Attention Incentive Audit
Predict a platform's behavior by tracing who pays and what earns revenue
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 94%
The Attention Incentive Audit starts with Hari's question: which technology is working in whose interests? Identify the paying customer, the product being sold, and the user behavior that increases revenue. On an advertising-funded social platform, Hari argues that more time and more behavioral data can increase value to advertisers, so product teams are pushed toward notifications, recommendations, and feeds that extend engagement. A subscription service can also want continued use, but the paying relationship may create a different incentive to satisfy the subscriber. Compare the platform's revenue-maximizing behavior with the outcome the user actually wants, then evaluate whether controls, design changes, or another business model would align them better. The audit predicts pressure rather than proving the intent or effect of every feature; each product still requires evidence.
Origin
Hari develops the model from interviews with former technology employees and contrasts advertising-funded social platforms with services where the user pays directly.
Core principles
- 01Product behavior follows revenue incentives
- 02The user and the paying customer may be different people
- 03Engagement metrics can conflict with user intent
- 04Changing incentives can change product competition
How to run it
- 1
Name the payer
Identify who provides the revenue: the user, advertisers, an employer, or another party.
Pro tip Separate the person using the product from the person buying access to them.
Watch out Do not infer the full business model from a single feature.
- 2
Name the product
State what the payer receives, such as access, attention, behavioral targeting, or a direct service.
Pro tip Follow both money and data.
- 3
Trace the rewarded behavior
Identify which user actions create more revenue or valuable data and which metrics employees are likely to optimize.
Pro tip Inspect defaults such as recommendations and notifications.
Watch out An incentive creates pressure; it does not prove every employee's motive.
- 4
Compare interests
Write the user's desired outcome beside the revenue-maximizing behavior and mark where they align or conflict.
Pro tip Use the user's long-term intention, not only the next click.
- 5
Redesign the incentive
Consider controls, product changes, subscriptions, collective ownership, or regulation that would make success depend more directly on the user's desired outcome.
Pro tip Test whether the alternative merely moves the same conflict elsewhere.
Watch out Policy effects and trade-offs require evidence beyond this diagnostic model.
In the wild
Hari asks why a social platform does not prominently help nearby friends meet offline. His explanation is that a successful meet-up takes both people away from the platform, conflicting with an advertising model rewarded by continued use.
→ The feature gap becomes predictable once user intent is compared with the platform's revenue incentive.
Hari contrasts Netflix, where the viewer pays, with advertising-funded social platforms. He argues that both seek continued use, but the direct payer relationship changes whose satisfaction the service must prioritize.
→ The comparison shows why apparently similar recommendation systems can operate under different incentive structures.
Common mistakes
Stopping at good intentions
The audit examines what the revenue system rewards, not only what a company says it wants.
Calling all technology the problem
Hari explicitly distinguishes technology itself from particular incentive structures and business models.
Treating pressure as proof
A business-model incentive predicts likely optimization pressure but does not establish the effect of every feature without evidence.
Is it for you?
Best for
People evaluating platforms, product incentives, recommendation systems, or attention-related regulation.
Not ideal for
Assuming every product feature is harmful without examining its actual business model, controls, and user value.
From the transcript
“what tech working in whose interests”
“today you're not the customer of facebook you're the product they sell to advertisers”
From the episode
How To Fix Your Focus & Stop Procrastinating: Johann Hari