When we audit how work moves through a business, we are not really looking at the tools. Each tool is usually fine. We are looking at the seams, because that is where jobs stall, and the stall is invisible in every individual system.
The anatomy of a dropped job
A quote is accepted in the CRM. To become a job it has to exist in the ops tool. There is no integration, so a person retypes it.
That person is out on Thursday. Nobody covers the retyping, because it was never written down as a task. It is something Marcus does.
Now the CRM says won. The ops tool has never heard of it. Both systems are internally consistent and both are confident. Nobody is wrong, and nothing is scheduled.
Three weeks later the customer calls to ask when someone is coming.
What is instructive is that no report would have caught this. CRM reporting shows a healthy pipeline. Ops reporting shows a full schedule. The failure is in the space between two dashboards, and nothing looks at that space.
Why the gaps form
Not incompetence. Growth.
At five people the gaps do not exist, because the same person does both sides and remembers. At fifteen you have split the roles but the handoff is still a habit rather than a step. At thirty the habit has broken and been patched with a group chat.
Every system you add creates new seams. Three tools is three pairs. Four is six. The tools get better and the operation gets more fragile, which is genuinely confusing to live through.
Finding your seams
Take your five most common job types and, for each, write down every point where information moves from one place to another. Include the informal ones. A text message counts. A shout across the office counts.
For each handoff, ask four questions:
- Who does it? If the answer is a name rather than a role, that is a risk with a holiday schedule.
- How do they know to? If the answer is that they just check, it will lapse.
- What happens if it does not? If nothing surfaces, you have found a place where work can disappear silently.
- How long is it usually late? Ask the person, not the system. They know.
Most businesses we do this with find between six and twelve handoffs, and two or three that everyone immediately recognises as the source of recurring pain.
Three ways to close a seam, in order of preference
Delete it
The best handoff is the one that does not need to happen. Sometimes the second system does not need the record at all, or the step exists because of an old constraint. Ask why before you build anything.
Make it automatic
If information genuinely has to move, move it without a person. This is where a proper integration earns its cost, and it is usually far less work than people assume because you are moving a handful of fields, not merging platforms.
Make the failure loud
When neither of the above is possible, at least make silence impossible. A daily check for records that are won in one system and absent in the other. It does not fix anything. It means a dropped job surfaces the next morning instead of when the customer calls.
If a step can fail without anyone finding out, it is not a process. It is a habit with good luck attached.
Where AI fits, and where it does not
It is good at the reconciliation: reading both sides, noticing what does not line up, drafting the summary of what looks stuck. Frequent, boring, and cheap to be wrong about, which is exactly the shape you want.
It is a poor choice for the transfer itself. Moving a record from one system to another should be deterministic. Same input, same output, every time. That is ordinary integration work and it should stay ordinary. You do not want anything probabilistic between your quote and your schedule.
Use it to watch the seams. Use plain code to close them.
