Conflicting Reports From Two Different Systems: Decision Tree

Why this matters

The moment two systems that are supposed to agree show different numbers for the same thing, whatever confidence you had in either one drops fast. It happens constantly in a small shop running more than one tool: the field-service platform says one revenue figure, the accounting software says another; the scheduling tool shows a different job count than the invoicing report; a spreadsheet someone maintains on the side disagrees with both. The instinct is to pick whichever number is more convenient and move on, which is exactly how a real problem stays hidden. Reconciling the two properly, once, is far cheaper than making decisions off a number you privately do not trust.

Start here: do not just average the two or pick the nicer one

Splitting the difference between two conflicting numbers, or defaulting to whichever one looks better, both guarantee you are wrong. Neither number was derived by averaging or by optimism, so neither fix addresses the actual cause. Resist the shortcut and work through the comparison properly, even though it takes longer up front.

Step 1: Confirm the two reports are actually measuring the same thing

The most common cause of a "conflict" is not a data error at all, it is two reports answering slightly different questions while using the same label. Before assuming either number is wrong, check:

• Different date ranges. One report might run calendar month, the other a rolling 30 days, or one counts by the date a job was scheduled and the other by the date it was invoiced. A one- or two-day boundary difference can shift a count meaningfully.
• Different inclusion rules. One system might count a cancelled job as still in the pool, the other might exclude it. One might include tax in a total, the other might not. One might count a job with multiple visits once, the other might count each visit.
• Different definitions of the same word. "Completed" in a scheduling tool might mean the appointment happened; "completed" in an accounting tool might mean the invoice was paid. Same word, genuinely different events.

If the definitions or ranges differ, there is no real conflict, there are two correct answers to two different questions. Document which definition each system uses so the next person does not rediscover the same confusion, and decide which definition is the one you actually want to track going forward.

If the definitions and ranges genuinely match and the numbers still disagree, move to the next step. Something is actually broken.

Step 2: Identify which system is closer to the source of truth for this data

Not all systems are equally authoritative for a given number. Work out which one is closest to where the underlying event actually happened.

• For money actually collected, the accounting or payment system is almost always the source of truth, because it reflects what actually cleared, not what was scheduled or invoiced.
• For jobs actually performed in the field, the field-service or scheduling system is usually closer to the truth, because it reflects what a technician did, not what got billed for it.
• For anything manually re-typed into a spreadsheet from either system, the spreadsheet is the least authoritative of the group by default, because every manual re-entry is another chance for drift, staleness, or a typo.

This does not mean the "closer to the truth" system is automatically correct, only that it is the better starting point for tracing the discrepancy.

Step 3: Trace the actual gap, one record at a time if needed

Pull the specific records behind both numbers for a narrow window (a single day or week) and compare them line by line rather than staring at the two totals. You are looking for one of a small number of patterns:

• A record present in one system and missing in the other. Something did not sync, or someone forgot to enter it in the second system at all.
• A record present in both but with a different value (different amount, different date, different status). Points to a timing gap, someone updated one system and not the other, or a manual override happened in only one place.
• A record duplicated in one system. Points to a double-entry or a sync that ran twice.

Once you find the actual pattern in a small sample, it almost always explains the larger discrepancy, and you rarely need to trace every single record to be confident in the cause.

Step 4: Fix the sync or the process, and pick one system as the reference going forward

• If the gap traces to a manual entry step that sometimes gets skipped, either automate the sync between the two systems, or designate one system as the master and have the other pull from it instead of both being entered independently.
• If the gap traces to a broken or partial integration, this is a technical fix, and until it is resolved, treat the receiving system's numbers as unreliable for anything you are making a decision on.
• Whatever you decide, name one system as the reference for this metric going forward and use it consistently in every review. Chasing whichever number matches your expectation that day is how this problem becomes permanent instead of a one-time fix.

The recap

Do not average or cherry-pick between two conflicting numbers. Check first whether the two reports are really answering different questions with the same label; if so, document the difference and move on. If the definitions genuinely match, work out which system sits closer to the source event, trace the gap on a small sample of real records, fix the sync or process behind it, and name one system as the ongoing reference for that number so this does not resurface next month.

References

• U.S. Small Business Administration (SBA), small business record-keeping and reconciliation practices
• Trade-standard practice for reconciling field-service and accounting data
• See related: A Metric Was Wrong Because of Bad Data Entry; The Report That Should Trigger an Immediate Conversation