Your enterprise architect watches the procurement team accept a three-year SaaS agreement that lists integration as a one-time setup line item. Any follow-up work to connect the platform to existing inventory or finance tables triggers extra daily rates and change-control gates.
The vendor's implementation partner completes the initial data feed in four weeks and departs. Three months later the first batch of order records arrives with mismatched unit-of-measure codes, yet the contract treats the fix as new scope rather than part of the original delivery.
Finance now tracks every integration hour as an unplanned spend item. The architecture team therefore delays deeper mapping work until the next budget cycle, leaving the platform running on partial data and manual exports that nobody records as operational cost.
Consultants who negotiated the original deal reappear for each variation. Their incentive sits in extending the engagement, not in handing over the mapping logic so internal staff can maintain it after the first round of corrections.
The procurement policy still measures success by the date the licence invoice clears and the vendor signs off the statement of work. No metric tracks how many downstream systems remain unlinked or how many staff hours now sit outside the platform.
Next quarter the same team will approve another tool. The pattern repeats because the contract language never prices the cost of learning the integration once the first real dataset arrives.
Shift the commercial model so integration discovery work is funded inside the base agreement for the first twelve months. Otherwise the organisation keeps buying clean-looking platforms that stay expensive to actually connect.