Know the change you wish to bring. And then track it.
We as designers spend a lot of our energies in understanding the problem and solving them.
What we often overlook is the impact it would create. What change will it bring to the users, if the design goes out live in the world. And most importantly, how would one know if the designs had the intended impact?
This is where deciding success metrics before releasing the designs are very important. Define what success looks like. Share this with other stakeholders and have a common understanding.
What would change in the product/experience if this is a success?
Defining proxy metrics for this change can help us gauge the impact in an unbiased fashion. Derive these metrics from your vision. Ask your PM/Data scientist about the product metrics that can help you do this. If the said data point is not being captured yet, work with them to define it and set it up.
There might be a lot of unintended changes too; but knowing what's unintended will help you learn more about your users. And what's better than learning !
A great recent example of this, in my opinion, is Twitter shutting down Fleets in August.
Twitter recently announced that it will shutdown its ephemeral Fleets feature, citing that “we haven’t seen an increase in the number of new people joining the conversation with Fleets like we hoped. (https://blog.twitter.com/en_us/topics/product/2021/goodbye-fleets)”.
This is what success looked like to them and it was unsuccessful in doing that. Instead they saw users using it more to amplify their own tweets, instead of new people joining in conversations. Worked but not as intended.
Don't just make things work. Make them work as intended!