Measurement
Part of Radio advertising measurement explained for UK media buyers
Five radio advertising key metrics for an England client report
Report radio advertising key metrics in England with five headline numbers, a seven-field sheet, a worked cost per thousand and a quarterly review.
What to take away
- Put five numbers on the first page of a client reportreach, frequency, cost per thousand, response volume and cost per response.
- Report station-delivered audience and campaign-attributed response side by side, never blended into one figure.
- Give every number a row on a seven-field sheet, so a metric means the same thing in every update.
- Review the sheet each quarter and retire any metric nobody acts on.
Choose the five numbers for the headline
Most campaigns in England need five numbers in the headline: reach, frequency, cost per thousand, response volume and cost per response. Each answers one client question, so none should be blended with another. Anything else belongs in an appendix.
| Number | The client question | Where it comes from |
|---|---|---|
| Reach | How many adults heard the spots? | Audience data for the stations bought |
| Frequency | How often did they hear them? | Audience data for the stations bought |
| Cost per thousand | What did each thousand listeners cost? | Media cost and reach |
| Response volume | What did people do next? | Your own tracking: calls, site visits, code redemptions, branded search |
| Cost per response | What did each response cost? | Media cost and response volume |
Add average hours for sponsorship buys, and share of category spend where a benchmark exists, using benchmark research you have checked. The cost guide and its data sources explain where the media cost line comes from.
Work one cost per thousand before you send it
Cost per thousand is the media cost divided by reach in thousands. A buy costing £20,000 that reaches 400,000 adults shows a cost per thousand of £50. Put that figure beside response volume, so the client sees what the money bought and what it produced.
- Take the media cost£20,000.
- Convert the reach to thousands400,000 adults is 400 thousand.
- Divide the cost by the thousands£20,000 divided by 400 gives £50.
Cost per response follows the same sum, with response volume in place of reach. Show both figures to the client, with the calculation written beside each, so they can check the working.
Fix the sheet behind each number
Each of the five numbers needs a row with seven fixed fields. Without them, a metric means one thing in January and another in April. The sheet keeps an agency and a client looking at the same thing, and it makes an audit possible later.
- Metric name, written the same way each time.
- Definition, in one sentence.
- Source system, named.
- Reporting period, with start and end dates.
- Attribution window, if the metric is response-based.
- Owner, the person who signs the number off.
- Notes, for any change in method.
Date the report against the RAJAR calendar
RAJAR publishes its survey results on a fixed quarterly cycle, so a report dated before the survey lands will be late or built on the previous wave. Check the RAJAR release schedule before you promise a board a chart.
RAJAR quarterly reporting calendar
- Before bookingConfirm RAJAR release dates
- Quarterly cycleSurvey results published
- After releaseBuild report on latest wave
- On reportState period data covers
State the period the data covers, not the day you downloaded the file. The audience research guide covers how the quarterly data is built, so you can say which wave a figure belongs to.
Fix the attribution window for response numbers
Response metrics need a defined window before you quote them. Radio response often arrives after the spot, sometimes days later, so fix a window, for example 14 days from first transmission, and state it on every report. A shorter window changes the answer, so note the trade-off.
Mixed windows across stations make comparison meaningless. Log the window in the tracking platform as well as the report, so the two cannot disagree. The attribution methods article compares the ways to tie response back to a campaign.
Say how sure you are
Survey-based audience data carries a margin of error, and response data depends on tracking coverage. State both. A reach figure presented as exact invites a challenge you cannot answer. Where a number is modelled, say so and name the inputs.
Set a tolerance for what counts as a real change, and leave smaller movements out of the summary. If site visits are counted with tracking pixels, small pieces of code embedded in a website or an email, ask the client what is in place. Teams that have been caught out before will find the recurring faults catalogued in measurement mistakes.
Review the sheet each quarter
Metrics drift as tracking changes. Review the definitions each quarter, log changes in the notes field and re-baseline where needed. Retire any metric nobody acts on, and replace it with one that drives a decision. A metric that quietly changes meaning is worse than one you retire.
Once the sheet is stable, a reporting dashboard can carry the same fields, so the definitions you agreed travel with the numbers.
Common questions
Which numbers go on the first page?
The five headline numbers: reach, frequency, cost per thousand, response volume and cost per response. Add the written attribution window beside any response figure.
Should I report reach and response together?
Report them side by side, never combined. A blended score hides which part of the campaign worked.
How long should the attribution window be?
Set it in writing before the campaign starts. A 14-day window is common for direct response, but the right length depends on your tracking setup.
When should a metric be retired?
When nobody acts on it. Replace it with a number that drives a decision, and record the change in the notes field.
