The annotated journal
How to set KPIs and marketing objectives
An objective states the change you want, a KPI measures it, and a threshold turns the measure into a decision. How to build that chain, one measure at a time.

An objective states the change you want. A key performance indicator is the measure you will judge that change by. A threshold is the value of that measure which makes you do something different. Most plans define the first 2, skip the third, and produce reports nobody acts on.
This page sets out that chain and how to build one. Measuring the effectiveness of your brand messaging and measuring the impact of your stakeholder communication apply the same chain to their own subjects.
The 4 links
| Link | What it is | The test it has to pass |
|---|---|---|
| Objective | The change you intend to cause, with a size and a deadline | Someone outside marketing would recognise it as worth paying for |
| Measure | The one number that tells you whether the change happened | It has a named source you can open today |
| Threshold | The value at which you act differently | You wrote it down before you saw the data |
| Decision | What you do when the measure crosses the threshold | It names an action, an owner and a date |
A break anywhere in the chain makes the rest inert. An objective with no measure cannot be judged, and a measure carrying no threshold and no decision produces a chart that gets presented and then filed.
Writing the objective
Targets set this way usually follow the SMART pattern, the acronym George T. Doran proposed in 'Management Review' in 1981: specific, measurable, assignable, realistic and time-related. The letters have mutated since, and the achievable-relevant-timely variant is one of several in circulation. Which variant you use matters less than the 2 parts most objectives are missing, which are a number and a date.
2 failure modes account for most unusable marketing objectives.
The first is an objective with no business consequence. "Grow website traffic by 100% this year" and "increase Instagram followers" are both measurable, and neither says what would be different if it happened. Nobody in finance would fund either as stated.
The second is an objective set from what marketing wants to do rather than what the business needs. A launch objective of £1 million in new revenue next quarter can be met while the acquisition cost leaves the product unprofitable. An objective has to survive contact with the people who own the numbers it affects: sales, finance, and whoever owns the product.
A worked example. A small restaurant chain wants to increase revenue by 10% over 12 months. That is a business goal rather than a marketing objective. The marketing objective sits underneath it and names the contribution marketing is accountable for: 10% more qualified enquiries for the catering side within 3 months. The numbers here are illustrative. The structure is the part that transfers.
Choosing one measure
Choose one measure for each objective. Teams that pick 3 spend the review arguing about which to believe.
A usable measure passes 3 checks:
- Named source. You can say which system holds the number and open that system now. A measure whose source is "we would need to set up tracking for that" is a project you have not started yet.
- Existing baseline. A number means nothing on its own. A conversion rate of 4.3% is neither good nor bad until it sits against 3.1% last quarter, or against the 6% the plan assumed.
- Able to move. The measure has to be capable of changing inside the period, through work you control. Annual retention cannot report on a quarterly objective.
Where the outcome is slow or expensive to measure, a proxy is legitimate on one condition: write down what would make the proxy stop tracking the outcome. Qualified enquiries proxy for revenue until the definition of "qualified" drifts, which it tends to do in the quarter the target is missed.
Keep the measure and the reading apart. The conversion rate is the measure. 4.3% is the reading. Reviews go wrong when people defend a reading instead of examining the measure that produced it.
Setting the threshold
The threshold is the value that changes your behaviour. Write it before the data arrives, because a threshold chosen afterwards is a description of what already happened.
Set 2 of them. The first is the level at which you continue as planned. The second is the level at which you stop and re-plan, which is harder to write and is the one that saves money. For the enquiries example: at or above 8% growth by the halfway point, continue; below 4% at that point, stop the campaign and re-examine the offer rather than the advertising.
Record the baseline before you change anything. The first period of collection has no comparison target. Its job is to exist, so that the next period has something to sit against.
Naming the decision
Against each measure, write the decision it feeds. If no decision consumes a number, take the number out of the plan. Background reporting is fine and belongs somewhere other than the objectives document, where it competes for attention with the measures that trigger action.
Each decision needs an owner and a date. Without them a threshold gets crossed, discussed, and left.
Reviewing
Review the set on a fixed cadence, quarterly for most marketing objectives, and put the review in the calendar next to the decisions it feeds. A quarterly review that happens is worth more than a monthly dashboard that does not.
Trend each measure against your own baseline. A published benchmark comes from a different market with different definitions, and comparing against it produces argument rather than a decision.
The plan on one page
| Objective | Measure | Source | Baseline | Threshold | Decision |
|---|---|---|---|---|---|
| 10% more qualified catering enquiries in 3 months | Qualified enquiries per month | CRM enquiry log | 40 per month | Below 44 at week 6 | Marketing owner reviews the catering page offer by week 7 |
| Reduce customer churn by 5% in 6 months | Monthly churn rate | Billing system | 3.2% | Above 3.2% for 2 consecutive months | Account owner runs 5 exit interviews and reports at the next review |
Both rows are illustrative. The columns are what transfers.
Objectives written this way depend on knowing what your buyers are trying to do, which is a separate piece of work: see how to understand your audience and make sure your message resonates.
Fill in the decision column first
Open your current objectives document and add a decision column. For every measure already in it, write the action you would take, and who would take it, if that measure moved against you. The measures left with an empty cell are the ones to remove, and what remains will be short enough to review properly.