How to tell whether your marketing is working
Start with the customer action you need. Then work backwards to the evidence that would justify continuing, changing or stopping the activity.
A campaign can generate attention without producing the outcome the business needs. Before judging its performance, define what a useful result is: an appropriate enquiry, a qualified opportunity, a purchase or another action that fits your business.
Keep that definition alongside the report. It should be possible to explain why a number matters without relying on its size, its color in a dashboard or whether it improved compared with last week.
- 1Attention
Was the right audience reached?
- 2Useful action
Did a relevant enquiry or purchase happen?
- 3Business result
What value and cost followed?
Choose the outcome before the report.
SBA marketing guidance recommends setting goals, tracking costs and comparing marketing and sales spending with the revenue generated. It also acknowledges that some tactics are difficult to measure. [1]
Choose one outcome and explain what qualifies. An enquiry count becomes more useful when the team agrees which enquiries are relevant. Keep downstream checks visible so a channel producing many low-fit responses is not automatically treated as the best performer.
Use consistent costs and observation windows.
Decide which costs the comparison includes and apply the same rule to each activity. Keep a record of campaign dates, major changes and the point at which you expect the outcome to become observable. Avoid comparing a mature campaign with a new one as though both have had equal time.
Where appropriate, connect the initial action to later purchases or customer quality. Check that the underlying events work and that duplicates, internal tests or missing records are understood. A precise-looking ratio built from unreliable counts is still an unreliable guide.
Separate assigned credit from proof of cause.
Google defines attribution as assigning credit to interactions along the path to an important action. A model determines how that credit is shared. Its documentation shows that models can treat the same path differently. [2]
Name the model and data limits when presenting the result. If you need to know what would have happened without an activity, consider whether a suitable controlled test is feasible. A before-and-after comparison can be affected by other changes; an attributed sale alone does not settle that question.
Put it into practice.
Choose one current activity. Write its intended outcome, cost rule, measurement window and decision date. Check the events, then agree what evidence would justify continuing, adapting or stopping. Keep uncertain attribution visible instead of filling the gap with a confident estimate.
The boxes are a reading aid and are not saved. Download the worksheet to keep your notes.
Download the practical worksheetSources & reading
External sources support the factual background. The prompts and next steps are editorial recommendations, not promises of a particular result.
- 1Marketing and sales
U.S. Small Business Administration. Marketing section: goals, budget, cost tracking and comparing marketing/sales costs with revenue. Some channels are difficult to measure.
- 2Get started with attribution
Google Analytics Help. Explains how attribution assigns credit to interactions along a path to an important action. Different models can assign credit differently.