Analyze potential causes for the decline in daily viewership on Netflix and propose appropriate solutions

I’d first clarify what “viewership” means, verify the decline, and identify where users are dropping off. That would help me distinguish between fewer people choosing Netflix and people trying to watch but struggling to finish.

1. Clarify the product, scope, and metric

Netflix lets users watch movies, series, and documentaries across devices. For this case, I’ll focus on streaming across the platform.

Before investigating, I’d ask:

• Does viewership mean unique viewers, streams started, streams completed, or watch time?

• Is the decline global or concentrated in particular markets or devices?

• Did it happen suddenly or gradually?

• What period are we comparing against?

Here, the case defines daily viewership as the number of streams completed each day, with a gradual 15% decline over six months.

The calculation would be:

Percentage change = (Current daily completed streams − Baseline daily completed streams) ÷ Baseline daily completed streams × 100

I’d clarify what qualifies as “completed,” including the treatment of credits, repeated viewing, and content resumed on another day. I’d compare equivalent periods using weekly averages to reduce daily noise.

This distinction matters because a decline in completed streams doesn’t necessarily mean fewer people are watching. Users might be watching longer titles or spreading their viewing across more sessions.

2. Verify that the decline is real

I’d check whether the completion definition, tracking events, reporting dashboard, or data pipeline changed during these six months.

Even if the definition stayed the same, a tracking failure could make completed streams appear lower. I’d compare completion events with playback progress and other independent playback signals.

Next, I’d break down the decline by:

• Device, operating system, and app or player version.

• Country, region, and time of day.

• New versus existing subscribers and subscription plan.

• Content category, title length, and release date.

• Relevant demographic segments, where appropriate data is available.

The case tells us the decline spans platforms, regions, and user groups. That makes a narrowly local issue less likely, but I’d still compare the size of the decline across segments. “Everyone is affected” doesn’t mean everyone is affected equally.

I’d also check historical seasonality. A first-time decline of this magnitude deserves investigation, but previous holiday or release-cycle patterns can still help establish the baseline.

3. Explore external factors and changing viewing habits

I’d examine whether the wider market offers an explanation.

Are competing platforms seeing similar declines? Have competitors introduced unusually compelling content, discounts, or bundles? Have regulations, negative publicity, strikes, weather events, or major disruptions affected viewing or content availability?

In this case, there’s no strong evidence of these changes, and the issue appears specific to Netflix. I’d lower their priority.

However, two behavioural signals deserve follow-up:

• Some users are spending more time on short-form content.

• Affordability may be affecting subscriptions, particularly in India.

I’d also consider changes in work routines and household spending. These could reduce the time or money available for entertainment.

I wouldn’t treat these signals as the answer yet. An affordability issue concentrated in one market is unlikely to explain a global decline by itself. Similarly, a shift toward short-form content needs evidence that it is reducing Netflix completion, rather than simply happening alongside it.

4. Locate the drop in the user journey

I’d map the journey as:

Open Netflix → Browse or search → Open a title → Start playback → Continue watching → Complete the title

I’d also review changes to pricing, business strategy, marketing, recommendations, and product releases during the same period.

The case provides the following evidence:

I’d verify both absolute volumes and conversion rates. Stable rates alone can hide changes in the number of users moving through the journey.

At this point, I’d focus on why users abandon content after starting it.

5. Test the strongest hypotheses

I’d investigate three explanations:

The interviewer adds that the video player was recently updated and users have reported playback problems. That makes a player regression my leading hypothesis.

However, I’d still validate the connection. I’d compare updated and unaffected versions, check whether errors occur immediately before abandonment, and reproduce the reported problems.

I’d also reconcile the timing: a gradual six-month decline should align with the rollout or increasing exposure to the faulty player. If the update happened much later, it cannot explain the entire decline.

My working diagnosis would be that playback issues are contributing to lower completion. I’d call them the primary root cause only once the timing, affected sessions, and recovery evidence support that conclusion.

6. Address the problem in the short and long term

My immediate priority would be restoring reliable playback.

In the short term, I’d:

• Pause the affected rollout and roll back or disable the problematic change where feasible.

• Fix and test the specific buffering, crash, or playback defects identified.

• Release the fix gradually and monitor affected sessions.

• Communicate with users who reported the issue once the fix is verified.

After playback is reliable, I’d consider targeted reminders or relevant recommendations to help affected users resume watching. Promotional offers could be tested if there is evidence of lasting disengagement, but they would not be my first response to a technical failure.

In the longer term, I’d:

• Strengthen player testing across devices, operating systems, and network conditions.

• Use staged releases with playback-quality monitoring and rollback criteria.

• Monitor the full journey so completion problems surface earlier.

• Invest in delivery infrastructure if the investigation identifies CDN or network bottlenecks.

I’d keep exclusive-content expansion as a separate hypothesis. It could help if we find a content-engagement problem, but it would not resolve a player defect.

7. Measure recovery against the original problem

My primary outcome would remain daily completed streams, measured consistently against the original baseline. I’d use completion rate alongside it to account for changes in playback starts.

For completion rate, I’d follow comparable groups of started streams over a consistent window, allowing for users who finish later.

Supporting metrics would include:

• Playback error and crash rates.

• Rebuffering time as a share of playback time.

• Mid-stream abandonment.

• Playback-related complaints per 1,000 streams.

• Completion among previously affected users.

• Viewer satisfaction and return behaviour.

Total watch time would provide additional context, rather than replace the original metric. I’d also check that improvements don’t come at the expense of video quality, startup time, or data usage.

I’d compare the repaired experience with an appropriate baseline or unaffected group, accounting for content releases and seasonality. That would help establish whether the fix caused the recovery.

My recommendation is to prioritise the player investigation and restore playback reliability first. Users are still arriving, finding titles, and pressing play. The strongest evidence points to a problem in their ability to keep watching.