Research is so important. The closer you are to your customer, the more credibility you have. If you're fortunate enough to have a research org at your company, even better. My first weeks at Airbnb were in large part spent sitting in with the research team as they interviewed our users.
In terms of timing, the earlier you develop that understanding of and relationship with the customer the better. Great research leads to well-informed, strategic insights that will serve as the bedrock for your launch.
So, I never use all these types of research, but here's a great menu of research phases that I keep at the top of my Research Plan Template. Usually at most I'd pick one of these per phase, that is most important and will produce the most actionable insights.
Phase 1 - Early stage
Phase 2 - Early stage **This stage is so crucial if you don't already have data to support the direction pre-design phase**
Phase 3 - Mid stage
Phase 4 - Launch stage
Phase 5 - Late stage
Our product teams are great at this and we glean a lot from them during a launch but we’ll also do a lot of customer interviews on our own. I also always learn a lot from talking to our own Service team, but I specifically work on software for service teams. Often we also run a big survey to collect data points that support our narrative. For instance recently we wanted to figure out how many people invest in customer listening and advocacy tools. So we ran a big survey and that asked questions related to how service teams approach this. The data kind of became it’s own launch and added a ton of social proof to our message. Example: https://blog.hubspot.com/service/state-of-service-2019-customer-first
Foundational qualitative research should be done as soon as you are able, and also ahead of any major product shifts or market expansions. We've also explored qualitative research with "the ones who got away" - lost, paying customers. Understanding where we've failed them or where competition has become a better alternative has also been really helpful - sometimes it's a solution we already have that they weren't able to find easily and we can make changes without largely changing the product. The quicker you can engage lost customers, the better, because once they've left their desire to give feedback (or even remember) greatly diminishes the more time has passed.