Switching Agency Software Without Losing a Payout Month

Two tools that both hold creator revenue can look almost identical in a demo and still be completely different to move between. The reason has nothing to do with features: it is that half of what you see in an agency tool is stored somewhere, and the other half is worked out on the fly. The first half migrates. The second half stays behind, and takes your history with it if nobody exported it as a value before the account closed.

What actually carries over when you switch agency software?

Only what the old tool stored as a fact. Everything else is a calculation, and calculations do not travel: you take their results or you take nothing.

That single distinction sorts every field in the system into four categories, and it is worth doing that sort explicitly before a vendor call rather than after one.

Category	Typical fields	What to do about it
Carries cleanly	Creator accounts, monthly gross and net totals, closed payout amounts, fan identifiers, price lists	Export, import, reconcile once
Degrades in transit	Fan tags, notes, conversation history, timestamps and their timezone, sale-to-chatter attribution	Export early, expect to reduce it to a summary
Rebuilt by hand	Splits, commission bases, permissions, shift structure, voice guides	Re-key deliberately; never trust a mapping
Lost	Audit trail, who changed what and when, inbox read state, anything the old tool computed live	Screenshot or export as a fixed value before cancelling

The fourth row is the expensive one. Past commission is usually not a stored number at all. The old tool held a rate and a base and produced the figure when you looked. Cancel the account and every past statement becomes unreproducible, which matters the first time a creator queries a payout from before the switch. Export closed statements as documents, not as data.

What never migrates, whatever the vendor says?

The operational memory. Tools store transactions well and context badly, and context is what a chatter actually opens a thread with.

• Thread history in usable form. Even a complete text export loses the structure: who wrote which line, which message carried the sale, what was sent and never opened.
• Tag vocabulary. Two tools rarely mean the same thing by the same word, so tags import as strings and stop being a segmentation. The mapping is yours to redo, by hand.
• Attribution. The join between a sale and the person who made it is the first casualty of an export, and it is the join every commission calculation depends on.
• Anything a person knew. Why this creator has a different base, which fan must never be offered a discount, which chatter covers a gap on Sundays.

None of this is a reason not to switch. It is a reason to write the summary down before you go, while the old tool is still open: one line per active fan and one line per creator rule is a morning’s work, and it is what a chatter opens the first thread with on the new tool.

In what order do the steps have to run?

Rules before data, closed history before live data, people last. Eight steps, and the order is the part that fails rather than any individual step.

1. Export everything while the old account is fully live and paid. Open every file. An export you have not opened is not an export.
2. Freeze the closed months. Take past statements as documents, dated, in a folder nobody edits.
3. Configure the rules by hand in the new tool. Each creator’s split, each chatter’s base, the price grid. This is the step you deliberately do not automate.
4. Reconcile one closed, already-paid month. Match gross, net of the platform’s cut, and the amount actually sent to each creator, to the cent.
5. Load the live layer. Roster, active fans, tags reduced to the vocabulary you chose in step 3.
6. Run one full payout cycle in both tools. Compare before anyone is paid.
7. Cut over. Old tool to read-only on an announced date, new tool authoritative from a named cycle boundary.
8. Cancel the subscription and keep the export. Put the renewal date in your own calendar; unused subscriptions renew silently.

Step 4 is where migrations actually fail, and almost never on arithmetic. A split pointed at gross when the contract says net returns a plausible number, which is precisely why it survives the check. The vocabulary is set out in net revenue, and the same reconciliation seen from the creator’s side is in tracking creator revenue.

When in the year should you switch?

On a closed and already-paid cycle, in the flattest stretch of your calendar you can find. The calendar constraint is stronger than most agencies expect, because a migration consumes exactly the attention that a busy period also wants.

Window	Why it is tempting	Why it usually goes wrong
The day the contract renews	The licence saves money	Renewal dates rarely land on a closed cycle, and the deadline forces step 6
During or just after a promo push	Revenue is high, morale is good	New threads flood the inbox and the team has no spare hours for a second interface
Immediately after the holidays	The calendar looks empty	The reversal tail from the previous weeks is still landing, so closed months keep moving
A quiet month, mid-cycle	Nothing is competing for the team’s attention, and the last cycle is already reconciled	Nothing. This is the window
While onboarding a new creator	It feels efficient to do both at once	Two sets of unfamiliar rules configured by the same person in the same week

The underlying constraint is that step 4 needs a month whose numbers have stopped changing, and refunds and chargebacks keep rewriting recently closed months. Which parts of the year are structurally noisy for your roster is the subject of seasonality in creator revenue; the rule for this page is narrower: never reconcile against a month that is still moving.

What breaks in the first cycle on the new tool?

The same four things, in roughly the same order, on almost every migration.

• The base. Configured against gross where the contract says net, or the reverse. It shows as a small, constant proportional gap in the same direction every time, which is how you recognise it and why one month is enough to catch it.
• The exception creator. Everyone with standard terms imports fine; the one on different terms is the one the tool models badly, and she is usually not the smallest account.
• Permissions. Chatters get either too much or nothing, and the second one gets fixed fast while the first one does not get noticed at all.
• Reversals. A refund belonging to a pre-migration month arrives in a post-migration cycle and has nowhere to land, so it either disappears or is deducted twice.

Handle the last one by deciding, in writing, before cutover: prior-period reversals are recorded against the old frozen statement and deducted from the next new-tool payout as a named line. Agencies that leave this undecided end up arguing about it with a creator, which is a bad first experience of a new system, and the creator on the receiving end has no way of telling a configuration error from a change of terms.

How do you know the migration is finished?

When nobody maintains anything in the old system, and one named person can reproduce last month’s payout from the new one without help. Those are the only two tests worth applying, and both are answered by watching behaviour rather than by reading a checklist.

• Has anyone opened the old tool in a full cycle? If yes, the migration is not finished.
• Can a second person produce this month’s statements? If only one can, you replaced a tool and kept the single-person dependency.
• Did a creator get a statement she could read without a call?

What no migration changes is what actually moves revenue. Our 42,852 conversations put the optimum pitch after around ten exchanges, with roughly a third of conversion lost when it lands before the sixth message, and they put the 2am-6am window at the bottom of the day and the evening at the top. No tool on either side of a switch writes that message or staffs that hour. Choosing which tool to move to in the first place (and whether the honest alternative is still a spreadsheet) is handled in creator management software and spreadsheet vs agency software.