Public opinion polls enter the public domain for a reason. Focusing on the poll sponsors of public polls gives us considerable insight. It tells us how and why public opinion becomes part of the national conversation.

Overall, the analysis shows that research firms themselves (33%) and the media (25%) pay for a majority of polls. Non-media sources are responsible for 28% of the polling releases.

Since 2016 there has been fluctuations in the amount of polling done by different types of clients. Late 2018 and early 2019 is characterized by a larger amount of non-media sources. Media polling, which was down in the early part of the year, surged in the late summer and fall of this year.

… a surprising and perhaps troubling share of releases have no sponsor identified.

Type of Poll Sponsors of Publicly Released Polls

In assessing who the sponsor is, there is often no clear information. Two types of cases emerge. First, what are labelled as media releases may, in some cases, not actually be paid for by the media outlet. The media may have been given exclusive early access to the poll.

Second, many polls featured on market research company pages do not have a sponsor identified. Some companies explicitly acknowledge that they funded the research but not all do. We will discuss some of this below.

Type of Content Supported by Sponsors

Content is not supported equally by the four types of sponsors. The media are the most likely to sponsor releases about the state of the economy (24%). Research firms are the most likely to sponsor releases about electoral and political competition (51%) followed by public policy issues. Non-media clients focus primarily on perception of life and world as well as public policy.

Perceptions of Life and World — Polling as Content Strategy

Almost a quarter (22%) of polls released in Canada fit into a category that we call Perceptions of Life and World. In some ways, this is a residual category. If the poll is not about the economy, political competition or public policy it fits into this category. 

But what is in this category? Effectively, these are polls mostly about consumer behaviour and interests. How we bank. What flavour of ice cream we prefer. The key is that there is no inherent call to action on the policy front. Advocacy for policy change polls are not included in this category.

The purpose of this polling by non-media sources is to provide content about their business. So, a pet store does a survey on pets. A bank surveys on mortgage behaviour. The results are released and linked to the product offering.

News Media | All about the Story

It is not a new insight that the news media tend to focus on the principles of news; drama, conflict and change. As such, horserace coverage (38%) is the top category of all releases.

Economic issues (24%) and policy (32%) are also important. Note that the media rarely sponsors polls about general perceptions of life and world. There is almost always a policy hook. 

Research firms Seeking Publicity

Electoral and party competition is the primary category of poll releases by research firms. These polls represent easy and short questions that can usually garner media attention. 

The media and research firm polls release polls that are fairly similiar in terms of research area. The main difference is the slighter higher share of polls on perceptions of life and world for research firms.

There is a natural symmetry between the news media and research firms. The goals of research agencies is to get coverage. The goals of the media are to write informative stories. Polls provide something new. The synergy is so strong that many media releases may not have involved money changing hands — they are defacto research firm polls.

Note that research firms includes several organizations committed to conducting and sharing public polling information in a non-partisan manner.

Unknown Poll Sponsors, Really

It seems hard to believe that a poll could manage to get into the public domain without some acknowledgement of who paid for it. Someone, either a client or the research firm itself, did. In 14% of cases the poll sponsor is unknown. An unknown percentage of these are self-commissioned by a research firm.

If a research firm does not explicitly say that they did not have an external client pay for the survey questions, should we just assume there was no external client?

For some polling the answer might be yes. For example, most elections and party competition polling by major research companies was probably not commissioned by someone else. The share that is allocated to unknown drops to 12% if we treat elections polling as research firm funded,  

Nevertheless, 43% of all public releases where the sponsor is unknown are about public policy. And 23% of all public policy releases do not have a sponsor listed. For these, there may have been a real client who does not want to be known. In the absence of explicit acknowledgement that the polling was self-commissioned that are treated as unknown. These are releases we should be concerned about.


Knowing the sponsor of a poll provides you with some interesting insight into why the poll is getting into the public domain. In most cases, it is clear. In a surprisingly large number of cases the sponsor is unclear. And it should not be. Poll sponsors, particularly research firms, need to be more transparent.

Fifty-eight per cent of all polling gets released into the public domain by the news media or by research firms. This impacts, in some cases, what gets asks and what is covered. Questions that are more political charged are more likely to be asked.

Nevertheless, the poll sponsors of public policy polls are often non-media sources. These polls contribute to the public discourse but are designed to influence the policy process/ Canadians rather than inform.

In fact, non-media (private clients) polling is interesting not only because of its role in influencing policy but also in the questions it asks of Canadians that are not politically motivated. These polling releases are designed either to pique interest in the brand or to support a sales process.

Updated from original report published in September 2018.