Stakeholder Management And Metrics

The previous post covered TikTok's swipe-up nudge, which drove a 900% lift in engagement. But TikTok's success also created a problem no one had planned for.

The same recommendation engine that was brilliant at surfacing content people wanted to watch was equally brilliant at spreading content that was wrong.

When a Feature Helps One User and Harms Another

Unverified information began spreading rapidly on TikTok in several countries. In at least one case, incorrect content contributed to real-world civil unrest.

The product team needed to slow the spread without breaking what was working.

The solution was a pop-up. When a user tried to share a video flagged by others as potentially unverified, a message appeared before the share completed -- informing them that other users had questioned the content's accuracy. No automatic removal. No algorithm block. Just a moment of friction before the share went out.

That single intervention reduced the spread of misinformation by 70% in most of the countries where TikTok operated.

The engineering work was not complex. A developer could ship a conditional pop-up in hours. The hard part was the accuracy of the signal: which videos get flagged, and how confident is the system that the flag is correct?

If the flagging system incorrectly marks accurate content as unverified, the creator of that video is falsely accused of spreading misinformation. They feel wronged by the platform. They leave.

And when creators leave a platform that depends on user-generated content, the entire product breaks down.

Creators produce content. Users watch it. The positive response motivates more content. More content attracts more users. Interrupt any part of that loop, and the whole thing stalls.

This is what product management means when it talks about walking a thin line. The feature that reduces misinformation is the same feature that, implemented carelessly, drives away the creators who make the product worth using.

Stakeholder Management: Everyone Your Product Depends On

TikTok's misinformation case makes explicit a concept product management formalizes as stakeholder management: any product decision affects more people than the obvious user, and missing one can cause the product to break in ways that data does not immediately explain.

For TikTok, the stakeholder you might overlook is the content creator. The platform cannot exist without them.

For Uber, the obvious pair is driver and rider. But consider who else matters:

• Payment providers: if a payment gateway goes down, users cannot complete trips. Payment companies are a dependency, not a background service.
• Government regulators: speed limits in the app come from local governments. Women's safety features like the SOS button are now legally mandated in India. If a regulation changes, Uber's product changes too -- regardless of its own roadmap.
• Corporate accounts: businesses that pay for employee travel are a separate user type with different needs and different expectations.

For the NPCI and UPI, the picture is even more complex. UPI is not built for individual consumers alone -- it is built for every bank in India, each with its own API, its own transaction limits, and its own failure modes.

When HDFC Bank's server is down for an hour, total UPI transaction volume drops measurably. A developer at a fintech who has not mapped every bank's edge cases will ship code that handles 80% of scenarios and fails unpredictably on the rest.

In payments, that is enough to destroy user trust permanently.

Stakeholder management means mapping every entity whose behavior can affect your product's success -- and building with those dependencies explicitly handled -- before you write the first line of code.

Hotstar's Sticky Notification

Disney+ Hotstar faced an engagement problem during cricket matches. Users would open the app, watch five minutes of an IPL match over lunch, close it because they had to return to work, and then not come back -- even if the match was still live.

Hotstar makes money from advertising. More watch time means more ad impressions. Every user who forgets to return is lost revenue.

A standard push notification does not solve this. Users either dismiss it or tap it, at which point it disappears. Hotstar needed something that stayed visible throughout the match, updating continuously, giving users a reason to return without requiring them to open the app first.

The solution was a sticky notification: a notification that remained in the Android notification bar even after "clear all" was tapped, and that updated dynamically with live match data.

Current score. Key events. "Dhoni has hit a six -- watch now."

A single swipe down from the top of the screen showed the score. One tap opened the app at the live match.

The feature significantly improved both engagement and return rate during matches. But it came with a critical caveat.

When other apps in India copied the pattern without the same level of user love for their product, users found the persistent notifications irritating and uninstalled. Android eventually added a per-app setting to disable sticky notifications -- driven largely by Indian apps over-using the feature.

The lesson is not that sticky notifications are bad. It is that aggressive product techniques work when users genuinely want to stay close to your product, and backfire badly when they do not. The technique is not the decision -- the underlying product quality is.

Engagement, Retention, and Monetization

These three terms define almost every product decision a developer or PM will make. Knowing which one a feature is optimizing for changes how you build it.

Engagement is how much time a user spends in the product in a given session or day. TikTok's swipe-up nudge was an engagement feature. Hotstar's sticky notification was an engagement feature.

Retention is how consistently users return across days and weeks. A user who opens Swiggy once a month is a low-retention user. Daily practice streaks on learning platforms are retention features -- they bring users back by making the streak feel like something worth protecting.

Monetization is whether and how the user completes a valuable action: subscription renewed, order placed, ad viewed, transaction completed.

What counts as good numbers varies by product type. Once-a-month opens on an e-commerce platform is fine -- people do not buy clothes every day. Once-a-month opens on a food delivery app is a problem -- people eat every day.

The benchmark is set by what the product is for, not by an absolute standard.

The Essentials

1. Stakeholder management means mapping every entity your product depends on before you build -- users, creators, payment partners, regulators. Missing one is how products break in ways that data does not immediately explain.
2. TikTok's misinformation pop-up reduced false sharing by 70% with a single friction point -- but accurate flagging was as important as the feature itself. False positives drive away creators and break the content flywheel.
3. Sticky notifications work when users love your product. They produce immediate uninstalls when users do not. The technique does not substitute for the underlying product quality.
4. Engagement, retention, and monetization are the three levers every product feature pulls. Know which one you are optimizing for before you decide how to build.

Further Reading and Watching

• Good PM / Bad PM (Ben Horowitz, a16z) -- the definitive essay on what separates product managers who ship real outcomes from those who do not
• The Hierarchy of Engagement (Sarah Tavel, Benchmark) -- how engagement leads to retention and retention unlocks monetization, exactly the loop this post describes
• How to Set Product Metrics (Lenny Rachitsky, YouTube) -- practical walkthrough of choosing the right metric to optimize for at each stage of a product