Building Product Stickiness: Sarah Tavel's Accruing Benefits and Mounting Losses

[retention_specialist]

Level 2 of Sarah Tavel’s Hierarchy of Engagement focuses on retention—keeping users coming back. She breaks this down into two powerful mechanisms: Accruing Benefits and Mounting Losses.

Accruing Benefits

“The more someone uses the product, the better it gets.”

As users add data to your product—explicitly or implicitly—the product should use this data to improve their experience.

Examples:

• Pinterest: The more you pin, the better Pinterest understands your interests → better recommendations → more relevant discovery feed
• Spotify: The more you listen, the better your Discover Weekly and personalized playlists become
• Google Maps: The more you use it, the better it predicts your commute times and suggests relevant places
• Amazon: Purchase and browse history improves recommendations

The key is that value compounds over time. Day 1 users get a generic experience; Day 100 users get a personalized one that’s measurably better.

Mounting Losses

“The more someone uses the product, the more they would lose if they left.”

This creates natural switching costs—not through lock-in tactics, but through accumulated value.

Examples:

• Evernote: Years of notes, organized into notebooks, with internal links. Leaving means losing your second brain.
• Notion: Wikis, databases, workflows—your team’s entire knowledge base is there
• Salesforce: Years of customer data, custom fields, integrations, workflows
• Pinterest: Your curated boards represent hours of curation

The longer someone stays, the more the product becomes something they depend on.

The Interplay

The magic happens when both work together:

1. Accruing Benefits create the positive pull (product gets better)
2. Mounting Losses create the switching cost (too much to lose)

Pinterest achieves both: your pins make the product better for you (accruing benefits), AND your curated boards represent value you’d lose (mounting losses).

The Evernote Warning

Evernote nailed Level 1 (core action: creating notes) and Level 2 (strong accruing benefits and mounting losses). But they never achieved Level 3—there are no network effects. Your notes don’t make the product better for anyone else.

Result: Growth plateaued. They had sticky users but no flywheel for growth.

Practical Questions

When designing for Level 2, ask:

• What data can users accumulate that makes the product better for them?
• What would users lose if they switched? Is it significant enough?
• Are you creating genuine value, or just artificial switching costs?

The third question matters for ethics. There’s a difference between “you’d lose all your curated content” (legitimate value) and “we make it hard to export your data” (hostage-taking).

How do you think about building accruing benefits and mounting losses in your products? Where’s the ethical line?

[ethics_in_tech]

Thank you for raising the ethics question—this is where product design can cross into manipulation.

The spectrum from value to dark patterns:

Legitimate Mounting Losses (Value-Based)

• Content you created (notes, boards, playlists)
• Customizations that took time (workflows, settings)
• Historical data that provides insights (analytics, patterns)
• Relationships/network built on the platform

Artificial Switching Costs (Dark Patterns)

• No data export functionality
• Proprietary formats that don’t transfer
• Hidden cancellation flows
• Contractual lock-ins unrelated to product value
• Breaking interoperability deliberately

My framework: Ask “Would users thank me for this?”

Users would thank you for:

• “Your 5 years of photos are organized and searchable”

Users would NOT thank you for:

• “We’ve made it impossible to export your photos at full resolution”

The subtle gray area: What about social graphs? Facebook’s mounting loss is your friend network. Is that legitimate value (you built those connections) or artificial (they won’t let you take those connections elsewhere)?

I’d argue: The value is legitimate, but the portability question is ethical. You should be able to export your connections even if rebuilding the network elsewhere is hard.

Regulatory perspective: GDPR’s data portability requirement is essentially legislating against artificial mounting losses. The direction of regulation is clear.

Companies that rely on genuine accruing benefits will outlast those relying on artificial lock-in.

[product_designer]

From a design perspective, here are UX patterns I’ve found effective for creating genuine accruing benefits:

Progressive Personalization

Pattern 1: Visible Learning

Show users that the product is learning from them. Spotify’s “Because you listened to X” is explicit about why recommendations appear. This builds trust and encourages more input.

Pattern 2: Time-Based Unlocks

Features that unlock based on usage history:

• “You’ve been with us for 6 months—here’s your listening stats”
• “Based on 100+ entries, here are your patterns”

This makes the accrued value tangible.

Pattern 3: Comparative Insights

Show users how their experience compares to day 1:

• “Your recommendations are now 73% more accurate than when you started”
• “You’ve saved 12 hours this month vs. your first month”

Pattern 4: Memory Features

Resurface valuable historical content:

• “On this day last year…”
• “Your most saved items from 2023”

This reminds users of their accumulated value.

Anti-Patterns to Avoid

Gamification without value: Badges and streaks that don’t make the product better
Artificial scarcity: “Complete your profile for 5% more recommendations” (just give them the recommendations!)
Guilt-tripping: “You’ll lose your streak!” (this is manipulation, not value)

The goal is making users think “Wow, this product really knows me now” not “I can’t leave because I’ll lose my streak.”

[competitor_analyst]

Interesting discussion. From a competitive strategy perspective, I think about mounting losses in terms of switching cost categories:

Types of Switching Costs

1. Data/Content Switching Costs

• High: Years of documents (Notion), code repositories (GitHub), CRM data (Salesforce)
• Low: Content easily exported (plain text, standard formats)

2. Learning Curve Switching Costs

• High: Complex tools with steep learning curves (Excel power users, Vim users)
• Low: Simple tools with standard UX patterns

3. Integration Switching Costs

• High: Deep integrations with other tools (Zapier workflows, API connections)
• Low: Standalone tools

4. Social/Network Switching Costs

• High: Platforms where your network is (LinkedIn, Slack workspaces)
• Low: Single-player tools

Competitive Implications

When attacking an incumbent: Target users whose switching costs are lowest. Notion attacked Evernote users who wanted better organization—the note content was portable, so the mounting loss was mainly familiarity.

When defending against competitors: Increase genuine accruing benefits, not artificial lock-in. Artificial lock-in breeds resentment and motivates users to switch when any alternative appears.

The data portability paradox: Offering easy data export actually increases trust and reduces churn anxiety. Evernote’s excellent export options probably kept users longer than lock-in would have.

Question: How do mounting losses interact with network effects? If your product has strong network effects (Level 3), do mounting losses matter less?