Overview
Over October 2-4, 2020, in response to new SMS rules promulgated by Twilio, Vote Rev and Vote Rev Action Fund tested three ways to incorporate “Reply STOP to unsubscribe” into outgoing vote tripling SMS. These messages were sent to early voters in Wisconsin and North Carolina, with the goal of mobilizing them to remind friends to vote. In total, we texted 62,277 voters during this experiment, roughly half each in the two states. The three versions of the text were:
Version 1 (beginning): (Reply STOP to unsubscribe.) Hi [name], it's [name], a volunteer with Vote-Tripling .org. Public records show you voted in [state] - thanks for being a voter! We have a quick favor to ask: can we count on you to remind 3 friends to vote?
Version 2 (middle): Hi [name], it's [name] , a volunteer with Vote-Tripling .org. (Reply STOP to unsubscribe.) Public records show you voted in [state] - thanks for being a voter! We have a quick favor to ask: can we count on you to remind 3 friends to vote?
Version 3 (end): Hi [name], it's [name], a volunteer with Vote-Tripling .org. Public records show you voted in [state] - thanks for being a voter! We have a quick favor to ask: can we count on you to remind 3 friends to vote? (Reply STOP to unsubscribe.)
Key results for each version are shown below.
Tripler rate: v1 ≠ v3 (p=.000); v2 ≠ v3 (p=.000)
Names rate: v1 ≠ v2 (p=.008); v1 ≠ v3 (p=.063); v2 not distinguishable from v3 (p=.401)
Opt-out rate: v1 ≠ v2 (p=.000), v1 ≠ v3 (p=.000), v2 marginally distinguishable from v3 (p=.137)
Of the three versions, Version 2 is the clear top performer across all metrics, with these differences in most cases significant.
A few things worth noting:
Relative to putting the warning in the middle, putting the warning at the end suppresses the opt-in rate considerably, but has relatively little impact on the opt-out rate. It is also possible that those who do opt in are more committed, based on the higher relative name provision rates.
Relative to putting the warning in the middle, putting the warning at the beginning raises the opt-out rate considerably but has relatively little impact on the opt-in rate.
Across the board, including this language radically increases opt-out rates compared to previous vote tripling implementations, and modestly decreases opt-in rates.
Still have questions? Email hi@voterev.org
