Political & Public Opinion Intelligence
Message Test Scorecard
Which campaign messages persuade, for whom and in which language.
- Typical length:
- 6 pages
- Format:
- Live dashboard and PDF
- Channels:
- SMS, USSD, WhatsApp, Field enumerators
Best message
Message D
68% believable
Lift over control
+17 pts
band +11 to +23
Messages dropped
1 of 4
Message C
The evidence
- Message D68%
- Message B61%
- Message A52%
- Control51%
- Message C47%
| Message A | Message B | Message D | |
|---|---|---|---|
| Boda boda riders | 38 | 49 | 63 |
| Mama mbogas | 41 | 55 | 66 |
| Farmers | 36 | 52 | 61 |
| First-time voters | 44 | 47 | 52 |
What this tells you
- 1
Local language wins
Message D, delivered in the local language, is rated believable by 68%, 17 points above the control at 51%. Message C in Sheng scores below control.
- 2
First-time voters respond differently
Message D lifts intent least among first-time voters at 52%, against 61% to 66% elsewhere. They need their own message.
- 3
Message C should be retired
At 47% believability it falls short of the control, and the gap sits outside the margin of error.
Decisions this supports
- Which message leads the campaign
- Whether to build a separate message for first-time voters
- Which language mix to use in radio and SMS
Method and confidence
Surveyed, n=2,500 respondents randomized into 5 groups of about 500, via SMS (40%), USSD (30%) and WhatsApp (30%). Margin of error plus or minus 4.4 points per message group at 95%. Differences under 5 points are treated as noise.
Closer labels every figure as surveyed or observed, with sample size and margin of error where they apply. All numbers on this page are invented to show the format and do not describe real Kenyan opinion, markets or organisations.