Don't assume feedback is easy, free and helpful

Matt Wensing tweeted about product feedback couple months back and it had been living rent free in my head for a while. As we navigate the current Software Development Lifecycle (SDLC) and my imminent shift away from product building role in the future, I wanted to unpack for the current crop of builders who will have to dabble in product management.

Feedback is a privilege, so build something great. 98% of things die without knowing why, or how much they suck.

— Matt Wensing 🐙 (@mattwensing) August 9, 2026

He had this underrated tweet about how feedback is a privilege. And then followed it up by adding some more context.

Shipping something that you know is weak so you can iterate based on feedback is based on the belief that users will engage with it for long enough to form a reaction, and then use their time to tell you how it can improve. This is a very risky assumption.

— Matt Wensing 🐙 (@mattwensing) August 9, 2026

I see this in my work as well. Creators who are moving into these roles as we become more podular as an organisation. They think that we will get feedback from users. We won't. Getting good feedback is really expensive.

When was the last time you gave honest feedback as a consumer?

Often times when I listen to engineers or PMs, who recently moved into the role of product management, justify their feature/MVP rollout on the assumption that feedback will improve it further. I ask them how do they know if they would get feedback from customers?

This gets even trickier in B2B where the user of the software or tool may not be the beneficiary of the tool. We could use proxies like product metrics but any deduction of those metrics should be backed by a hypothesis. If we rollout a half baked feature and people bail out mid-way through the flow, was it because the feature we built was not valuable or was it an information architecture problem.

Let me take this tweet about coding agents becoming ubiquitous yet not raising the bar of IA.

despite coding agents becoming amazing, most software isn't getting *better*

as in, you can tell the human engineers behind the software are not actually great at IA, sensible user flows, etc 🧐 agents dont push back enough on poor human judgment

— Sara Du (@saradu) August 8, 2026

The above sentiment is a symptom of improving on feedback.

Taking feedback is not about validating your previous decision but learning the emergent behaviour based on your decision. It requires a sense of care and is often costly.

We are inundated with survey questions requesting us to rate a service like airlines, hotels and online purchases. The number of times I replied can be counted on fingers and each time I did, I had some not nice things to say.

B2B is different, your customer are looking for modifications to existing solutions and they would like us to build it for them. They also would not like to pay the cost of building it for them. The loud and vocal customers may not be always the ones pushing the product in the direction that improves it for everyone.

Built to order (BTO) is an interesting new possibility where the customer can also pay for the cost of building the solution that could improve their work. I am yet to see success on this front but I am highly motivated about learning how we can make it work.

To close it out, build with an opinion and test your hypothesis. The easy part is that doing the reps is much easier with coding agents, the harder part is that you learn from those reps and figure out how to build an opinion.

Things like North star metrics and weekly business review are important practices to discern what you learn for each of the reps.

Round up

This week I was prepping for my employer's annual flagship conference where I am presenting a few sessions. Prepping for these sessions took the entire week and I finished 80% work on them just before I typed this out. Slides are still work in progress.

like slides being pretty was never what made them good? the pretty part is the last 10%, and it’s sort of easy… people who are good at making slides and storytelling (which is hard!) kinda just pick it up as they go and add it onto important decks https://t.co/VBYc1PcBTi

— Maia Bittner (@maiab) September 27, 2026

I was searching for an angle and repurposing some slides, but it didn't pan out. I had to re-do the entire thing but I am much more in grip with delivering a message on my beat. Since I live in a region where I act as a reporter observing the events unfolding without an invested stake in it.

Links that resonated

I liked this Muse write-up from Facebook. I am an Instict user because I don't trust facebook. My instinct has a burner facebook account for casual marketplace browsing. But this gives a good outline on how an agent should work.

Introducing Muse.ai

Interesting take from Tom Tunguz on the need for narrow AI. I believe most of the data needs can be solved by Decision models like Jev by Typesafe.ai

AI comes for the if statement

Sign off

I started working on this with a budget to finish it within a single coast to coast flight ride.

I am going to unpack Built to Order and why Venkatesh Rao's Art of Gig may have some frameworks in place to make it successful.

In my plane ride, I met a father son duo who run a swimming pool construction and services business in NYC. We started talking about how archaic and difficult their software is and what they need. I started discussing with them about agentic AI and how the future will evolve. We are exchanging numbers when we land to continue the conversation. They are psyched to explore but would like someone to build to order. I will be thinking about this and writing more on it for sure.

Signing off till next time,

Vivek , experienced the longest ever runway gridlock