When to automate
You will know if you have a candidate for an automation when you have: a frequent manual process, consistent data, and a trigger.
1. Frequent manual process
The frequent part in frequent manual process is whatever allows you to get the job done. Here are a few examples of some manual processes you may see in your day to day:
- Every week I update a GitHub ticket with data from a Google Spreadsheet.
- Every day I need to know how many people are going to be in the office, including visitors.
- I need to know who at the company has anniversary next month and how many years so we can order them a personalized cake.
2. Consistent data
Having a consistent data source can make or break your ability to automate. Check to see that your data:
- Lives on an application or service that has an API.
- Has attributes or identifiers that will allow you to get return data. For example, if you need to trigger an event based on a specific date, then you need to find a date field in your data.
3. A trigger
Your automation needs a trigger to know when to run. Your automation could run on a schedule:
- Every 5 minutes
- The first Monday of the month
- Every day at 5:30 pm Pacific Time
Or your automation can run in response to another event. For example, if you send a tweet with a hashtag, then you could trigger an automation to save the contents of that tweet to a spreadsheet.
Should I build it custom?
Before creating something from scratch, check to see if a service exists that can do it for you. Services like Zapier, IFTTT, and Apple’s Shortcuts can make automation as straight forward as connecting your accounts and clicking a few buttons. Slack also offers built-in features and apps to help automate processes.
If services cannot connect in the way you need or if you have privacy concerns, then you’ll need to create a custom automation.
Where can I host a custom automation?
You have a lot of options to host your code and schedule it or listen for changes. Here are a couple that I’ve worked with:
- Amazon Web Services (AWS) - You can upload your code to an S3 bucket and then use a Lambda for trigger and/or schedule it. Examples:
- Glitch - You can create an Express app that will listen for changes to a webhook and run code and more. Examples:
Is my automation helpful?
You should regularly check-in with your automation to make sure it still accomplishes your goal, especially in the company of others.
- Does the automation add noise? If your automation posts too often or presents wrong information, users will begin to distrust it and ignore it. Evaluate your automation for quality over quantity.
- Is the automation working against your workflows? Your automation should work with your team. Find ways to make interacting and working with the automation easier for all.
What’s next?
Finally, have a work-in-progress mindset with your automation. Your workflows and productivity constantly change, so should your automation. Ask for colleagues for feedback and see how others interact with your automation, and then iterate.