Knowing When To Automate The Process

A few months ago, Lizzie Seedhouse published an article discussing the time it takes to make a quality connection, as well as the process that CopyPress has implemented to do so.  This was a great post for current and potential clients to see what challenges some of our staff have to overcome everyday. I however, read this from a different perspective, which is that of a developer looking to make as much of this process as easy and time saving as possible through the use of automation.

Before a process can be automated, there must be a full workflow created, implemented, and tested; any issues, or conflicts needs to be noted and tracked, and solutions need to be thought of. Depending on the task, this is something that can take days, months, or even years to achieve.  For example, in 2005 a group of Google Engineers were able to develop a robotic vehicle, also known as a driverless car.  This is something that has been in the works for hundreds of years, and without all the tests and knowledge that came to them from cars that needed drivers, they wouldn’t have been able to successfully do this.

shutterstock_98570228Image via 1000 Words /

When we began testing our connections software over a year ago, we were not fortunate enough to have the extensive knowledge and data that these Google Engineers had, and we learned that very quickly.  Referring back to some of our outreach process, the overall workflow has not changed to much since we considered automating, but at the time we didn’t have any solutions when a new issue would occur.  Let’s take a look at some of our process and breakdown what would happen if we tried to automate it.

Mining Domains

shutterstock_61446970Looking at the same UK Fashion data Lizzie references in her article, completing this step with human interaction and through automation would not return the same results.  Software can easily find the top 250,000 domains in Alexa, as well as determine which of these are UK based, however when trying to find a specific category like “fashion sites”, it would produce varied results.  Letting our system determine the category of a site is possible, but is not as accurate or subjective as the human eye, and this would be the first failure in the automation process.


Just as Lizzie mentioned, bloggers are real people.  Just like you, they do not want spam or junk email, and that is exactly what they would get if we had an automation process for reaching out to these bloggers.  The way a connection is being made currently is a true guest blogger reaching out to another blogger who they feel could be a good connection. Deciding which blogger would find true value in a guest post is not an objective decision.  If we were to decide to automate this process we would be going against this, and we would be faced with another failure in the automation process.

shutterstock_80248693Decide what works

Nobody will argue that when done correctly automation can change the way a company is built, but when done prematurely or incorrectly, it can also be the downfall of a company.  Knowing when to automate a process is much harder than knowing how to automate a process, and it is something that should not be rushed. Getting feedback from everyone on the team and keeping an open mind is essential to progressing toward automation.

About the author

Eric Kaufman