Not being in the business of voter suppression, I couldn't tell you anything authoritatively. But if I was, I'd use the data to find who has a spotty or non-perfect record with the hypothesis that they're more susceptible to suppression methods than someone with a near perfect record.
Then the next step would be targeting locations that lean towards one party or using a combination of retargeting ads +obvious party selectors to run a BS phone or ad campaigns that disenfranchise or provide misleading information. Like this. Which was a real thing that happened:
https://en.wikipedia.org/wiki/2011_Canadian_federal_election...
"I know you'll probably vote for $foo. If I check after the election and find that you did|didn't vote, bad thing $bar might just happen to your friends and family."
Edit: It's probably not a particularly good threat, but it could maybe be used to suppress certain targeted demographic groups that tend to vote one way or the other.
This seems like a pretty ludicrous threat to make. You could just submit a blank ballot or vote for the opposing party and your hostage-taker would have no way of knowing.
Party registrations are public, and you can make a pretty accurate guess that someone who is registered to one party is likely to vote for that party, too.
I can imagine a situation where someone's boss doesn't like a certain party or politician, and they decide to take it out on their subordinates that they suspect will vote/have voted for them. Seeing a D or R next to an employee's name and that they voted could be enough evidence for them to face discrimination or threats not to vote.
Then the next step would be targeting locations that lean towards one party or using a combination of retargeting ads +obvious party selectors to run a BS phone or ad campaigns that disenfranchise or provide misleading information. Like this. Which was a real thing that happened: https://en.wikipedia.org/wiki/2011_Canadian_federal_election...