July 17, 2017
In the Mood for Marketing
Targeted marketing works when it is done in the right place and at the right time. Getting the place right is increasingly feasible with today’s location-aware technology. But getting the time right depends on knowing when a person is in the mood to receive the message.
That’s the goal behind the Centiment, a machine learning startup that applies AI to analyzing the expression of emotion “to make advertising more ethical and efficient for everyone.” It does that by creating what it calls “thought-driven AI and Google for feelings.”
In a phone interview, Micah Brown, the company’s CEO and founder, explained how this approach is a game-changer that makes ads more effective by taking cues from “moods rather than price and brand.” The ads are designed to fit with how the individuals in the target market “are feeling at the current time rather than what you think is best from them.”
The mood assessment offers two strategic advantages for marketers. One is that it allows them to deliver ads to people while they are in that receptive mood. The other is that they can see the feelings particular ads generate in their target market.
Centiment’s Emotional Search Engine takes a read on expressions of feeling and words used in connection to trending topics, cities, and other subjects over the previous month. That can be particularly useful for those who want insight into how people are reacting to campaigns, including:
- Marketing campaign managers looking to assess the impact of paid media efforts
- PR and marketing departments to quickly assess mood of customers
- Content marketing teams looking for sentiment overview for trending news
- Creative ad managers looking to benchmark campaign creative against competitors using social sentiment analysis
It can also take the emotional temperature by location or context in real-time to direct marketers where and when to target ads effectively.
Centiment already boasts a number of case studies for businesses that benefited from its insights in a variety of industries. Here’s a video of Centiment in Action:
I asked Brown how Centiment developed the AI required for the sentiment analysis. He credited its partnership with IBM Watson with granting them the resources that got them “50% (through) the journey.” The remaining half, building “the emotional component took about two years” — and involved a number of people with advanced expertise in data science and psychology.
Centiment also got a boost from being in Sprint’s Corporate Accelerator program. As a telecom provider, Sprint was able to provide data about people’s smartphone usage that gives a picture of individuals in particular places. Unless users opt out, Brown explained, every time they use the phone, they are sharing data about their usage. While the data Centiment gets is anonymized, they can get the figures for a location that they can “cross reference with Twitter.”
The correspondence of tweets and location can reveal segmented insight into sentiment. That Twitter data can also surface if a specific individual falls into “the same general frame” or expresses a different sentiment than the group.
For users not on Sprint, access to Facebook can provide data on user sentiment. Brown said that Instagram and Snapchat are more difficult for them to work with, but they “are looking into incorporating” those channel in the future.
What’s on the agenda for the immediate future is setting up offices in Kansas where it will build on its previous sentiment analysis work as it collaborates with the University of Kansas Medical Center on a neuroscience study that seeks to uncover “the links between human thought and sentiment analysis.” In addition to contributing to science, Centiment expects the study to reveal valuable marketing insights.