Support and Reputation Management Campaign
This post is part of a series of guides demonstrating how to create CrowdValence campaigns for various use cases. In this guide, we’ll walk through a sample campaign setup to notify the user of posts mentioning a topic.
Here, we’ll use Tesla as our example brand. In this case, Tesla’s goal is to get notified of posts from customers or potential customers that mention issues with their cars for their support team to address. Because many posts are quite long, CrowdValence will also summarize them. Finally, we’ll ask CrowdValence to suggest response ideas for Tesla’s support team where relevant. The end result will be a campaign that finds posts mentioning the topic, then summarizes and sends them to us if they are relevant.
Below, we can see sample posts found by this filter, and how CrowdValence has categorized, summarized, and suggested response ideas for each of them.
We’ll start by going over the basic filtering options most useful for this use case, then go into our more advanced filters.
Basic Filters
Keywords
To start, we’ll simply add Tesla. This will monitor any posts containing the keyword, whether in their title, self-text, comments, or even as text in an image (like a customer posting a screenshot of an email from Tesla).
We might also want to add specific model names like “Model Y” or “Model X” for posts that don’t mention Tesla specifically. You’ll notice these multiword keywords have quotes around them. This ensures that we search for posts that contain the full phrase “Model Y” and not simply Model or Y.
Negative Keywords
In Tesla’s case, many posts containing the keyword “Tesla” are not relevant to monitoring customer feedback and brand perception. For example, there is a high volume of posts discussing Tesla’s stock, posts related to the inventor Nikola Tesla and other uses of the word less relevant to this use case.
To filter out these lower hanging fruit, we’ll add some negative keywords, which will filter out posts that contain them.
Subreddits
For now, we’ll leave subreddits as “all” so we can track posts form every subreddit, as there are a large number of different subreddits where Tesla cars are discussed. If we decide to narrow our scope later, we can always edit the list to monitor only specific subreddits.
Contextual Filtering
We’ll also describe to CrowdValence the exact kinds of posts we want to be notified of. CrowdValence will then evaluate each post to see how well it fits this description, and assign it a Relevance Score. This will help us quickly discern which posts are more or less relevant to our use case.
Additionally, we can use this Relevance Score to filter out posts below a certain threshold. For example, we can specify to only notify us of posts that are above 50% relevance if we notice most posts below that score aren’t relevant to us. We recommend doing this only once you’ve started receiving posts, to ensure that you don’t set the filtering threshold too low.
Suggesting Responses
While many users simply want to be notified of relevant posts, for this example we’ll also show you how to get CrowdValence to suggest response ideas.
Because CrowdValence has already read through and understood each post, I’ll also ask it to suggest response ideas where it would recommend responding based on the post’s content. A quick example is below, where we describe the kind of ideas we want, the persona we would be replying as, and for what kinds of posts we want those ideas generated.
