Hi, thank you for the article, I enjoyed reading it.
I'm glad to share that my data team is on the way to create values in all the 3 types, especially the Tier 3.
However, I still find Tier 2 project sometimes hard to manage both progress and stakeholders expectation. You said that "be careful about shoving this type of work into the agile framework.", so is there any other methods that you recommend for this project type? (I'm working on something similar to Marketing media mix)
That’s a wonderful question! Because it’s research, the value is somewhat unknown but not completely unknown. For example you might be researching how LTV could be used in your company. The value of LTV itself is well known, but its value to your company is unknown.
As far as managing expectations goes, you need to frame the research question in terms of value delivered. If you’re exploring LTV usage, then the research goal should be to determine the viability of using LTV in your company, nothing more. So make sure everyone is clear that you won’t be implementing anything as part of the effort.
Of course you have to attack the question from multiple angles and even get feedback from stakeholders along the way, but the important part is that everyone agrees on the goal and the timeframe. After thst it’s standard project management. Make sure you timebox the work end everyone should be happy.
I would suggest you find and use a template for R&D projects that specifies concrete deliverables even if the decision ends up being “no go”
Thanks for the article Ergest. It reflects well how I think about value.
To part 2, i.e. R&D, I would like to add the terminology “hypothesis testing” in which you might test (weak) relationships in your KPI tree(s) - which you described in terms of finding causal relationships. Together with the organisation’s strategy it should inform what to work on next.
Finally I would not disregard the value of “answering questions”, in case you find value in them. I like to see my colleagues as subject matter experts with good ideas and questions. Also, you can decide whether you deliver them a workable prototype with reduced effort or a production-ready “data product”.
Yes, I consider hypothesis testing as part of R&D which puts it in the context of unpredictable outcomes. I also agree that answering questions does have value after the basic metrics have been set up. I plan to explore this further
Hi, enjoyed the classification here. I was wondering what happens when one of the R&D project do yield significant impact? E.g., a multi-touch attribution model that works well enough for the business. Should they remain as R&D, since there might be ongoing research that's still required to improve, while a good enough version "graduates" to Tier 3?
Hi, thank you for the article, I enjoyed reading it.
I'm glad to share that my data team is on the way to create values in all the 3 types, especially the Tier 3.
However, I still find Tier 2 project sometimes hard to manage both progress and stakeholders expectation. You said that "be careful about shoving this type of work into the agile framework.", so is there any other methods that you recommend for this project type? (I'm working on something similar to Marketing media mix)
That’s a wonderful question! Because it’s research, the value is somewhat unknown but not completely unknown. For example you might be researching how LTV could be used in your company. The value of LTV itself is well known, but its value to your company is unknown.
As far as managing expectations goes, you need to frame the research question in terms of value delivered. If you’re exploring LTV usage, then the research goal should be to determine the viability of using LTV in your company, nothing more. So make sure everyone is clear that you won’t be implementing anything as part of the effort.
Of course you have to attack the question from multiple angles and even get feedback from stakeholders along the way, but the important part is that everyone agrees on the goal and the timeframe. After thst it’s standard project management. Make sure you timebox the work end everyone should be happy.
I would suggest you find and use a template for R&D projects that specifies concrete deliverables even if the decision ends up being “no go”
Thanks for the article Ergest. It reflects well how I think about value.
To part 2, i.e. R&D, I would like to add the terminology “hypothesis testing” in which you might test (weak) relationships in your KPI tree(s) - which you described in terms of finding causal relationships. Together with the organisation’s strategy it should inform what to work on next.
Finally I would not disregard the value of “answering questions”, in case you find value in them. I like to see my colleagues as subject matter experts with good ideas and questions. Also, you can decide whether you deliver them a workable prototype with reduced effort or a production-ready “data product”.
Yes, I consider hypothesis testing as part of R&D which puts it in the context of unpredictable outcomes. I also agree that answering questions does have value after the basic metrics have been set up. I plan to explore this further
Hi, enjoyed the classification here. I was wondering what happens when one of the R&D project do yield significant impact? E.g., a multi-touch attribution model that works well enough for the business. Should they remain as R&D, since there might be ongoing research that's still required to improve, while a good enough version "graduates" to Tier 3?
Great question! Yes they’d have to graduate to Tier 3 in order to provide continuous value.