Combining Polling with Modeling to Drive Cost Savings

Toward the end of the 2018 election cycle, we had multiple clients and committees complain about the high cost of conducting both polling and predictive analytics (often known as – modeling or scoring). They wondered why these projects couldn’t be done at the same time.

The cumulative results were great to have – thorough understanding of the electorate and custom scores on issues to target voters – but even at the congressional level, candidates were paying $30k+ for their poll and then another $40k+ for predictive analytics.

That got us thinking? Why does polling have to live separately from predictive analytics?

We always tell our clients that it’s less expensive to add questions to a survey as opposed to fielding another poll. So why not add modeling questions to existing polls being fielded?

Throughout 2019, we tested this approach and supplemented the modeling questions with additional completes beyond the poll sample size.

We also had canvassers working for our clients ask the modeling questions at the door. This gave us another data source that cost essentially nothing extra.

The results were awesome! 

Since Cygnal’s multi-mode polls are already less expensive than traditional phone-only pollsters, it was only marginally more to also conduct the predictive analytics project. 

Our congressional-sized customers were able to get a benchmark poll (with modeling questions added) for $20k, and the addition of 9 – 10 custom scores was only about $19k more. So for $39k, our clients received the polling needed and the scoring desired for two-thirds the traditional cost.

The savings was reinvested into voter persuasion and GOTV, a much better use of resources than overpaying for polling and predictive analytics.

So as we enter the heat of the 2020 cycle, keep in mind that you don’t have to pick between polling and predictive analytics if you work with a firm that is focused on freeing up money to win races.