Find the evidence for your next onboarding, trial or monetization test. Compare the references, the connected journey and the growth decision each workflow supports.
A good-looking paywall is a starting point. A growth decision needs the journey, the changes teams have made, and a reason the idea fits your product.
Lazyweb is built for product growth optimization: captured experiments, connected app trees, screen clustering and recommendations grounded in your app’s goals and constraints.
10 paywall experiments ↓ How the research connects ↓ Compare capabilities ↓
Ali Abouelatta. Worked on retention and monetization at Duolingo. He tracked purchase flows across roughly 220 consumer apps, returned every six months to study changes, and used those patterns to sharpen the roadmap. That research habit became Lazyweb. Ali’s research habit.
Francesco Polizzi. Led growth at Discord and was Head of Product at Photomath, following growth work at Dropbox and Mozilla. The team built Lazyweb around the work of finding, prioritizing and testing product growth opportunities. Founder backgrounds · Photomath interview.
Price framing, trial length, plan structure, fallback offers and CTA choices. Every example includes the captured pair and a test idea.
Curated from Lazyweb’s experiment catalogue. Control / Experiment are catalogue labels for observed product changes; these captures do not confirm randomized A/B tests or conversion lift. Prices are historical.
01 / 10 · Pixelcut · Price framing
The yearly plan switches from $59.99/year to $4.99/month. The annual billing disclosure remains.
Test the display unit; monitor annual purchase conversion and billing-related refunds.
02 / 10 · PictureThis · Trial offer
Seven days becomes 14 days, with “TODAY ONLY” copy. The annual price stays $29.99.
Separate trial length from urgency in your test; measure paid conversion after the trial.
03 / 10 · The New Yorker · Trial offer
The monthly trial shrinks to one week while the annual plan offers four weeks.
Test plan-specific trials; measure plan mix, trial-to-paid conversion and cancellation.
04 / 10 · Any Distance · Pricing
The annual price changes from $30 to $50; the trial CTA and comparison remain.
Test willingness to pay; measure net revenue alongside purchase conversion.
05 / 10 · Photomath · Plan structure
Horizontal cards become vertical rows. The selected annual price falls from $69.99 to $39.99.
Treat this as a bundle of changes; isolate plan layout before attributing an effect to it.
06 / 10 · Hallow · Fallback offer
The one-month option drops from $3.99 to $0.99; a scholarship link and testimonial appear.
Test an accessible fallback; monitor paid retention and movement away from the annual plan.
07 / 10 · Gentler Streak · Trial offer
Seven days becomes one month; the yearly price changes from $49.99 to $54.99.
Test the trial and price separately; compare revenue after trial and renewal retention.
08 / 10 · Crunchyroll · Offer and CTA
The monthly renewal changes from $11.99 to $13.99; “Start Subscription” replaces the trial CTA.
Do not credit the CTA alone. Monitor starts, paid conversion and net subscription revenue.
09 / 10 · Genie · Trial control
An “Enable Free Trial” toggle and new benefit list replace the reviews-led paywall.
Test trial choice against a fixed offer; monitor trial use, paid conversion and refunds.
10 / 10 · Halide · Price framing
A $0.99/month headline becomes $19.99/year. “Continue” replaces the trial-and-subscribe CTA.
Make the commitment clear; measure paid conversion and billing-related support contacts.
Lazyweb connects the captured screen to its journey, related patterns and your product’s context.
Screens, connected flows and new versions.
Visual similarity groups related treatments; experiment labels describe the changes.
Your category, product, competitors and constraints narrow the research.
A cited reference, a hypothesis, a metric and guardrails.
Conceptual illustration of the research layers. Recommendations are hypotheses to test, not proof a pattern works.
| Research job | Lazyweb | Mobbin | Refero |
|---|---|---|---|
| Captured paywall experiments | Control / Experiment pairs, dates, what changed and a test hypothesis. | Screens and version history; dedicated experiment catalogue not documented. | Screens and flows; dedicated experiment catalogue not documented. |
| Connected app trees | Screens, actions and branches in one app map. | Step-by-step flows. | Step-by-step flows with goals, actions and responses. |
| Changes over time | Screen and flow versions plus annotated product changes. | Version history is included on paid plans. | Version-history workflow not documented in the checked MCP tools. |
| Pattern discovery | Screen clustering, experiment categories and similar references. | Search, categories and agent synthesis. | Styles and visually / functionally similar screens. |
| Fit to your product | Product context, direct competitors and similar companies shape research. | Provide your context to the connected agent. | Provide your context to the connected agent. |
| Prioritized growth report | Your screen, references, test ideas, success metrics and guardrails. | Agent research from references; a dedicated growth-report workflow is not documented. | Agent research from references; a dedicated growth-report workflow is not documented. |
Checked October 5, 2026. “Not documented” means the reviewed first-party pages do not establish that workflow, not that the feature cannot exist. Mobbin plans · Mobbin MCP · Refero tools.
Examples inspected October 4, 2026. Mobbin and Refero also offer flows and agent research; these selected screens are not a benchmark of their full capabilities.
In this August 2026 Cal AI capture, account choices come before the trial invitation, followed by a reminder promise conditional on notifications. Read the sourced answer.
Open the ordered capture · Inspect the full app tree
Growth hypothesis: trial reassurance and annual-plan emphasis are separate levers. The reminder has a notification dependency; the selected annual plan has a billing commitment. Choose a lever that matches your bottleneck, then measure trial starts, paid conversion and retention separately.
Inspect the full app tree for the later plan selector and App Store sheet. The capture does not establish a completed subscription or delivered reminder.
Brainly’s January 2026 capture added post-trial price and cancel-anytime copy under the trial CTA. Inspect the documented changes.
First-party access pages checked October 5, 2026. Public examples and installed reports are different outputs.
| Tool | Free output | Paid access | Verify |
|---|---|---|---|
| Lazyweb | This ten-example sample, public answers and MCP setup. Installed non-Pro accounts receive partial reports. The full experiment library is plan-gated. | Pro receives full reports with priority capacity; Team includes Pro access. | Setup and access · Plans |
| Mobbin | Limited apps and sites; up to three collections. Flows, search and MCP are excluded. | Pro includes flows, search and MCP. Team and Enterprise also include API access. | Plans · MCP |
| Refero | Limited search results and library preview. MCP is excluded. | Pro, Team and Lifetime include MCP: up to 8,000 tool calls per monthly usage period per licensed user. | Plans · MCP docs |
Choose Lazyweb when the next decision is which growth change to test in your product. Choose Mobbin for broad manual screen and flow research, or Refero for web and iOS references, styles and Research mode. Both also support agent research. Lazyweb does not yet offer web-app flows; verify the exact platform and product coverage you need.
Flow research workflow · Experiment research workflow · Growth Report workflow
AI product growth toolkit that connects real screens, flows, app trees and captured changes to opportunities worth testing in your product.
A curated library of shipped iOS and web app screens, websites and step-by-step flows for benchmarking UI patterns.
A curated web-and-iOS reference library with screens, user flows, styles, Research mode and an official MCP.
| Criterion | Lazyweb | Mobbin | Refero |
|---|---|---|---|
| Best for | Use Lazyweb when deciding which onboarding, trial or monetization change to test in your product. [1] | Use Mobbin for broad manual iOS and web app-flow research, with a paid plan for flows, search or MCP. [14][15] | Use Refero for web and iOS references, styles and Research mode, with a paid plan for MCP access. [26][50] |
| Pricing | Human browse and MCP setup are free. Pro is $39/month for the complete toolkit except competitive tracking. Team is $99/month for up to five seats and competitive tracking. [4] | Limited free tier. The October 4, 2026 pricing page lists Pro at $10/month and Team at $16/member/month, billed yearly; Enterprise is custom. Confirm local checkout totals. [14] | Limited free preview. Pro, Team and Lifetime include MCP. Individual plan pricing could not be verified on October 4, 2026 because the pricing card did not load; check the live pricing page. [50] |
| Library depth | 281k+ real app screens across iOS apps and marketing pages. [1] | Vendor-stated 621,500+ screens, 323,900 flows and 1,428 apps on the October 4, 2026 homepage; counts are not independently audited. [15] | Vendor-stated 142,000+ screens and 12,000+ user flows on the October 4, 2026 MCP page; these are catalog claims, not independently audited counts. [34] |
| Platform coverage | iOS apps and marketing pages today; web-app flows are not available yet. [1] | The October 4, 2026 homepage advertises iOS apps, web apps and websites. Verify other platform coverage in the current catalog. [15] | Websites, web apps and iOS apps, with screens, flows and styles. Check the exact product and platform in the current catalog. [26][34] |
| MCP / API | Agent-first MCP setup across Codex, Claude Code, Cursor, Antigravity, hosted Claude connector paths, and other agentic platforms. [3] | Official MCP server on Pro, Team and Enterprise; API access on Team and Enterprise. Free excludes both. See current documentation for authentication and usage limits. [14] | Official MCP with OAuth or Bearer authentication. Pro, Team and Lifetime include up to 8,000 MCP tool calls per monthly usage period per licensed user. [26] |
| Agent readiness | Agents research flows and experiment patterns, then use your product context to develop and prioritize growth hypotheses. [3] | Genuinely agent-ready: the official MCP returns screen images, annotations, and metadata with natural-language search inside Claude Code, Cursor, Codex, and v0 — but only with a paid, authenticated account. | Official MCP and a research-first skill for searching products, styles, screens and flows with a paid authenticated account. [26] |
1. Lazyweb
2. Lazyweb Research
3. Lazyweb MCP install
4. Lazyweb pricing