In this episode Elizabeth delves into the fascinating world of data and personalization with Hamish Marshall, former National Campaign Manager for the Conservative Party of Canada and seasoned expert at the crossroads of data and politics. They explore the evolution from the Two-Step Flow to the One-Step Flow of Communication, examining how data personalization and the changing media landscape have shifted the way information is disseminated in political campaigns. Hamish shares real-world insights into the practicalities and challenges of data-driven campaigning, touching on the nuances of voter targeting, the effectiveness of door-knocking, and the potential of personalized campaign strategies.
Additional Resources:
Check out www.polcommtech.ca for annotated transcripts of this episode in English and French.
In this episode Elizabeth delves into the fascinating world of data and personalization with Hamish Marshall, former National Campaign Manager for the Conservative Party of Canada and seasoned expert at the crossroads of data and politics. They explore the evolution from the Two-Step Flow to the One-Step Flow of Communication, examining how data personalization and the changing media landscape have shifted the way information is disseminated in political campaigns. Hamish shares real-world insights into the practicalities and challenges of data-driven campaigning, touching on the nuances of voter targeting, the effectiveness of door-knocking, and the potential of personalized campaign strategies.
Additional Resources:
Check out www.polcommtech.ca for annotated transcripts of this episode in English and French.
Elizabeth Dubois: [00:00:00] Welcome to Wonks and War Rooms, where political communication theory meets on the ground strategy. I'm your host, Elizabeth Dubois. I'm an associate professor and University Research chair in Politics, Communication and Technology at the University of Ottawa. My pronouns are she/her. Today we're talking about the One-Step Flow and personalization in campaigns with my guest, Hamish Marshall. Hamish, can you introduce yourself, please?
Hamish Marshall: [00:00:23] Sure. My name is Hamish Marshall. I've been involved in politics for more than 20 years, and my career has always been at the intersection of data and politics. The idea that while politics isn't a science in the strict sense, it's not all art either and there's a little bit of math involved, or there should be. I've been involved in campaigns from coast to coast and in other countries as well. But the biggest job I had was [when] I was the [Conservative Party of Canada] National Campaign Manager in the 2019 election.
Elizabeth Dubois: [00:00:55] Wonderful. Thank you so much for being here. I'm really excited to chat today. We're going to talk about data, personalization [and] the way that impacts campaigning. But what I want to start with is this idea of [the] One-Step Flow of Communication [Consult Bennett and Manheim’s article,The One-Step Flow of Communication], which is this theory that was developed in the early 2000 in response to a changing media environment and all this availability of data. So, as usual, I'm going to give a brief kind of overview from an academic perspective of what this theory actually is, and then you're going to tell me whether or not it makes sense.
Hamish Marshall: [00:01:28] Sounds good.
Elizabeth Dubois: [00:01:29] All right. So to understand the One-Ftep Flow, we need to go back a little bit into the history of political communication and media studies. Back in the 1950s, the Two-Step Flow Hypothesis was fairly innovative [Consult: The Two Step Flow of Communication by Katz]. It was this idea that we have opinion leaders who intervene in the flow of information. So rather than mass media, political campaigns, big brands, anyone who had resources to broadcast, rather than them directly affecting the wider public, what we actually saw was these opinion leaders who paid attention to their messages, and then they repackaged them, repurposed them, filtered them, curated them for a smaller set of everyday associates, families, friends, colleagues [Consult: Personal Influence: The Part Played by People in the Flow of Mass Communication by Katz and Lazarsfeld. Also, check out our previous episode, The Two-Step Flow and Opinion Leaders with Nick Switalski]. And those people were paying less attention to political news, for example. But they did pay attention to the opinion leader. And so we had this kind of two hops that the information flowed through. As the media environment changed, we stopped all watching the same evening news. For example, we started getting our own little slices of the information environment. We started seeing more personalization algorithms being used across social media, started seeing more personalization and micro-targeting in advertisements. This is within politics and outside of politics. The hypothesis is that we don't need opinion leaders as much as we used to. We get the kind of filtering and curation through this technology that's being integrated into our information environments. And so instead of primarily relying on a Two-Step Flow of Communication, we're back to a One-Step Flow of Communication. There's going to be a lot to unpack here. But before we get into it, what are your initial thoughts? Does this resonate? Are there questions you have about it? Things we should clarify? What do you think?
Hamish Marshall: [00:03:15] Yeah, I think saying that we're either in one or the other is a mistake. While the technology might change, and there are certainly many people who would get their information in a One-Step Flow way, as you describe it, there's many, many who don't. And there's many, many people who do both, right? The experience is not uniform. And that's not just by person. So like, you know, my parents who are in their 80s, overwhelmingly get their news almost entirely from watching the 6:00pm news
Hamish Marshall: [00:03:46] And reading a physical newspaper, [which is] very mediated by what the editors are saying and deciding what's important. And the columnists they like, and that sort of thing. So there might almost all be entirely in that [Two-Step Flow] bucket, but I think everybody else gets a bit of both. I mean, my own experience is [that] I get information directly from politicians, but there's other things that are filtered through, and not necessarily filtered through in a sort of old newscaster kind of way. For instance, sort of an odd personal example, I have several friends who are very, very, very invested in the conflict in Israel with Hamas right now [Consult this article that provides background to the ongoing conflict between Israel and Hamas]. And I would say I get 80% of my information on what's happening on a daily basis through the Instagram Stories they post.
Hamish Marshall: [00:04:32] So that would be, I guess, classically, [Two-Step Flow]. It's a different kind of second step. It's not the editor of The Globe and Mail, it's my friend and what he thinks is interesting to post. But that is very much where I'm getting my, at least initial, information on the conflict. So I can see there's a transition maybe happening. And I think it's happening to some extent. But, the technology that has allowed that transition to happen has really been around in some way or another for 20 years, but behaviors are now only beginning to catch up with it.
Elizabeth Dubois: [00:05:00] That makes a lot of sense. And the idea of pitting these theories against each other is problematic because you're right, we all get information from a variety of different sources. And the conflict that you were describing, I think is a really good example, particularly because we know there are so few journalists in Gaza. One of the main ways that information about what's happening gets out is via Instagram right now, and it is less likely to be following, you know, here's my one news account that I follow on Instagram. The culture of using Instagram and other social media tools is usually to be following a whole bunch of different accounts, and the way you get information in those platforms involves piecing together a lot of different information.
Hamish Marshall: [00:05:45] And that's obviously curated by the people who are putting it on and whatever their agendas and biases are. But, the only thing I would disagree with you [on] is that I would argue that even getting information from a journalist is still really a Two-Step Flow. The idea that a journalist, especially in a conflict like that, is entirely unbiased and presenting just the facts is almost certainly untrue. You're getting something that's filtered through a journalist, a newsroom, an editorial, and a legal process which may or may not reflect things that happen on the ground, but not everything that happened on the ground, and certainly not the context, how one side or another sees things.
Elizabeth Dubois: [00:06:21] I think what you're bringing up is [an] important point. There are a ton of different layers of filtering that happen on all information.
Hamish Marshall: [00:06:30] Right.
Elizabeth Dubois: [00:06:30] Whether it's in a conflict or not. In a conflict, it's extra heightened and there's so much extra political and economic pressure. But across the board, any information we receive has been filtered by a variety of different sources. So one of the things that's really important about the Two-Step Flow hypothesis is that filtering level in the Two-Step Flow is a personal influence filter [Consult: Who to Trust on Social Media: How Opinion Leaders and Seekers Avoid Disinformation and Echo Chambers by Dubois, Minaeian, and Beaudry]. It's using social pressure and social support. The idea of 'you're going to want to change your opinion to agree with me', because we're in the same social circle and you're going to feel ostracized if you have a very different opinion
Hamish Marshall: [00:07:04] Right.
Elizabeth Dubois: [00:07:04] Or, all of my friends vote. What do you mean you're not going to vote? That's not cool. The peer pressure kind of approach. And then the idea with the One-Step Flow is we need that less or we're exposed to it less, and it's okay that we're exposed to it less because we can target using data and using this kind of changed media environment.
Hamish Marshall: [00:07:26] I think there's a lot of truth to that, both from politicians and politics. We've certainly been trying to do custom targeting to one extent or another, my entire time in politics with, I would say, a variety of levels of success.
Hamish Marshall: [00:07:41] The thing that's interesting to me now is that first with [Prime Minister Justin Trudeau] and now with [Pierre Poilievre] on slightly different platforms, the followings, the number of people following those leaders themselves and choosing to get information 'unfiltered' directly from the political leader is really significant,
Hamish Marshall: [00:08:02] I mean, Poilievre's [YouTube account], I don't know what the [follower] count is today, but the number even before he was leader, he had over a couple hundred thousand YouTube subscribers. Well, no [other] politician in Canada [has] anything else close to that.
Elizabeth Dubois: [00:08:14] Yeah.
Hamish Marshall: [00:08:14] And Trudeau now [has] 7 or 8 million fans on Facebook. I don't sure what that number was before he got elected, but it was a lot. Obviously he's gotten a whole bunch more since becoming Prime Minister, but it was a lot [before he was elected Prime Minister]. Even then, the people choosing to follow him on Facebook since he [became] Prime Minister are choosing to get information as [Trudeau] wants it presented. And the ability to go directly to non-trivial numbers of Canadians is what's really changed comparatively recently, despite best efforts for targeting and things earlier than that.
Elizabeth Dubois: [00:08:44] Yeah, that is really interesting. And it kind of fits with this idea of politician influencers?
Hamish Marshall: [00:08:51] Right.
Elizabeth Dubois: [00:08:51] We're getting used to social media influencers as being this active agent in our information environments on social media. And then to see these politicians borrowing similar tactics and techniques and then themselves becoming hyper-popular on the internet.
Hamish Marshall: [00:09:05] Yeah. And I think the Poilievre case is a really interesting one, because Poilievre was a Minister in the Harper government, didn't have a critic portfolio [in opposition] until 2017, or I can't remember what it was, but [the critic portfolio] wasn't very high profile. [Poilievre] became the Finance Critic in 2017, which he held through to 2020 or 2021 [Poilievre’s wikipedia page and Poilivre’s House of Commons page detail his roles in government and opposition]. And in that time he built up an audience that he didn't have when he started. I think Trudeau's growth was very popular in the media. He was getting a lot of coverage in the media, and he was also growing on social media at the same time [here is an article from one year into Trudeau’s first term as Prime Minister which discusses his social media use], which drove the other, who's to say? Probably a bit of both, right? They probably drove each other.
Elizabeth Dubois: [00:09:42] Mhm.
Hamish Marshall: [00:09:43] Whereas Poilievre, by the time he decides to run for leader after Erin O'Toole [the leader of Conservative Party of Canada from August 2020 to February 2022] resigns, he's more famous on the internet than he is in the public media. So in many ways, he's the first leader in Canada who's come at this with [an] almost social media or internet first approach. His source of strength, the people that he signed up [as members of the Conservative Party of Canada] [here is an article that discusses the 2022 Conservative Party of Canada leadership race, which Poilievre won, and the number of members of the Conservative Party of Canada] in that leadership race came not because he was famous and translated that into social media, but because he was big on social media. Then that translated into mainstream media and political attention and fame.
Elizabeth Dubois: [00:10:18] Yeah, that's [an] interesting transition to see how that works. And, you know, of course, Trudeau has been very popular on social media for a long time, but he also had this celebrity status before [Consult: Justin Trudeau and the play of celebrity in the 2015 Canadian federal election campaign], which is a very different set of dynamics, even if it ends up with both of them having a lot of followers.
Hamish Marshall: [00:10:34] And then there's the rise of other people. So I don't know if you would have come across him, there's a guy in [British Columbia] called Aaron Gunn.
Elizabeth Dubois: [00:10:41] Mhm.
Hamish Marshall: [00:10:41] [Aaron Gunn] is a filmmaker [and] Facebook personality. Aaron, the last time I checked, had about 50,000 fans on Facebook. [The] vast majority of them in British Columbia. [He] talks about [British Columbia] politics primarily [and] makes really interesting YouTube shorts and long videos and really grew [his brand] because of his social media following. [Aaron Gunn] was a social media influencer in politics that way, and with a huge following. [Former Premier of British Columbia] John Horgan had, I don't know, [25,000 to] 30,000 fans on Facebook. He was twice the reach of the guy who'd been Premier for 4 or 5 years at that point. Aaron's now the federal conservative candidate in the northern part of Vancouver Island [here is an article that discusses Aaron Gunn’s successful effort to become the candidate for the Conservative Party of Canada in the federal riding of North Island - Powell River]. He just won the nomination a few weeks ago. And his ability to win that nomination, his ability to [potentially] get himself [elected as a Member of Parliament] entirely stems from the social media activism.
Elizabeth Dubois: [00:11:30] That's super interesting. And there's examples internationally too. There's [a] Brazilian politician, for example, in São Paulo, who was very popular on Twitch and gaming on YouTube and still games on YouTube. I'm forgetting his name right now, but I'll add into the show notes and link to [his content] [Elizabeth is referring to Brazilian politician Kim Kataguiri. You can read more about him on the Kim Kataguiri Wikipedia page and look at Kataguiri’s Youtube content].
Hamish Marshall: [00:11:46] It's really cool.
Elizabeth Dubois: [00:11:47] It's really, really interesting to see. And this kind of builds off some of our earlier episodes in the season on the influencer industry [Consult previous Wonks and War Rooms episodes: Political Influencers with Nate Lubin, Social Media in Politics with Dave Sommer, and Influencers with Taylor Lorenz] and parasocial relationships [Consult: More Than Just a Tweet: The Unconscious Impact of Forming Parasocial Relationships Through Social Media]. So we won't rehash all of that right now.
Elizabeth Dubois: [00:12:00] But again, links [will be provided] in the show notes to the past episodes. Okay, so we've talked a little bit about the idea of campaigns, particularly leaders having this social media presence. What are the other kind of aspects of data-driven campaigning and ways that data and personalization are informing campaign practices? Can you walk us through the basics?
Hamish Marshall: [00:12:22] Going back to sort of the beginning, when we first started doing this, it was overwhelmingly, [and] in many ways still is, [the] ability to deliver direct mail. It's making a determination that certain groups of people are interested in certain things. That could be from knocking on the door and asking them what they're interested in [for more information on door knocking, also referred to as canvassing, you can consult this Wikipedia page on political canvassing], or it could be informed guesses based on the neighborhood they lived in. At the most basic level, it's [that] there's a hyper local issue in one neighborhood and people running for office. We're going to write a letter or postcard or something on that issue and send it to every home in that subdivision. And that has become more and more complicated as the amount of data the political parties have access to. Although the ability to do that in [Canada] is severely limited by Canada's privacy laws [for reference and additional reading on Canada’s privacy laws, is the 2022-2023 Annual Report to Parliament on the Privacy Act and the Personal Information Protection and Electronic Documents Act].
Elizabeth Dubois: [00:13:16] Mhm.
Hamish Marshall: [00:13:17] Talking to people I know in the [United States of America] where they can go and buy from credit bureau data [consult this article for more information on the use of data in political campaigns in US elections], they can go and buy and read [3,000 to] 5,000 data points on you: how old you are, what type of car you drive. Did you buy it outright? Did you borrow to lease it? Did you borrow for it? How much? How much [did you pay] for your house? Are you married? They have an unbelievable amount of individualized, personalized data. And in Canada, our ability to get that data because of our privacy laws, is basically nonexistent. And certainly not at the scale [of other jurisdictions like the United States of America].
Elizabeth Dubois: [00:13:49] Mhm.
Hamish Marshall: [00:13:49] That severely limits what we can do. The other interesting limiter in this, [and] we can make a lot of informed guesses. Or when you're doing online advertising, you can target based on age, gender and geography effectively, and you can build lookalike audiences based on data you collected. But, our ability in most political campaigns to be able to go and say, I want to reach every Canadian who's, you know, health care is their top issue and run YouTube ads to them or whatever is really extremely difficult.
Elizabeth Dubois: [00:14:23] Because it's hard to identify the people who would resonate with those advertisements, or because you can't get all of the people you think you're trying to get?
Hamish Marshall: [00:14:32] You can make inferences from polling. You can say, we know people over 55 or more concerned about health care, so let's target health care ads [at] people over 55. You can do that. But you don't have all the data points like political parties do in the United States, who can say, we know that this person who is spending money on this type of product is probably more likely to have these sort of health concerns or has health concerns in general. Therefore we can advertise directly to them. And the ability to do super targeted, individualized targeting is starting to come in with some of the connected TV options. Now that there [is] ads on things like Amazon Prime or a variety of other things that are not conventional TV, that you can advertise quite interesting things through. There is some of this beginning to appear, but it's really not as developed as people like to think it is. There's this perception out there that there's some of us sitting in a war room with a big computer saying, 'target segment 12 and do this'.
Elizabeth Dubois: [00:15:36] Hahaha.
Hamish Marshall: [00:15:37] As much as I would love that to be true, the instruments are a lot blunter than I think the public would perceive.
Elizabeth Dubois: [00:15:42] Yeah, there's some really interesting research that Kate Dommett, who's a professor in Sheffield in England, has done with some colleagues that looked at data driven campaigning around the world [Elizabeth is referencing Just what is data-driven campaigning? A systematic review by Dommet, Barclay, and Gibson]. Most of the studies have only really been done on Canada, the US and Western Europe. So around the world actually means the West [understanding of “the West” has been derived from The Clash of Civilizations by Huntington]. What [Kate Dommet] and her colleagues found was privacy legislation and data restrictions drastically impact how campaigns use data, and what digital and data-informed campaigning actually looks like. So what I'm hearing from you is Canada is not really as targeted as the US context. I would gather from my knowledge of privacy legislation in Europe that maybe we have a little bit more access to data than some of the European counterparts.
Hamish Marshall: [00:16:34] One end is Germany, right? Where the German privacy laws are [you can read about German privacy laws and political campaigns in this article on how German political parties have been accused of breaching data protection rules. Also, here is a link to an article about the German government's support of banning the use of personal data in political campaigning]. I [was] talking to some campaigners in Germany. They have the voters list, [and] I might be overstating this [as] it's been a few years, but they have the voters list they get from the elections authority, and then they have their list of results they get from door knocking saying, we're supporting you or SDP or whoever it is. They can't actually merge those.
Hamish Marshall: [00:16:54] They can say, I knocked on your door and Elizabeth says she's not voting for us so I can keep that down [on a] separate list, but they can't actually go and connect that to your record where it says 'lived at this house for this number of years'. So their ability to target is almost none. It's basically, [looking] at past election results and [deciding] which neighborhoods you're going to go after. It's very, very, very restrictive. The UK allows more than Canada, but not nearly as much as the US [learn more about the UK’s guidelines for the use of personal data in political campaigns] . So that has a huge, huge, huge impact on it. And then the tools, most of the tools that are developed for, especially online targeting, are developed not for political advertisers, but for commercial advertisers, naturally, because that's who pays Google and Facebook and everyone else all the money. So you're stuck with tools that are designed to do something different than what you want them to do [and] you have to make do.
Elizabeth Dubois: [00:17:44] That makes sense. And I think it's probably worth noting that privacy legislation [is] important. They help protect a lot of people.
Hamish Marshall: [00:17:53] 100%. Absolutely.
Elizabeth Dubois: [00:17:54] It's really tricky when we don't know how the technology is going to evolve next. What are your thoughts in terms of what it's going to look like for campaigning, for political campaigning? You're talking about using these tools that haven't been developed for politics, but you shoehorn them in. We also know that big tech companies are constantly evolving their approaches, and the way we consume content is also evolving. Do you have guesses at what we should expect next from this data-driven approach to campaigns?
Hamish Marshall: [00:18:28] I think we'll continue to see a refinement of online advertising. The big thing that has been promised that hasn't come true thus far, and perhaps will, is the really good connection between real-life connections and digital. In the 2015 British election, both the Tories and Labor bragged [that there were] all these articles about how they said they went and knocked on your door and you said you cared about health care [and they would] have health care Facebook ads running to you within 48 hours or something [Consult this journal article about the political marketing strategies used in the 2015 UK election here], which sounds amazing and incredible and is almost certainly untrue. But they certainly had ads running. Whether they were reaching the same people or not is another story entirely. And, we tried to do that in the 2019 campaign where we would be collecting information at the door and then said, Facebook said, if you have someone's address, we can build advertising lists based on someone's physical address and there'll be like a 65% matching.
Hamish Marshall: [00:19:26] So if we go door knocking and we find 10,000 people one day across the country who care about tax cuts but say they're undecided, we can start running ads to those people about Andrew Scheer's [Andrew Scheer was the leader of the Conservative Party of Canada from May 2017 to August 2020] tax cut plan or whatever it was. And this is going to be great. We can hit them directly. We've got good address data because we've got Elections Canada data like this is good high-quality address data [consult: Elections Canada’s collection of personal information and data privacy practices]. We found in practice that the matching rate was well under 10%.
Elizabeth Dubois: [00:19:50] Wow.
Hamish Marshall: [00:19:51] So yes, in theory, we knocked on your door and you said you cared about issue X and Y. And we can go and talk about issue X and Y. But in practice, the number of people we were actually effectively able to deliver that to was very, very minuscule. And maybe, [there are] things we could have done that we should have started six months earlier and done a whole variety of things to improve that matching number. But I really doubt that the matching number that everyone was telling us was 65% was really ever achievable.
Hamish Marshall: [00:20:20] There [were] other things in that campaign we did where we tried to advertise to people based on the type of newspaper articles they were reading. There's different advertising networks out there that will say, we know people are interested in the way they pitch. It is okay. They're reading car reviews, they're interested in buying a car. So, you know, advertise to people who have shown an interest in buying it or interest in cars. Great. We were trying to do that, and it really broke down in our ability to actually create meaningful-sized audiences was very, very difficult. And this is something that, you know, years and years and years ago, the first campaign that did a whole bunch of 'microtargeting' was Bush's campaign in 2004 [consult this link to learn more about microtargeting practices in US elections]. It was the first one that the media really talked about. A few years after that, I talked to a professor at some university in Texas who was the guy behind that targeting. It was [an] interesting conversation because I was expecting him to be [say that] “targeting is the best, I've built my career on it, we did this amazing stuff, we got the president reelected”, [instead] he was [saying that] there's massive limits to this. The limits to targeting are very real. And he says, you can go through and you can cut up the electorate in 35 different ways and say, we need segment A and segment G and this and this thing, and we're going to put it together. And what you find is when you put all these segments together, it never adds up to the number you want it to.
Hamish Marshall: [00:21:45] Say we need 100,000 votes in this county or whatever it is, and you go through and you say, okay, we've got this type of Republican, this type of Independent, and you're putting it all together. It never comes to that number. You always just need a percentage of people that you can't really define [or] put in a target box. I've always kept that in mind, I've built my career at this intersection of politics and data. But I'm also intensely aware of the limitations of it. If you try to run a campaign that's solely based on, 'we're going to talk to these three segments and ignore everybody else', it's almost always going to fall apart.
Elizabeth Dubois: [00:22:20] That's such an interesting observation, and it's so interesting that 20 years later, it's still the case even as technology has evolved. And that makes me think some of the people who are the prime target voters, the undecideds who are still likely to go vote, maybe they're the ones we don't have the most data about.
Hamish Marshall: [00:22:42] 100% correct. I'm a public opinion researcher. I do polling all the time, and the number of people who say they're undecided in a poll who also are more likely to say, ["I don't know"], to 35 other questions. We have this idea of undecideds [thinking] 'I spent a lot of time thinking about politics, and I thought about this leader and this leader, and I'm weighing the pros and cons, and I haven't decided to vote' [consult this article that discusses, and interviews, undecided voters from the 2021 Canadian federal election]. Statistically, there are people like that, but overwhelmingly undecided voters are people who are not paying a lot of attention, [and who] maybe end up caring about 1 or 2 issues. But when you are much more likely to say, ["I don't know"] on a whole bunch of other questions all the way through a survey,
Hamish Marshall: [00:23:17] [Undecideds are] less likely to be engaged because if they were super engaged, they're more likely to have an opinion on who they're voting for. Right?
Elizabeth Dubois: [00:23:24] Totally.
Hamish Marshall: [00:23:24] So, you're absolutely right. And there are many ways that true undecided voters are the hardest to connect with. There's this ongoing debate about [contacting] people by phone for polling or [contacting them] online using panels, and that's another whole podcast, public polling methodology. But, everybody says, the number one question I get when I say we're gonna do this poll online is, do you get enough old people? Yeah, yeah, old people are online. The group that's hardest to reach, both on phone polls and online polls, is men under 35. Like brutally difficult to get them to give an opinion.Just unbelievably hard. Women under 35 are difficult, but men under 35 are sometimes nigh on impossible, and they're largely just very disengaged from political discourse. When you do get them, the chances [are] that they're saying, "don't know", "don't know", "don't know.” There are obviously exceptions to that [trend], but [it] skews that way.
Elizabeth Dubois: [00:24:23] So what does a campaign do then? If these kind of online polls or phone polls and data-driven approaches are not doing it? What does a campaign do?
Hamish Marshall: [00:24:33] Anybody who comes in on a campaign, and whether you or any of the listeners or your students or whomever it is, is ever working in a campaign, and someone comes in and says, "I've got this data product that will solve all your problems", they are lying. There is no one product. My approach has always been to take a very multi-stream approach
Hamish Marshall: [00:24:54] You got this product [that] is [doing] cool data targeting? Great. We'll do that. We'll also do this other one. We'll also do this other one. We'll also do traditional door-knocking and we'll do everything else. We'll do it all because that's the only way to get a half-decent picture of what's actually happening. In the 2019 campaign, we had a sort of rough predictive analytics system. We had [an] analysis of the public polls that was going on. We had our tracking polls that were going on. We had individual riding polls that were coming in. We had door-knocking returns analyzed in a certain way. We had social media feedback. We had all these different things happening in order to give us a picture that was understanding and then [informing] who we should target, and then we would go and push out through all of those channels an ability to say, we're going to go target these folks. We think, you know, Group X or Y or people who are receptive to this kind of message. So let's go after them. And the interesting thing is that you can do lots of data stuff to reach people online, but I'm more and more convinced of the importance and efficacy of door-knocking,
Hamish Marshall: [00:25:54] Turning up at someone's door and having a conversation with them. Now, which door is you? Door knocking can be informed by data and everything else but [a] conversation with an informed and passionate volunteer who's got enthusiasm for the candidate who's supporting is massively persuasive.
Elizabeth Dubois: [00:26:09] And typically door knocking happens within either your own community or communities kind of nearby. So even if you don't know this person personally, there's a sense of [similarity].
Hamish Marshall: [00:26:23] I also think it's the ultimate personalization. We do all this stuff to talk about personalization, which is right. And we try to narrowcast [Consult: The Mind-Set to Share: An Exploration of Antecedents of Narrowcasting Versus Broadcasting in Digital Advertising to read about the difference between broadcasting and narrowcasting] these messages. But if I'm a half-decent door knocker and I'm having a conversation with you at your door, I'm going to respond to the things you say, I'm going to respond to your body language, I'm going to respond when you say "I care about issue X or Y", and then we're going to have a conversation about that. If you say, my kid goes to school down the street and it's kind of lousy, let's have a conversation about education. It's in many ways the ultimate personalization. And it's interesting that sometimes the oldest technology is actually the most effective. It's the one-on-one conversation.
Elizabeth Dubois: [00:26:59] That's really fascinating, and I appreciate that you were able to kind of lay out all of these different strategies and then also show where door-knocking comes in and is still super important. It makes me think of relational campaigning, which became more popular, particularly during the pandemic when we were in the stage of don't go outside, don't talk to strangers. And there was a lot of emphasis on [the idea of if we] can get campaign volunteers to send text messages or social media posts to their friends and family. Can we get them to email predetermined talking points, essentially key messages? What are your thoughts on that sort of tech-enabled, personal influence replacement for door-knocking?
Hamish Marshall: [00:27:43] I mean, I think that's interesting. I've never seen it done really well at scale.
Hamish Marshall: [00:27:47] The technology in order to get to do it is pretty straightforward. The issue is the willingness of somebody to go and text everybody in their [contact list].
Hamish Marshall: [00:27:55] And they're going through the list [and thinking] well, "[I'll] text Jim, but not Bob. What about Michael? Sarah? I don't want to bug my cousin about politics". People end up self-editing. If people want to broadcast something, there's a personality issue here. One of the reasons why, in my view, Twitter has never grown beyond basically being more or less the same size for quite some time [is that] there's only a certain percentage of the population who wants to put their thoughts out there for everybody, right? I'm not on Twitter, [and I have] no interest in ever being on Twitter. I [have] Facebook to show pictures of my kids and talk about things that are personal things in my life. The idea that the general public can interact with that is abhorrent. Now, I'm pretty political. Most [of] my friends are political, but even then I put very little political stuff on my personal feeds. The [idea] that I would email my dad's old friend and say, "have you considered voting conservative" is very not me and a lot of people would personally never do that. There's a limit to that. The one thing that has been done effectively and some campaigns do it and it's an old technique is FRAN letters (Friends, Relatives, Acquaintances and Neighbors), but not in the traditional sense let's give a list of everybody in your address book, and we'll send them a letter, because most people don't have address books. [Another] problem is it doesn't really exist in writing, how many people live in my riding that I know that I connect with regularly. I guess some parents that go to my kid's school, But, not a lot, right? And everybody else's networks are strewn across the country or across the world, so it's more limited. So the FRAN letter, which I've used successfully in a bunch of campaigns, especially local campaigns, is you have somebody who is willing to take a letter and you say "I live on Main Street and I'm going to take everybody who lives within three blocks of me on Main Street, and I'm going to write a letter saying I'm Hamish and I live on Main Street and I really care about this in our community, and that's why I'm voting for this candidate". Then the campaign goes and prints 75 [or however many copies of it], and goes and drops them in every house for 3 or 4 blocks. Suddenly it's not a letter from a candidate, it's someone who's got a life. To your point earlier about someone who's got a life that looks a lot like yours. Talking about a connection now, I think that's really good, especially for local campaigns. I don't think that works necessarily as well at federal, maybe provincial elections. But if, especially something like a city council election where you might not have strong views about your councilor, if someone's saying, "I'm supporting our councilor to get re-elected because she helped get my garbage issue fixed" well, my neighbor said that, you know, that's great news to hear, right? So that kind of personalization can be done. But the text everybody you know, and tell them to vote this way or that, I've yet to see that done well.
Elizabeth Dubois: [00:30:30] That's interesting. That's such a wonderful, tangible example, and I can totally see how that could fit into the larger strategies. You're using a whole bunch of data about what neighborhoods are more likely to be winnable. And then you figure out which are the people where we've got volunteers active already. And so you use this data-informed approach to target who you're going to then use this FRAN letter approach, which is much more kind of old school, personal social influence style.
Hamish Marshall: [00:31:01] Folksy? Totally.
Elizabeth Dubois: [00:31:02] Yeah. That's fascinating. Well, this has been a fantastic conversation, but we are running out of time. Um, so the very last question is a little pop quiz. Hopefully, you've been properly engaged for our whole chat.
Hamish Marshall: [00:31:16] You're going up? Yeah.
Elizabeth Dubois: [00:31:18] Short answer, [how would you] describe to someone else what is the One-step Flow?
Hamish Marshall: [00:31:24] The One-Step Flow is basically communications flowing directly from the source or a media outlet to the consumer, not mediated through a local or a community or a neighborhood or a personal influencer or opinion.
Elizabeth Dubois: [00:31:39] Yeah, exactly. In our current media environment, it's usually very data-informed, [and] usually involves a lot of personalization approaches that are relying on personal data.
Hamish Marshall: [00:31:53] Right.
Elizabeth Dubois: [00:31:53] Awesome! Well, thank you so much for your time. I've really appreciated chatting with you.
Hamish Marshall: [00:31:57] My pleasure.
Elizabeth Dubois: [00:32:03] That’s our episode on the one-step flow, on data personalization [and] what it looks like for campaigns. I hope you enjoyed it. If you’d like to learn more about any of the topics we’ve talked about today, you can check the show notes and look at the transcripts that are available in English and French - we annotate them with links to a whole bunch of different resources. Head over to polcommtech.com for more.
[00:32:27] This season of Wonks and War Rooms is supported in part by the University of Ottawa's University Research Chair in Politics, Communication and Technology. I also want to acknowledge that I am recording from the traditional and unceded territory of the Algonquin people, and I want to pay respect to the Algonquin people, acknowledging their long standing relationship with this unceded territory.