For retailers whose store managers do not believe the online numbers

Store Sales Attribution: Who Gets Credit When Channels Blur

A customer is served in your Geelong shop, is shown a colour that is not in stock, and the staff member orders it for them. The item ships from the Melbourne warehouse to the customer’s home. Who made that sale? Ask four people in the business and you will get four answers, and each of them affects a store ranking, a bonus and a manager’s willingness to keep helping customers who then buy online.

It has to be solved when the sale is captured rather than in a month end report, with one documented rule per order type, attribution that survives partial shipments and refunds, and a commission calculation people can reproduce. This page covers how to build that, and the specific ways it goes wrong.

Book an Attribution Review (03) 9003 0111

Why This Is Harder Than It Looks

4 claimants

Can each claim one online order

The fulfilling site, the collection store, the staff member who helped and the customer’s usual store

1 rule

Per order type, decided once and written down

Two rules in two reports is how finance and operations end up ranking stores differently

Ex GST

The base most commission schemes should use

Paying on GST inclusive values overstates the base by about ten percent of the ex GST amount

Same day

Visibility staff need before they believe the numbers

A month end spreadsheet nobody can reproduce loses that trust quickly

Four Decisions That Make Attribution Trusted

Staff stop trusting sales credit for predictable reasons. These four decisions address the most common ones.

Attribution is captured, not inferred

The only reliable moment to record who influenced a sale is when it happens. A staff identifier entered on an in store order, a clienteling link a staff member sent, the collection location chosen at checkout, the site that actually picked the parcel: all of these are facts at the time and guesses afterwards. Reconstructing attribution later from postcodes and timing produces a number that changes every time the logic is revisited, which is exactly why nobody trusts it. Capture it at source and store it on the order.

One rule per order type, chosen for the behaviour you want

Click and collect, ship from store, in store order for home delivery, phone order taken by a shop, online order in a store’s catchment: each needs its own rule, and the choice is commercial rather than technical. Crediting the collection store encourages stores to promote collection. Crediting the fulfilling store rewards the picking effort. Splitting rewards both and complicates everything. Decide deliberately, write it down with effective dates, and accept that changing the rule mid quarter changes people’s pay.

Once set, attribution is immutable and auditable

If a report rebuild can change who was credited with a sale from last March, then no store ranking is stable and every bonus conversation becomes an argument. Attribution should be written onto the order record when the order is created, kept with it through splits and partial shipments, and adjusted only by explicit, logged corrections with a reason and an approver. Amendments are legitimate. Silent retrospective recalculation is not.

Returns and refunds are part of the design, not an exception

A commission scheme without clawback pays on revenue that later walks back through the door, and one with badly designed clawback punishes staff for a decision a customer made two months later. Decide whether commission is earned at sale, at dispatch or after the change of mind period, decide how a refund in a later period is handled, and make it visible to staff at the time so they are not surprised in a pay cycle. Exchanges deserve their own rule, since treating one as a return plus a sale produces odd results.

The Six Parts of Attribution and Commission

The first three are about data. The last three are about trust, and pay, which is why the audit trail matters as much as the maths.

Facts, not guesses

Capture at the source

Staff identifiers captured on in store orders without slowing the transaction, collection location recorded at checkout, fulfilling site recorded at pick, clienteling links carrying an identifier through to the order, and phone orders tagged to the store that took them. The practical challenge is making capture fast enough that staff do it every time, because partial capture is worse than none: it produces a league table of who remembered.

Documented

The attribution rule set

One rule per order type, versioned with effective dates, owned by a named person, and published to the stores rather than kept in a finance spreadsheet. Includes the tie breakers: what happens when a clienteling link and a collection store disagree, or when an order is split across two sites. Published rules let stores plan around them instead of disputing the results.

Per line

Carrying attribution through the lifecycle

Orders split across locations, ship partially, get cancelled in part, get edited by a service agent, or get exchanged. Attribution has to travel with the lines rather than the header, so a split order credits correctly and a partial cancellation reduces the right store’s number. This is where first attempts at an attribution report often break without anyone noticing.

Net, not gross

Returns, refunds and clawback

Refunds matched back to the original attributed sale even when processed in a different store or a different period, with the clawback rule applied consistently. Returns accepted in a store that did not make the sale are a common and useful case: the returning store deserves recognition for handling it while the original credit is adjusted. Both facts can be recorded without arguing about one number.

Reproducible

Commission calculation

Thresholds, tiers, team and individual splits, an ex GST base, and a period that aligns with your pay cycle rather than your accounting calendar. Every calculation keeps its inputs so a query can be answered by showing the workings. Commission forms part of pay, so the treatment of superannuation and any minimum award entitlements needs your payroll adviser’s confirmation rather than an assumption in a formula.

Trusted

Payroll handoff and staff visibility

An approved output that flows to payroll on a schedule with an audit trail, and a same day view for staff and managers showing what has been credited and what is pending. A dispute path with an owner, because disputes will happen and slow answers undermine the scheme.

What Changes in Practice

Task	Traditional	Attributed Properly	Notes
In store order, shipped from the warehouse	Counts as an online sale	Credited to the staff member	The single change that most affects whether staff keep helping customers who then buy online.
Click and collect order	Credited to the web channel only	Credited per the published rule	Either the collection store or a split. What matters is that the rule is known in advance.
Order split across two sites	Credited to the header location	Credited per line	Line level attribution is the difference between a report that survives and one that gets abandoned.
Refund in a different store	Reduces the wrong store	Matched to the original sale	Recognise the handling store separately rather than fighting over one number.
Exchange for a different size	Counted as a return and a sale	Handled by its own rule	Otherwise apparel commission numbers inflate in a way nobody can explain.
Commission calculation	Spreadsheet from a changing report	Calculated with inputs retained	Being able to show the workings resolves most disputes in one conversation.
Staff checking their numbers	Wait for a month end statement	Same day view, pending flagged	Visibility of what is still pending prevents the pay cycle surprise.
Changing the attribution rule	Reports quietly recalculated	New version, effective dated	A rule change affects pay, so it needs a date and an announcement rather than a silent deployment.

Where Sales Credit Loses Credibility

Attribution reconstructed after the fact

Deriving credit from delivery postcodes, timing or customer history gives a different answer every time the logic is touched, and staff work that out quickly. Once they believe the number is arbitrary, the scheme stops changing behaviour and starts generating complaints. Capture the attributing fact at the moment of the transaction and store it on the order. If you cannot capture it reliably for a particular order type, say so openly and exclude that type rather than estimating.

Two definitions living in two reports

Operations reports store performance one way and finance reports it another, both are internally consistent, and the rankings differ. Every meeting then starts with a debate about whose number is right. Agree one definition per order type, hold it in one place that both reports consume, and if a second view is genuinely needed, label it clearly as a different measure rather than letting it circulate as the truth.

No clawback, or clawback nobody expected

Paying commission on sales that are later refunded means paying twice for nothing, and it concentrates in categories with high return rates. Applying clawback without telling anyone is worse: it produces a pay cycle where several people are unexpectedly short and the scheme loses whatever goodwill it had. Set the earning point explicitly, publish the clawback rule, show pending amounts before they are paid, and get your payroll adviser to confirm that deductions and adjustments are handled correctly.

Capture that slows the transaction down

If entering a staff code adds fifteen seconds at a busy counter, it will be skipped exactly when volume is highest, and your attribution data will be systematically biased against your best trading periods. Design capture around the flow of work: a scanned badge, a short code, a default that is right most of the time and can be changed. Then measure the capture rate per site and treat a low rate as a design failure rather than as a discipline problem.

Commission treated as reporting rather than pay

Once a number feeds pay it inherits obligations. Commissions paid to employees count towards the super guarantee, and from 1 July 2026 all commissions are qualifying earnings for that purpose. For staff covered by an award or agreement, commission is generally paid on top of the minimum wage, and commission only pay is allowed only where the award or agreement provides for it. Records need to be kept. None of that is a reason to avoid commission, but it does mean the calculation needs an audit trail, an approval step before payroll, and sign off from your payroll adviser on the treatment rather than a formula somebody wrote in a spreadsheet three years ago.

Individual performance data handled carelessly

Per staff sales data is personal information about employees, and league tables have a way of travelling further than intended. Restrict access to individual numbers to the people who need them, aggregate where you can, be clear with staff about what is measured and how it is used, and note that the employee records exemption in the Privacy Act 1988 is narrower than often assumed: it covers a private sector employer’s own handling of employee records for the employment relationship, and does not extend to contractors or service providers handling that data for you. Take advice if you are unsure.

How Yes AI Approaches Attribution and Commission

The rules agreed and published before the build

We work through each order type with operations, finance and a couple of store managers, and produce the rule set with effective dates. Involving store managers early is how you find the order types head office does not know about.

Capture designed for a busy counter

Attribution recorded at the moment of sale in a way that takes seconds, with capture rates monitored per site so gaps surface as a fixable design problem rather than as an argument about diligence.

Built, hosted and monitored by us

The flows run on a managed cloud automation layer that we operate, with every calculation retaining its inputs and same day alerting if a feed fails. A missing sales extract raises an alert rather than quietly understating a store’s month.

Honest advice about scope and about schemes

If your point of sale and ecommerce platform can already carry a staff identifier through to reporting, the answer may be configuration rather than a build, and we will say so. We will also say when a commission scheme is too complicated to be motivating, because a rule nobody can predict changes behaviour less than a simpler one they can.

How the Work Runs

Five steps. The rule set and capture usually come first, with commission calculation following once the data is trusted.

Map every way a sale can happen

We list the order types in real use, including phone orders, staff assisted online orders, endless aisle, collection and the awkward ones a manager mentions in passing. Then we check what data each one currently captures, which is usually less than assumed.

Agree the attribution rules

One rule per order type with tie breakers, plus the earning point and the clawback rule for commission. Agreed with operations, finance and store representatives, and checked by your payroll adviser before anything is built.

Specify capture and calculation in plain English

What is captured where, how it travels through splits and refunds, how the calculation works, and what staff and managers can see. Written so a store manager can check it, and approved before build.

Build, run in parallel, compare

Built with attribution written at order creation and calculations retaining their inputs. Run alongside the current method for at least one full period, and explain every difference rather than averaging them away, because those differences are what earn trust.

Operate, monitor capture, extend

Same day visibility for staff, an owned dispute path, capture rate monitoring per site, and alerting on failed feeds. Rule changes are versioned and announced, never applied retrospectively without a decision.

FAQ

Should the fulfilling store or the collection store get the sale?

There is no universally right answer, which is why the decision has to be deliberate. Crediting the collection store encourages shops to promote collection and to look after the customer at handover. Crediting the fulfilling store rewards the picking and packing work, which matters if that work competes with serving customers on the floor. Splitting recognises both and makes every downstream report more complex. Pick the behaviour you want to encourage, publish the rule with an effective date, and revisit it annually rather than quietly.

How do we credit a staff member for an order the customer places later at home?

With a captured identifier, not a guess. The practical mechanisms are a clienteling link or a saved cart carrying a staff reference, a code the customer enters at checkout, or the staff member placing the order on the customer’s behalf in store. All three work, and all three need to be quick enough that staff use them consistently. What does not work is inferring it afterwards from the customer’s postcode or from which store they last visited, because that produces numbers nobody can defend.

Should commission be paid on GST inclusive or exclusive values?

Almost always on the ex GST value, since GST is not your revenue. Paying on the inclusive amount overstates the commissionable base by about ten percent of the ex GST figure, which is a material overpayment across a year and awkward to correct later. Whichever base you use, state it in the scheme document, and be consistent about how freight charges, discounts and gift card redemptions are treated, because those are the next three questions staff will ask.

When is a commission actually earned?

It is a choice with real consequences. Earning at the point of sale is simplest and pays fastest, but exposes you to refunds. Earning at dispatch or delivery reduces that exposure and delays recognition. Earning after any change of mind window has passed is the most conservative and the least motivating. A common approach is to accrue at sale, show it to staff as pending, and confirm after delivery. Whatever you choose, write it down, show pending amounts, and have your payroll adviser confirm how adjustments should be handled in pay.

Do we need to change our point of sale to do this?

Often not. Most modern point of sale systems can carry a staff identifier and a location on a transaction, and most ecommerce platforms can carry order attributes. The gap is usually between them: attribution captured in one system and never carried into the order record that reporting and payroll use. That is an integration problem rather than a replacement problem. Where the point of sale genuinely cannot capture what you need, there are usually workarounds worth considering before a platform change.

How do we handle returns processed at a different store?

Record both facts. The refund adjusts the original attributed sale, so net numbers stay honest and clawback works. Separately, credit the store that handled the return with the service interaction, because otherwise stores learn to discourage returns they did not cause, which costs you customers. Trying to express both in a single sales number is what creates the argument. Two measures, clearly labelled, resolves it.

Is a complicated scheme worth building?

Usually not. Schemes with many tiers, split rules and conditions are hard to explain, hard to predict and therefore weak at changing behaviour, while being expensive to build and to maintain. If staff cannot roughly calculate their own commission, the scheme is unlikely to change what they do. We will build a complex scheme if that is the decision, but we will tell you first that a simpler rule, visible daily, is usually easier for staff to act on.

Make Sales Credit Something People Trust

Book a call. We map every way a sale reaches you, agree the attribution rules, and give you a priced plan for capture, calculation and visibility.