By Chris Lundberg
As we’ve been going through dozens of demos and presentations, we’ve been heartened by the response. It’s been fun to see people’s jaws drop and the look of excitement as Frakture promises to solve real, hard problems that both organizations and companies are facing today. There’s also been some amount of healthy skepticism. We talk a lot about the Bots automating messaging, targeting, and reporting — and we show the process of working with the Bots and getting the final products.
In the words of a real person last week, we were asked, “Seriously?! Do you have magic ponies? Please tell me how this actually works!”
Since I’m sure others have the same question, and perhaps skepticism, I’m going to attempt to peel back the (virtual) carbon-composite exoskeleton just a little and describe how we tackle the problems facing marketers and organizers everywhere.
First, some terminology. At its core, all marketing can be boiled down to small numbers of people communicating effectively with large numbers of people — our terminology is based around this.
We think of every individual in the world as a (conveniently named) person. Not an email address, or IP, or other possible dupe. A significant number of our Bots are devoted merely to the task of identifying a person from a bunch of different possible sources. Sometimes we have email addresses from a CRM, sometimes a unique identifier from a database firm, sometimes a Twitter handle or Facebook ID. Sometimes we can’t uniquely identify a person, but this is rarely an impediment to many ways of communicating. Deduplication, data appends, and fuzzy matching are all techniques we use to ID someone.
A segment is any set of people. These could be based on geographic boundaries (people in DC or CA), demographic data (men or women), how you first came in touch with them (from an event, a list serve, a signup form, etc), based on activity (email clickers), or some combination (people likely to be watching NBC at 3AM Sunday morning). These are all segments, and a number of Bots are designed to merge segments, slice ‘em and dice ‘em.
Any action that happens at a specific time we call an interaction. This could be an email sent, an ad clicked, an event attended, a payment or donation made — literally anything that is tied to a specific time. Most interactions are also tied to a person, but some are not (Number of Facebook likes, etc).
A code is any set of interactions – all clicks from this one online add, all payments or donations from this one direct mail piece, all signups from this landing page, etc. Codes provide a much easier way to deal with the millions of interactions happening with most organizations. In different systems they are similarly called tracking codes, tags, or any number of other labels for grouping interactions together.
A message is any unique set of content sent out. One version of an email test, one online ad, one YouTube preroll video, one play of a radio or tv spot, etc.
A channel is a way of distributing a message to people. Facebook, Email, TV, online ads, Twitter, Direct Mail, etc. are all different channels. They often use different terminology, which is where the Channel Bots and Chatter Bots come into play.
The Mapping Process
Most of the work on the Bots goes into designing and creating Bots for each channel. These Bots map specific concepts into people, segments, interactions and codes. For example, for Facebook, each target audience (state, gender, custom audiences) is mapped to a Segment, each Facebook Ad is mapped to a message, with a price as an interaction, and each resulting click is also an interaction. For email, each send, open, and click is an interaction, each email address maps to one person, and each version of the email is a message. Some are harder than others, but the concepts hold very true no matter what channel it is.
On the sending side, the Channel Bots will translate “Publish a Message” into “Post on Facebook”, “Send Email” or “Tweet”.
Reporting & Analytics
Once data has been mapped and imported into these data types, reports put it all back together. Now numbers like Cost Per Click become “Count of Facebook Click Interactions on Message A”/“Cost Interaction for Facebook Ad A” … or more generically “Message A Click Code.Count”/“Message A Cost Code.Sum”. Reports can basically include any ratio or combination of interactions, and thus are extremely flexible.
Lots of work has gone into the details and testing of the Bots so that it *seems* like magic. But at its core, the Channel Bots and Chatter Bots translate information from or to a channel, and the Data Bots import it and manage segments of people, and generate reports.
See? No magic ponies. Just smart bots that work with smart people to get their jobs done.
Stay Tuned for even more information on how it all works in future posts – and sign up for our newsletter in the upper left sidebar!