AI Is Changing The Game During The 2020 Election Cycle

In the US, voters are still waiting for the results of the 2020 presidential election. With social distancing and early voting, candidates are using the full power of social media to mobilize their voters. Powered by conversational AI, Amplify.ai is changing the game during this election cycle.

Since 2017, the company has helped large brands and influencers stand out in social media’s crowded landscape by providing genuine, two-way, conversational engagement. The company applied proven technology to help many 2020 political campaigns, including five presidential campaigns, seven senate campaigns, three national political organizations, and several 501c’s engage voters. Along the way, they’ve learned a few lessons and found a way to help temper the polarization in social media. 

Mahi de Silva, CEO of Amplify.ai says, “In 2019, in our work with some of the biggest digital publishers in India, we came to realize that the political parties were big advertisers. It was the publishers that got us into the mix. And we started working with candidates for Prime Minister in the national election in India. It opened up our eyes on how we could bring conversational AI, to drive voter engagement.”

Personalized Outreach

Social media management can be cumbersome for large brands and companies. Due to the massive amounts of reactions and responses they receive from their posts, companies often outsource efforts to respond to these comments.  Amplify.ai uses its state-of-the art NLP, or natural language processing engine to scale out the engagement in a way that eliminates the grunt work of managing responses.

During this election cycle, when candidates cannot hold as many rallies and personalized outreach seems impossible while social distancing, social media outreach became a must. Digital outreach through using social media platforms can be engaging and personal. 

For example, there were typically thousands of comments and reactions per every social media post for any candidate’s primary campaign. These comments and reactions can only be valuable if they are analyzed, and responded to.

Amplify.ai uses its NLP engine to respond to the comments and analyze the comments and further engagement reactions. In this 2020 election cycle, their new feature opens up a “relevant” conversation with each engagement. Through sentiment analysis, the NLP engine determines whether the person is most likely to donate, find out more about the candidate, find out about policy, or help with outreach. Each subsequent conversation opened can be around a topic that engages the participant.

De Silva says, “In the primary, with our first U.S. political client, and a real digital pioneer was getting thousands of comments and reactions on places like Facebook and Instagram, far more than any social team could directly respond to. Typically, how we engage with a political campaign is we give them a dashboard and say, look, here’s what’s going on. Let’s develop a strategy of how you want to engage your audience. You create a retargeting campaign. You target just those users and try to engage them. When you have  25,000 people that are sending you comments, again, with our platform, we can take those comments. We can analyze them, and we can automatically respond to them.”

Tempering Polarization

In the past few years, social media platforms have been blamed for the increased polarization during election cycles. Each platform has its technology teams to use the power of AI to track hate speech and analyze negative feedback on social media posts. Amplify.ai is using their technology to go one step further in tempering the polarization on social media platforms.

AI is different from human intelligence. Speech that generates more polarization tends to be emotional or incite emotional feedback. By opening up conversations in an emotionally neutral way and sending responses that are more focused on actual action items such as “voting,” “campaigning,” “donation,” or “policy,” etc., Amplify.ai can help to steer overly emotional reactions to a more emotionally neutral baseline.

When you put people from both sides of the political spectrum in the same room, you will often find that they may disagree on some political issues, but they can usually find common ground. Some Democrats have the same family values as Republicans. They send their kids to the same types of schools, etc. They live in similar neighborhoods and shop at the same supermarket. They may be concerned about the same issues locally.

Conversations that start at the candidate level can help to remind everyone of their common humanity.

De Silva says, “In our previous company, we ran a lot of ads for Obama’s 2012 campaign. President Obama really inspired a lot of people in Silicon Valley to get involved. These were super-targeted ads, ad sets that were probably seen by 50 people in Ohio, micro-targeted with a message around what was going to be meaningful to them.” 

With conversations around micro-targeted ads that center around the common thread of humanity, Amplify.ai can help to temper some of the polarization generated from online political campaigns. At the same time, they can help candidates reach more people across both political spectrums.

Consent and Privacy

With micro-targeting and personalization, there’s always talk around privacy. Before beginning a conversation, Amplify.ai delivers simplified UX design to alert the participant to opt into the conversation if they want to participate. 

Instead of opt-in and opt-out using a privacy document or a feature list under an application’s settings, using ethical UX design practices to deliver easy to understand privacy features embedded in actual UX functions to maximize consent is a new trend in AI-enabled applications development.

Developing Better Policies and Relationships With Voters

Campaigns at the national levels can leave constituents in each locale feeling a bit left out. The consensus is that local issues are often not a part of the national conversation. Translating policy stands in terms of local issues can help constituents improve their understanding of the candidates’ policy stands. Feedbacks from these conversations can help candidates understand the policy issues better to help serve more members of their constituents.

De Silva says, “So if you’re a construction worker in Mississippi, you have a very different set of needs than let’s say, a high-tech worker in Portland. We generate an issue word cloud for these campaigns every day, and our team reviews them. It shapes the campaign that runs that day; again, we’re talking about tens of thousands of ad sets because they are hyper-targeted, but these teams are down at the state level. They can shape the messaging around that. We also use peer to peer persuasion.”

Local social media influencers, volunteers, and supporters with their own social media capabilities can amplify messaging using their own social media accounts.

De Silva says, “So, when you have very passionate people, that have expressed an interest in being a volunteer, we can actually give them capabilities to post their own persuasion, on Facebook and Instagram, for example, or Twitter, and then invite their friends into the conversation. It’s not necessarily just driven by the candidate and the issues of the candidate. It actually can be taken to that next level of persuasion, for, again, using the power of social media.”

By participating in these conversations, people feel more vested in the dialogues, in the policy stands, and in the political candidates they support. Simultaneously, when millions of people’s engagement are analyzed, the campaigns can understand what their constituents genuinely care about.

The granularity of these micro-targeted conversations that are generated on policy issues goes beyond polls. It can help candidates to drive policies that are truly important to their constituents.

De Silva says, “You look at some of the most highly regarded polls, they’re touching maybe 2000 people across the country, we’re touching millions of people. The real-time feedback we get is, just from a statistics standpoint, so much more relevant way of taking a pulse of the electorate and really understanding what messaging people are responding to.”

Machine learning can remember the history of every engagement. For candidates interested in seeing the record of what a constituent cared about versus what they care about today, this can be very powerful. A conversation can also be resumed at different levels of the re-targeting campaign.

De Silva says, “One of the things we’ve learned is that if you can make that conversation stateful and persistent, because what’s really frustrating to a lot of users is that, when they engage in like a human agent chat, it’s very ephemeral, you go to a website, you have a chat, and sometimes the agent is, chatting with five other people.”

De Silva says, “We’re working with a company in Europe that is in the financial services space. With the recent government directives, there are public APIs to banks to let people understand their loan terms and to help shop for refinancing for lower interest rates. This company can have a call to action to make it easy for consumers to refinance. In a conversation about refinancing, you need sticky, persistent conversation so that participants can exit and resume the conversation when they have all the information.”

Truth and Authenticity

Every day, with social media influencers advocating for authenticity, political candidates also realize the value of truth and authenticity in their relationships with their constituents. They know that their branding depends on the strength of the relationship with the members of their constituents. One of the best ways to build a strong relationship is through truthful and valuable conversations with their constituents.

De Silva says, “There’s a professor here at Stanford, Dr. Jeff Hancock, who’s written extensively about this concept of emotional contagion on social media. The basic premise is that if you see overly negative things about a brand or a candidate, you are much more likely as a social media user to pile on. But, if you can tamp down that negative sentiment, you can get a much more even keel response.”

Amplify.ai quickly realized that valuable information that reflects the constituent’s real sentiments, their actual policy stands, what they think about their candidate’s policy stands, etc. comes from even keel-ed responses.

The right way to ask a given question often will generate more “truthful” answers. By asking the right questions and receiving responses, Amplify.ai can evaluate whether the respondents are dependable. The population targeted can be refined throughout the campaign. The analytics from ad spend that the marketing team relies on can consistently become more accurate in the process.

De Silva says, “Machine Learning-derived sentiment analysis can help refine your audience data to find specific audiences for your next campaign, we basically say, let’s take all of the people that reacted negatively, whether it’s a comment or a reaction and target them. What happens is that the efficacy of your advertising goes up by about 20% because you didn’t target those people that are not the people you want to engage.”

De Silva says, “If we want to ask them a location question, instead of saying, ‘what state do you live in’, you asked a Trump supporter, ‘what state do you call home in this great nation?’ Again, we saw, 50% better response rate…The beauty of ML is that it lets you find the most productive path without having to spend hours pouring over raw data.”

Scaling Up

Scaling up is another reason that companies with hundreds of brands can consolidate their social media outreach efforts under one umbrella using Amplify.ai.

De Silva says, “We now give our customers a map of their customer journey, it’s not just a pretty picture, but this is an active tool that you can look at what is happening with the customer journey, where are they coming in? What means of demand generation is working best? If you’re using influencers, we have a beauty brand with 700 influencers, but there are about 15 of them that really matters. How do you instrument that? How do you reward them?”

Recently, Amplify.ai has just surpassed 10 billion engagements on its platform. With increased business from political campaigns, multi-national conglomerates with hundreds of brands, it is poised to drive the use of artificial intelligence on social media platforms.

De Silva says, “A typical enterprise uses platforms like Salesforce, Zendesk, Shopify, but we have a product that rides above all of that but gives you the visibility into the intertwining of those systems. We are very excited about delivering those AI-powered tools that help the human teams inside businesses operate more efficiently.”