Test a hoarding before you buy it: ad trials on simulated streets
Put two to four versions of an ad on a zone's hoardings and posters, let simulated residents pass them on their real routes, and see which one they notice, remember and would act on.
Outdoor advertising is bought on faith. You pick a message, pay for the boards, and find out weeks later whether anyone noticed. Parallel Earth lets you find out first, on simulated streets.
How a trial runs
- Write the question, for example: which Diwali offer works best in Rohini?
- Add two to four versions, each with its language.
- Pick hoardings, posters or both, and one to seven days.
- The versions rotate across the zone's boards every day, so none wins by getting the best spot.
- Simulated residents pass the boards on their real routes; the trial counts who passed, how often, and how many noticed, remembered and would act.
A first trial
We tried a sweet shop's Diwali offer in Rohini from 5 to 7 November, three versions, on a panel of 412 simulated residents.
| Version | Noticed | Remembered | Would act |
|---|---|---|---|
| "Sweets for the whole family", in Hindi | 42% | 31% | 11% |
| The same idea in Hinglish | 40% | 26% | 7% |
| "Flat 20 percent off, terms apply", in English | 34% | 14% | 0% |
The hour-by-hour chart had two peaks: the morning rush, and 7 to 8 pm, when Rohini was heading to the Ramleela. A festival week changes when an ad is seen, not just how it lands.
The full method, with the exposure distances and the noticing priors, is in our research notes.
Frequently asked questions
Can a simulation predict how many people will see my hoarding?
No, and we do not claim it can. It ranks versions of an ad against each other on a simulated panel. Real reach depends on real traffic, which the simulation does not count.
How do you stop the best location deciding the winner?
The versions rotate across the same boards every day, so each version spends time on every board.