A customer's billing address is one field, and in a company running a CRM, Tally, an ERP and a support tool it is usually stored four times. When the customer moves, one copy gets updated and three do not. Nobody records that as a cost, because it never appears as a line in a budget; it appears as a wrong invoice, a support call and an evening of reconciliation.
Where the cost shows up
Each department picked a good tool for its own work. Finance runs Tally, HR has its own platform, sales has a CRM, operations has an ERP, and support lives in a ticketing tool and WhatsApp. The cost sits in the gaps between them, and it takes five recognisable forms.
| Form | What it looks like day to day | Who carries it |
|---|---|---|
| Manual transfer | Copying values between sheets, emailing files, logging into four tools for one answer | Operations and accounts staff |
| Conflicting versions | Sales updates the CRM, billing invoices from older data | Finance, and the customer |
| Slow reporting | The monthly pack takes days to compile and arrives after the decision | Management |
| Repeated stories | The customer explains the same problem on every call | Support, and the customer |
| Workarounds | Scripts, macros and one person who keeps systems loosely in step | IT, and whoever wrote the script |
None of these is large on any single day. The time is spread so thinly across so many people that no manager ever sees it as a cost, which is why the arrangement is tolerated year after year.
How one stale field spreads
- 01Sales edits CRMNew address saved
- 02Billing keeps old dataNobody retypes it
- 03Invoice goes outTo the old address
- 04Support cannot see whyDifferent record again
One missed copy produces a billing error and a support call.
Nobody in this chain made a mistake. The sales executive updated the CRM as instructed. The billing clerk raised the invoice from the data in front of them. The design of the work guarantees the gap, so training people to be more careful does not close it, and neither does a circular asking everyone to update all the systems.
One owner for every field
Integration replaces the copy step with a defined path from the system that owns the data to the systems that need it. The first piece of work is deciding who owns what, field by field, and writing it down. One source of truth per field, and every other system reads from it.
| Field | Owner system | Systems that read it |
|---|---|---|
| Billing address | CRM | Tally, ERP, support tool |
| Credit limit | Tally | CRM, order entry |
| Order status | ERP | CRM, support tool, customer messages |
| Open complaints | Support tool | CRM |
Teams often disagree about this table, and the disagreement is useful. Sales may believe it owns the billing address while accounts insists on approving any change that affects GST invoices. Settle it before anything is connected, because a connection built on an unsettled owner will copy the argument into every system.
- Reports and screensread connected data, no compiling
- Reconciliation checkscompare owner and copies on a schedule
- Integration layersync, events, mapping, retries
- Owner systemsCRM, Tally, ERP, support tool
The reconciliation layer is the one most often left out, and the one that catches drift.
Once the owners are set, automated sync removes the manual entry and everyone sees the same current value. Reporting reads from the connected sources, so the monthly pack becomes a query rather than a compilation. Support, sales and marketing see the same customer history. The workarounds are replaced by connections that are documented and monitored, which means they survive the person who built them. Which shape that middle layer takes depends on the flow, and the options are compared in five enterprise integration patterns and when each one applies.
The cost hidden in workarounds
The IT cost deserves spelling out, because it is the least visible of the lot. Workarounds accumulate one at a time. A script exports one table from Tally. A macro reformats it for the CRM import. A person in the Noida office knows to run both on the first of the month, in that order, and to fix the date format by hand when it breaks.
Each piece was cheap when it was written and is expensive when it breaks, because only its author knows how it works. Over a couple of years the cost of maintaining that collection can exceed the cost of a proper connection. It never appears in a budget because it sits inside other people's jobs, and it disappears entirely from view until the author goes on leave in the week the auditors arrive.
Mistakes that cost the most
The comparison is usually done wrong from the start. Teams count only licence and hosting costs, so the hours spent on reconciliation never appear on the other side of the sheet. Others fix the report instead of the source, building a dashboard over mismatched data and then arguing about why each team's total is slightly different.
An export script becomes the integration by default, which works until the person who wrote it resigns. Connections get built with no reconciliation check, so both sides assume the other is correct and drift goes unseen for months. And the first project gets chosen by noise, starting with the system everyone complains about rather than the flow that causes the most rework.
Finding the cost in your own company
Ask each team for a list of the manual handoffs they perform in a normal week. For each one, write down:
- The two systems involved, and which of them should own the data.
- Who does the transfer, and how often.
- What goes wrong when it is late or incorrect.
- Whether anyone checks it afterwards, and against what.
Rank the list by how many people it affects and how visible the failure is to customers. The top entry is usually a better first integration than the one with the loudest sponsor.
Take a furniture maker that sells through dealers and its own showrooms. Its list might show that reconciling dealer payments against dispatches takes two people the first week of every month, and that nothing else comes close. That is the first project, even if the managing director's own complaint is about the CRM. An integration that removes one monthly reconciliation is easier to justify, and easier to finish, than a general plan to connect everything. Our notes on tools and approaches for a first integration pick up from there.
A good test of the result is whether a customer-facing question, such as the status of an order, can be answered from one screen without pinging another department on WhatsApp.
The handoff list costs a few conversations and nothing else. Make it before any vendor call, take it to that call, and use it to scope the first piece of enterprise application integration work, whoever does it.
