Simple work takes three systems and a spreadsheet.
The friction is obvious from the inside. What it does not tell you is whether the answer is integration, configuration, modernization, replacement, or a data layer underneath all of it.
What this looks like from the inside.
- 01
A routine task touches three systems, and one of them is a spreadsheet somebody maintains personally.
- 02
The same record is entered more than once, and the versions have started to disagree.
- 03
People have become the integration layer, moving information between tools that were never introduced to each other.
- 04
Month-end depends on a reconciliation ritual that only a few people know how to perform.
- 05
Nobody can give a straight answer about where a given piece of work currently stands.
- 06
The stack was accumulated over years of reasonable individual decisions. It was never designed.
Why the friction does not reveal the answer.
The symptom is identical across at least five very different causes. Systems that do not talk to each other produce the same daily experience as a platform that was configured for a business that has since changed, an aging system nobody dares alter, a workflow that was never designed end to end, or the absence of any agreed source of truth. Each of those calls for a different response, and several are expensive to get wrong. This is the condition where buying the obvious fix does the most damage: an integration layer over a process that should have been redesigned is a faster way to run the wrong workflow, and a replacement platform bought to solve a data problem inherits the data problem on day one.
What a responsible answer has to account for.
Before choosing a direction, these are the constraints that decide which directions are even available.
- 01
Which parts of the current stack are genuinely working and should be left alone.
- 02
Where the real source of truth is today, and who has authority to define a contested term.
- 03
What the workflow is actually supposed to accomplish, as distinct from what the software currently permits.
- 04
Which exceptions are rare edge cases and which are the business operating normally.
- 05
What the operation can absorb: how much change, at what pace, without stopping.
- 06
Which people are currently holding the process together, and what happens when the workaround is removed.
The classes of response.
Usually the answer is a combination, sequenced. What matters is knowing which one leads.
- 01
Integration
Right when the systems are individually sound and the seams between them are the problem. The workflow makes sense; the information simply does not travel. Wrong when it would automate a process that should not exist in its current form.
- 02
Configuration
Right when a platform can genuinely represent the work and was set up for an earlier version of the business. This is the cheapest good outcome available, and it is worth ruling in before anything is built.
- 03
Selective replacement
Right when one tool in the chain is the constraint and the rest are fine. Replacing the single component that cannot be made to fit is usually less disruptive than a platform migration, and far less than a rebuild.
- 04
Modernization
Right when the system does the correct thing but cannot be changed safely. The behavior is worth preserving; the foundation underneath it is what has to move.
- 05
Custom software
Right when the workflow is genuinely specific to how this business competes and no product represents it cleanly. Reserved for the part of the operation that is the business, not the parts around it.
- 06
A data layer
Right when the disagreement is about numbers rather than steps. If two systems are both defensible and neither is trusted, connecting them without settling definitions just distributes the argument faster.
What it looks like when the diagnosis lands.
A commercial construction firm was running field documentation on one platform and accounting on another, with the gap between them absorbed by people. The answer was not more integration between the two. It was one platform for the field, deployed across hundreds of devices, connected directly to the existing ERP so financial integrity stayed where it already worked. Annual operating cost fell by $2.5 million, and the reconciliation work disappeared rather than being automated.
Read the caseWhere this goes next.
The diagnosis decides which of these leads.
Most engagements here use more than one.