How does Mediarails Collect Social Data?

At Mediarails, we use a meta-service approach. This means that for many of our data discovery services, we use different and redundant approaches to get data. These services are both first-party  (built by Mediarails) or we integrate with third parties. We do this for many reasons:

• Coverage - We want to make sure that our customers can get the data they need in as many scenarios as possible
• Quality - We always strive to have high-quality data.
• Resiliency - Getting data at a scale level can be tricky. For example, Instagram might change the API or access (like it has happened in the past) and we want to make sure we have multiple ways to get data
• Velocity - We refresh data at different rates and different volumes
• Cost - We continually look for approaches to scale data and keep our solution affordable for our customers.

In the case of social media data and audience, we use a blend of first and third party services. For example:

• Audience data (followers, engagement rate, etc.) is updated weekly using proprietary IP
• Audience demographic is updated monthly using three different audience data providers
• Fake follower data in the Influencer Lookup is pulled when a search is enabled by the user, using a third party provider that analyzes an influencer's follower base for red flags such as odd usernames, no posts, and followers who have no followers themselves.
• Contact information is acquired using both proprietary services that use social media network APIs to analyze the profiles (for example, an influencer on Facebook might have the email in the "about me", on Instagram in the Bio, and in their blog under the "Contact us").
• Placement statistics rely on proprietary IP. We monitor the profile using the networks' APIs, detect the placements, and monitor their performance for seven days.

We are currently working on the ability to connect all social media handles and access the influencer's analytics. This approach will give us the ability to complement our current placement statistics. We'll not only be able to pull in additional details, but also get access to stories.