaaS or custom integration? A decision framework, not a preference
When Zapier, Make, Workato, or a vendor connector is the right answer, when custom engineering pays for itself, and the total-cost factors most teams miss.
Freshworks CRM integration: architecture patterns that hold up in production
A practical engineering guide to integrating Freshworks CRM with databases, products, billing, and warehouses: API surfaces, auth, rate limits, sync direction, conflict resolution, and audit.
ServiceNow integration patterns for engineering teams
Table API versus Scripted REST, incident bridging, CMDB alignment, SLA-safe state mapping, and loop suppression when connecting ServiceNow to engineering tooling.