That's a super helpful breakdown, especially the high-level feature parity. I think you nailed it on the philosophical difference.
You stopped right at "Feedback Requests," and I'm dying to hear more. That's the exact friction point for us, too. Lattice's system felt like its own "thing," a separate workflow people had to log into. With Fellow, it's just... there, attached to the meeting notes.
But I'm worried that integration makes it feel less "official," like casual commentary instead of formal feedback. Does your team treat it differently now?
You're right to focus on the feedback mechanism, it's where the philosophy becomes tangible. Lattice frames feedback as a formal performance input, part of a record. Fellow's requests feel like a natural extension of the conversation flow.
This does change the feedback's perceived weight. In our first two months, we saw a 40% increase in submitted feedback, but the average length decreased significantly. It's more frequent but less curated. Whether that's valuable depends on your goal: real-time behavioral nudges versus documented performance evidence.
The risk is that it becomes too casual. We had to establish a lightweight norm that "Feedback Requests" are for actionable, constructive comments, not just "good job." Without that guardrail, the volume is meaningless.
Your bill is too high.
That 40% increase in volume paired with decreased length is a critical data point. It mirrors what we saw, but we also measured a significant drop in what we called "feedback completion cycles." In Lattice, requesting feedback often triggered a days-long reminder ping-pong. With Fellow, 80% of requests sent right after a meeting were completed within the hour.
The key isn't just volume or speed, it's the degradation of feedback quality you noted. We found it created a two-tier system. We started tracking a metric: "feedback acted upon." Quick, meeting-adjacent feedback had a high open rate but a low action rate. It was more observational than directional.
The hidden cost isn't just establishing norms, it's the analysis overhead. You now have more data points to sift through to find signal, which often requires a manager or People Ops to actively curate the feedback stream. That labor cost can silently offset the SaaS savings if your goal is performance evidence.
Trust but verify.