Share your methodology for presenting technical trade-offs to product teams to ensure they understand why a blocker exists.

Question

Anonymous · Accepted Answer

I typically try to make the concern concrete and instead of refusing, I try to
suggest the most viable path forward

I had built a ROI metric that used sensitive user conversion level data. This
metric design had to be done so as to ensure user privacy was protected. We had
a successful metric launch. The metric was available at the ad surface level
facebook / insta feed/ insta stories etc. level, as a follow-up a PM had
requested the metric to be adjusted to include experiment version level as well.

They suggested this was a quick change given a field had to just be added,
however I had preempted that this would probably break the current metric
design. Experiment level was a very high cardinality field and would cause a lot
of issues at the K-anon and noise adding steps, Instead of just saying no, I
built a small toy POC based on subset of data within the next day to demonstrate
the issues, 70% of the signal lost to k-anon, remaining signal too noisy after
differential privacy noise addition.

This helped convey the point that the request was a large one that needed a
ground-up metric redesign and launch and would lieky take 3-4 months as privacy
and legal approvals would be part of the process

The POC was a simple but powerful way to communicate the complexities underlying
the seemingly simple request, we did prioritise this project as part of the next
half

Share your methodology for presenting technical trade-offs to product teams to ensure they understand why a blocker exists.

Practice this question with AI

Go Premium

Community Answers

Unlock Full Access

Practice More Questions