A field guide to polls in decision making
Polls like this - completely inconclusive - seem to fill slack channels everywhere. I intend to describe some reasons why this happens and a recipe to get decisions over the line. It’s aimed primarily at those readers charged with a team or the responsibility for a decision.
If you don't want or need the background information, click here for the conclusion.
What types of polls are there?
Many tools provide one "default" poll configuration but it seems usually optimized for the poll providers' engagement statistics rather than the effectiveness of your process. Here are some of the different poll parameters and the effects I have observed using them.
Forced choice polls
These are polls where everyone must cast a vote. I’d discourage their use for decision making purposes.
However inviting it is to compel fence sitters to pick it generally makes people very uncomfortable and people unsure of their decisions tend not to execute on them very well. This is enough of a reason to limit their use. There are uses for forced polls outside decision making however, such as aligning a group by compelling them to look at the same issues.
Many people make the assumption that fence sitters don’t care or are otherwise disengaged. In my experience this is quite rare - there’s usually another reason such as assuming they are not qualified or thinking that revealing their views might invite ridicule. This meta-information however is valuable in itself; it indicates how strongly a group feels about an issue, or how safe everyone feels to contribute. If you force people to enter a vote you risk losing such information.
Named vs Anonymous polls
Anonymous polls are those where you can’t see who voted, though you still see a running total. Named polls are those in which we can see who voted for what (or those that didn’t vote). Unlike most of the different poll types there isn’t an obvious choice here.
Deciding which is appropriate seems to depend on the overall psychological safety level of the group. Prefer anonymous votes if the group is large, if the group hasn't developed a productive way to approach conflicts, or if the participants don't know each other well. If the group is close knit and open, save yourself some time and use named polls.
I often see people struggling to navigate discussions on anonymous poll results with the tendency to want to de-anonymise voters. This is unwise as the reason for anonymous polls is to encourage psychological safety and trying to de-anonymise results will encourage the opposite response.
Here are some questions you can ask after anonymous polls that move the conversation along without de-anonymizing the results:
If some small percentage of people picked an option (sometimes those people will be brave!)
It seems a couple picked X, what made them think differently?
How much of a risk is missing something the minority know/think here?
If it's relatively even (people will be less fearful of de-anonymization here)
For those that picked X, what do you think made people pick Y?
What makes X better than Y?
Open vs Closed polls
An open poll is one in which participants can see the results update as people pick options. These polls tend to lead to very high engagement for polling apps because it turns the vote into a watchable drama. Unless the drama is the purpose I would avoid open polls.
This drama damages the poll itself because people see the opportunity to risk nothing if their “team” is winning. They’ll tend to sit and watch, only voting when they see their preferred option is likely to lose. Likewise in named polls people may deliberately support the losing option to show support for the person behind the option; to save them a potentially embarrassing landslide. This all tends to lead to 50/50 splits (or close to) on every topic.
Binding vs non binding votes
Binding votes are where it's decided in advance what a poll's results will mean, for example in an election we know that the winner will take power automatically. Non binding polls/votes are taken as informative and they are explicitly being used for information gathering. I’d generally advise against binding votes for small (1-100) groups because it's not necessary and the information a poll's results provide may change the type of decision you make.
The perturbation of results in binding polls occurs because knowing the conditions in advance - like a 51% requirement or a supermajority - changes the nature of particular votes. The vote that pushes something over the threshold for example is worth considerably more than the first vote cast because that could be the "winning" vote. This means there’s a real incentive to delay your vote especially if it is an open poll/ballot.
Although not exclusively a property of binding polls, the binding nature also prevents leaders from declaring the poll invalid for a justified reason. For example if someone on holiday had some crucial information that would have changed the outcome. The saddest outcome for a binding poll or vote is one where important information is revealed by the poll results itself. When we have votes like this, usually nobody is happy with the outcome and the action tends to get sandbagged.
Important polling principles
There's more going on here than just configuring a poll. Here are some of the meta-poll topics I've found important.
Decisions should be made by the people responsible for the outcome
Even in a binding vote/poll, someone sets it as binding with their authority (and so takes responsibility for the outcome).
Those new to responsibility tend to shirk this aspect of a vote. Oftentimes new managers for example will want to run a team entirely democratically, having periodic referenda on course changing topics. It’s a noble goal but there are a couple of almost insurmountable problems, namely:
a) Your team doesn’t want the responsibility
People love being asked, but if you’re charged with a team you’re paid for the responsibility and the people in your team aren’t.
People want to be listened to and understood, sure, but this does not mean they want to be deferred to on every topic (even if they say they do). What’s more, continually polling all decisions can create a paralysis when a unilateral decision is necessary; this opens the door to relentless discontent and a tyranny of complainers. You risk getting nothing done on this path!
This isn’t an excuse to be a tyrant of course. If you could do it yourself, you wouldn't need a team. It’s about creating a healthy homeostatic balance where the team can influence decisions and understand when decisions are made (by you) that you’re acting in the best interest of the group. It’s in their interest from there on to commit - to avoid sandbagging or continually contesting the call.
b) Your team doesn’t know all the ramifications of what they’re picking
If you’re charged with a team, someone expects you to look after them. The organization or social group is expecting you to be accountable for the result. This is how you will be evaluated - not the rest of the team. This is likely one of many things that you know that your team does not, or at least doesn’t really think about. Other things include what your peers are doing, the team’s reputation and what your boss is trying to get done.
This kind of asymmetry is rarely acknowledged in democratic systems but it’s why nearly every country has a representative democracy rather than a direct one. We pay people to collect information and make decisions on our behalf because it would be exhausting for each of us to maintain the same level of context on a topic.
Acknowledging asymmetry in information means that you know things that your team does not, but it also means that they know things you do not. That’s why you’re asking them for a vote, right?
Make it easy for people to change their mind based on what other people think
Most people create votes to ensure people converge on a position rather than as a tool to generate divergence, but this is a mistake.
The results of a poll change how people think about a topic as often it’s their first chance to see how other people think about something. If you feel strongly about something but you realise you’re the only one on the team, you may change how strongly you feel about it.
What’s more many votes are on non-binary topics where the best option is some blend of the various options (and not necessarily a compromise). This is only achievable by discussion with the poll votes available and the ability to ignore the results if it seems right to do.
Putting it all together; the vote, discuss, decide play
This "play" is described in David Marquet's outstanding book "Leadership is language". I recommend reading it!
Before you start
Explain the process and who will make the decision (and when). It doesn’t have to be you, but it does need to be a named person willing to be responsible for the outcome.
1. Vote
Create a poll in your tool of choice. Here's a cheat sheet for the options:
| Poll option | Choice |
|---|---|
| Non binding | You decide based on the outcome. Don't decide what happens in advance |
| Closed | No watch along drama here - the results are shown at the end |
| Anonymous | Decide based on psychological safety level of the group. If it's low provide anonymity, otherwise show what people pick |
2. Discuss
In the event of a landslide with >90% of people picking one option, quickly ask any dissenters or abstainees if they think we’ve missed anything (or if anonymous, ask the group what the dissenters/abstainees might be thinking differently).
In the event of an even split, ask a representative from each group why they voted the way they did. Follow the room here, do people naturally want to change their votes after a discussion? Can the different opinions be joined? What do people think should be changed in the option they didn’t pick? Any really strong opinions? Why?
3. Decide
Very likely the previous discussion will have ended in a concrete recommendation, but if it doesn’t make a call and move on. Likely you will never have more information than this. Explain why you think what you do, and if it’s something that can be reversed, when you will next consider a review.
