An RFQ arrives in Outlook. Somebody opens the print, works up a price, and types it into the quoting tool. If the job is won, somebody types it again into the shop ERP — part number, quantity, due date, routing. The job runs. Hours come off the floor on a paper traveler or a spreadsheet, and somebody types those into the accounting package. A customer calls to ask where their parts are, and somebody opens three screens to find out.
Nobody sold you a bad product. The quoting tool is good at quoting. The ERP is good at running jobs. The accounting package has been fine for fifteen years.
The problem is that there are five of them, and you are the integration.
The cost does not appear on any invoice
This is why it survives. Every other cost in a shop has a line item. Tooling has a line item. The lease has a line item. Insurance, payroll, the maintenance contract on the Mazak — all of it shows up somewhere you can see it.
The integration cost shows up at nine at night, in the office, moving numbers between screens that cannot see each other. It shows up as the half hour a production manager spends assembling a status answer that the system should have been able to give in four seconds. It shows up as the quote that went out Thursday instead of Tuesday because the person who prices work was busy being the person who keys data.
None of that is on a bill, so none of it gets managed. A shop will run a six-week project to shave two minutes off a setup and never once measure the forty minutes a day its highest-paid person spends as a data bridge.
Four places the gaps actually bite
The same number exists twice, and the two disagree. The hour your operator clocks at the machine lives in one system. The cost that lands on the job lives in another. When they are two numbers in two databases, every margin conversation turns into an argument about whose number is right, and the argument is unresolvable because both are defensible. Somebody eventually decides which system to believe, and from then on the other one is quietly wrong.
Re-keying is where accuracy goes. Not because anyone is careless — because transcription has an error rate and volume is volume. A shop quoting forty RFQs a month, each touched three times, is making a hundred and twenty manual transfers a month. At even a very good error rate that is several wrong numbers a month, and you will not know which ones.
Nobody owns the seams. Each vendor supports its own box. When the quote total and the job cost do not reconcile, neither vendor is wrong and neither vendor will help. The seam is yours.
Lead time is a guess dressed as a commitment. Quoting cannot see the schedule, so the delivery date on the quote comes from memory and optimism. The schedule finds out about the job after the customer has already been promised a date.
"Just integrate them" is a real answer, and it has a real price
It is not that integrations do not work. Shops run them, and in this market the full quote-to-actuals loop does exist — it just runs through two vendors and a piece of middleware between them.
Count what it costs before you choose it. There is a build, and a bill for the connector, and a person who understands it. Then one of the two vendors ships an update, the field mapping drifts, and for a few weeks the numbers are subtly wrong in a way nobody notices until a job closes badly. The integration is a system too, and it needs an owner, and in a thirty-person shop that owner is the same person who was already the integration.
What changes when it is one record
Not "the systems talk to each other." One record, which is different.
An RFQ becomes a quote, becomes a job, becomes an invoice, and never gets typed twice — not because a connector copies it, but because there is only ever one of it. The hour the operator clocks at the machine is the same hour that costs the job and posts to the ledger. There is nothing to reconcile because nothing ever left.
That sounds like a small distinction and it is the whole thing. Reconciliation is not a feature you can add. It is work that exists only because the data was split in the first place.
How to tell which side of the line you are on
Three questions, and they are worth asking of whatever you run now, including if you have no intention of changing anything.
Does the hour an operator clocks and the cost on that job live in the same database? If no, you are buying a reconciliation job along with the software, and you are paying for it in somebody's evenings.
When a customer asks where their job is, how many screens does it take? One is a system. Three is a filing cabinet with a login.
When an estimate misses, can you see which line missed? Not that the job came in over — which line. Material, setup, run, outside service. If the answer is a total, the next quote will repeat the same mistake, because nothing told you where it came from. That comparison is the whole mechanism, and it is the one most stacks cannot run.
The part nobody says out loud
The reason shops tolerate this is not that they have not noticed. Every owner reading this noticed years ago.
It is that replacing five systems at once sounds like betting the business, and replacing one of them makes the problem worse — now there are five systems and a migration. So the patchwork persists, not because it is good, but because every individual step away from it looks irrational in isolation.
That is a fair reading of the risk. It is also the reason this does not fix itself.
KG Ai ERP runs quoting, production scheduling, shop floor data collection, live job costing, suppliers, quality, CRM and a general ledger on one record. It was built inside a working precision machining shop. Request a demo — thirty minutes, RFQ to invoice, on a shop set up like yours.