I disagree with your analysis of the survey.
While it is true that more people polled want to have no allies in a game than any specific number of allies, 45% of people wanted at least 2 allies which is a slightly higher amount than the number of people who wanted no allies. There was also the 10% of people who said that the ideal number depends on the game. It stands to reason that even if some of them might usually prefer games with no allies, they'll admit some maps are better with them, and a majority of those people likely tend to prefer games with an ally cap above 0.
There is also the matter of the sample size being rather small. With a sample size of 16*, I do not believe that enough people answered to make the results a good barometer of how people feel about alliances in the game.
*I am referring to the "How many allies makes the perfect game?" poll which closed with 16 respondents.
I would also argue that the survey structure should be like this:
-No Allies is ideal, but if forced, I would settle for 1 ally
- ... 2 allies
- ... 3 allies
- ... many allies
-1 ally
-2 allies
-3 allies
-many allies
First of all, this structure lets you know how many people want allies in the first place, as well as how many alliances people prefer to have. I also omitted the "it depends on the map" option, as it does not give you information that could help you make decisions unless you are willing to post polls on a map by map basis.