When your systems do not talk to each other
Decide which system is right when two disagree. Then connect them.
Systems that do not talk are rarely missing a connector. They are missing a decision about which one is right. Name one owner for each record, count the hours going into moving data by hand, connect only the handoffs where work waits for a person, and monitor every connection.
Is it a fit?
Owner
One system owns each record and the rest read from it. That decision comes first.
Handoff
Connect only where work stops and waits for a person, not everything to everything.
Alarm
A failed sync tells you. A silent one is worse than no sync at all.
Start here. You get back what the overlap between your tools is costing, and which way to go.
Price the overlapWork it out in a minute
Four answers. Prices what the overlap costs you, then says which way to go.
The plan
Do these five, in order
A time against each one and a way to tell it is finished.
Name the owner of each shared field
One system is right when two disagree.
1 hour · Done when every shared field has one owning system
Count the retyping in hours
Who moves what, how often, how long it takes.
2 hours · Done when each handoff has an hours-a-month number
Connect the heaviest handoff, not the systems
One handoff. The rest waits.
1 to 2 days · Done when nobody retypes that one any more
Assume it will break
An alert to a person, a log of every run, a written manual fallback.
half a day · Done when you break it on purpose and someone is told
Re-time it, then take the next handoff
Write down the hours freed.
1 hour · Done when the hours are recorded and handoff two is chosen
What you get
Name the owner first
One system owns each record and everything else reads from it.
Count the retyping in hours
How many times one customer or job is entered by hand, by whom, and how long it takes.
Connect handoffs, not systems
The places where work stops and waits for a person to move it.
Assume it will break
Every connection fails eventually.
The order, and why it is this order
Start with the decision, because everything after it depends on the answer. Two systems holding the same customer will eventually disagree, and at that moment somebody has to know which one to believe. Naming that system takes an afternoon. It is also the step almost everybody skips, because it means telling somebody their tool is now the secondary copy, and that is a conversation rather than a configuration.
Then connect, and only the handoffs that measurement pointed at. The temptation is to do all of it while somebody technical is in the room, and that is exactly how a three week job turns into a permanent one. Two connections, finished and monitored, beat nine half built ones, and the second should not start until the first has run untouched for a fortnight.
If you would rather not run this yourself, we do it end to end, and the calculator on that page turns your team size and system count into what the retyping is costing you a year before anybody commits to anything.
Questions we get
Do I need something like Zapier or Make?
Often, and not always. For a handful of connections between mainstream tools they are the right answer and the cheapest one.
What they cannot do is decide which system is right, and that is the part that blocks people. Buying one before making that decision gets you the same argument at a higher frequency.
What if two systems both need to change the same field?
Then one of them still owns it, and the other gets a way to request the change rather than make it.
Two way sync on the same field is the source of most integration horror stories. It is possible, it is expensive to get right, and in a small business it is almost never worth it compared with picking a direction.
How do I know which system should be the owner?
Watch what people already trust. If everyone quietly checks the spreadsheet before believing the CRM, the spreadsheet is already the source of truth.
The honest move is either to make that official or to fix whatever made the CRM untrusted. Declaring a winner that nobody believes in changes nothing except where the arguments happen.
Our data is already inconsistent. Do we clean it first?
Clean only what the connection touches, and only once the owner is named. Cleaning everything first is a project that never finishes.
Expect the exercise to surface real disagreements about what a field means, which customer counts as active, when a job is finished. Those are worth settling and they are not really data problems.
More in the guides and every answer in one place.
Most of this is doable in house.
Read next
Whether to replace or connect, how far a no code tool gets you, and where the numbers end up once it works.
Shaheer leads the work, with engineers, writers, filers and analysts behind him. C-suite operations for a San Francisco AI company, Six Sigma on the process side, Anthropic certified on the Model Context Protocol, ten years across eight industries. See what we have built