Data silos are separate pockets of information, held in different systems or spreadsheets, that don't share with each other. They're the reason your systems disagree: each one holds its own copy of the same information (customers, products, jobs, stock, prices) and those copies drift apart. Information is retyped instead of shared, systems update at different times, and codes don't match between them. The fix isn't usually replacing systems; it's agreeing which system owns each piece of information, connecting the others to it, and having differences flagged automatically. This article explains what data silos look like, the common causes and how to fix each one.
What data silos look like day to day
You rarely see a data silo directly. You see its effects:
- the same customer has a different address in the CRM, the job system and the accounts,
- someone has to ring the warehouse to find out what's really in stock,
- a report has to be rebuilt by hand from three exports before the monthly meeting,
- one person is the only one who knows how a key spreadsheet works.
Each system is doing its own job. The problem is the space between them, where information stops instead of flowing on to the next person who needs it.
Common causes of disagreement
Retyping
Every time information is typed into a second system, it can be typed differently: a slightly different customer name, a missed order line, a different price. This is double entry (the same information typed into two systems), and timesheets are one of the worst cases: the same data entered three times.
Different codes
The same job, product or customer has a different code in each system. Without a shared code, systems can't be matched reliably.
Different timing
The job system is updated daily, the accounts at month-end, the spreadsheet when someone remembers. At any given moment, they show different pictures.
Different rules
One system counts revenue when a job is invoiced, another when it's complete. One includes freight in cost, another doesn't. Both are "right", but they don't agree.
Workarounds
Spreadsheets that adjust or override system figures, kept by one person and not visible to anyone else.
A typical example
For example, if your job system shows a job at a healthy margin but the accounts show it losing money, the cause is usually one of the above. Supplier bills might be coded to the job in the accounts but never entered in the job system. A variation might be approved on site and added to the job system, but invoiced weeks later. Or labour might be costed at a standard rate in one system and at actual payroll cost in the other. Each explanation is reasonable on its own, but together they mean two people looking at the same job see two different answers, and neither fully trusts the other's number. If that's familiar, see when your systems give different numbers and when you can't see job profit until it's too late.
How this shows up in different industries
Data silos look different in every industry, but the pattern is the same: information that stops in one system before it reaches the next. Some short examples:
- Trades and field service: the tech marks the job complete on a phone, but hours, materials and variations reach the office days later, so the invoice goes out weeks after the work. See jobs finished but not invoiced.
- Labour hire: onboarding, tickets, timesheets, payroll and invoicing each live in a different place, so every new placement adds more rekeying in the office. See why the office grows with every placement.
- Manufacturing and fabrication: hours on the floor and material issued to the job never meet the original quote, so quote against actual is rarely compared. See why job cost is a guess until it's shipped.
- Grain, timber, hay and quarry: weighbridge tickets sit in the weighbridge software, contract tonnages in a spreadsheet and invoices in the accounts, and they drift apart load by load. See weighbridge tickets, contracts and invoices.
- Wholesale selling online: the online store, the stock system and the accounts each keep their own stock and order records, so a sale in one isn't always seen in the others. See online store, stock and accounts out of step.
- Any business with credit customers: accounts puts a customer on hold, but that flag never reaches sales, dispatch or the techs, who keep supplying them. See customer on hold across systems.
How to fix it
- Pick the figure that matters most, such as job profit, stock or revenue, and fix that first.
- Agree which system owns each piece of information, for example customers in the CRM, products in the stock system, invoices in the accounts. This is what people mean by a single source of truth (one place everyone takes the number from).
- Use one code everywhere for jobs, products and customers, so systems can be matched.
- Connect the systems so information flows from its owner instead of being retyped. See ending double data entry.
- Agree the rules for how figures are counted, and write them down.
- Flag differences automatically, so mismatches are caught the same day, not at month-end.
What it looks like when it's fixed
On Monday morning, the job system and the accounts show the same margin for last week's jobs, because supplier bills, labour and variations were entered once and flowed to both. A customer put on hold by accounts shows as on hold on the sales screen and the dispatch list within minutes. When a figure does disagree, an automatic check flags it that day, and someone fixes it while the details are still fresh.
When it's bigger than a quick fix
Steps 1, 2, 3 and 5 are mostly agreement and housekeeping you can do yourselves. Connecting systems and building automatic checks is where help is worth it, especially with several systems or entities. If your systems were added one at a time and never joined up, see disconnected business systems. For how we approach the fix, see one set of numbers, when the problem is between your systems and connecting your systems.
When the silos are joined up, you can get a near-real-time financial picture instead of waiting for month end. New to some of these terms? Our plain-English glossary explains them in a sentence or two each.
Frequently asked questions
What are data silos?
Pockets of information held in separate systems or spreadsheets that don't share with each other. They're why the same figure can look different depending on where you look.
What causes data silos?
Usually systems added one at a time without being connected, plus spreadsheets built to fill the gaps. Each new system gets its own copy of customers, products or jobs, and the copies drift apart through retyping, different codes and different timing.
Why don't my job system and accounts match?
Usually because of retyping, different codes, different timing or different rules for counting. Finding which one applies to each difference is the first step to fixing it.
Do I need to replace my systems to fix data silos?
Usually not. Agreeing which system owns each piece of information, using shared codes and connecting the systems fixes most disagreement without replacing anything.
How do I stop systems drifting apart again?
Connect them so information flows automatically from its owner, and set up automatic checks that flag differences as soon as they appear.
If your systems never quite agree, book a free initial chat. We'll help you find where they part ways and what it would take to fix it.