A marketing report should let you see what was spent, what happened and how confidently the two can be connected. Start with orders or completed deals, then use advertising and analytics reports to explain the route.
Keep three calculations separate
ROAS = attributed revenue ÷ advertising spend. Hypothetically, HK$150,000 of attributed revenue divided by HK$50,000 of ad spend is 3× ROAS. This excludes product, fulfilment, agency and other costs.
Acquisition cost = defined acquisition costs ÷ new customers. State whether the numerator includes sales staff, production and agency fees. Count new customers consistently, rather than mixing orders with customers.
Marketing ROI = (incremental contribution before marketing − marketing cost) ÷ marketing cost. State the period, included costs and how incrementality was estimated. Attributed revenue alone does not establish incremental contribution.
Begin with the business record
For ecommerce, reconcile paid orders, refunds and cancellations. For a service business, connect the enquiry to qualification and the eventual deal. Use stable identifiers and define duplicates so one person is not counted repeatedly.
A simple lead record can use Lead, Disqualify, Qualify and Deal. Define the criteria with the sales team, and keep dates for status changes. A campaign with inexpensive enquiries can still be expensive if few become suitable customers.
Distinguish an event from a result
A visit, a contact-button click, a successfully saved form and a completed sale are different events. Map them separately. Verify that the right event fires once at the intended point and that the record reaches the business system.
Google Ads supports different conversion sources, including website and offline activity. Choose and configure the business action deliberately instead of treating every interaction as an equally valuable conversion.
Attribution explains credit, not causation
Platforms may assign credit to the same purchase using different windows or models. Do not add their attributed revenue totals together. Compare settings before comparing performance, and reconcile them to the underlying order or CRM totals.
A change of attribution model can move credit between channels without creating another sale. A useful report shows its model and date range, includes the sales-cycle delay, and explains what remains unobserved. Check the options available in the current analytics product rather than relying on old model lists.
Build a review that leads to a decision
- Start with spend, valid outcomes and contribution where available.
- Break down qualified outcomes by campaign or source.
- Record measurement gaps, delayed outcomes and changed offers.
- Choose an action: repair a broken route, change a message, improve follow-up or adjust a test.
- Keep a change log so the next review can assess what happened after that action.
For a new campaign, report small samples honestly rather than inventing a benchmark. Where a controlled experiment is practical, it can help estimate incremental effect. Where it is not, explain the limits of observational comparisons.
Our tracking and optimisation service covers this connection between campaigns and business feedback. The budget guide shows how those definitions support spending decisions.
Editorial review: 30 September 2026. Practical guidance; examples are not performance guarantees.
