How we test

1. App discovery

We discover apps to test through:

• Reader submissions via the contact form — about 40% of our test queue.
• App store monitoring — iOS App Store and Google Play AI-companion category, monthly sweeps.
• Search trend data — what people are searching for in the AI companion space.
• Industry monitoring — funding announcements, product launches, beta lists.
• Competitor mentions — apps mentioned in other reviews we trust.

Affiliate program existence is NOT a criterion for testing. If an app is popular or interesting, we test it whether or not we can monetize the review.

2. Testing protocol

Every app gets the same testing protocol:

Session 1 (1-2 hours): First impressions

• Sign up with a fresh account (separate email, separate payment method)
• Complete onboarding without skipping
• Try 3-5 default companions for ~20 messages each
• Test claimed free tier features end-to-end
• Note: signup friction, paywalls, dark patterns

Session 2-3 (2-4 hours): Depth testing

• Pick one companion. Have an extended conversation (100+ messages).
• Test memory across context: "What did I tell you about [X]?"
• Test character consistency under stress (sarcasm, contradiction, emotional pivot).
• Test NSFW capability across the spectrum (vanilla, edge cases, refusals)
• Test customization depth (personality, appearance, voice, kinks)

Session 4+ (2-5 hours): Feature breadth

• Test image generation (count seconds, count successful renders, note quality)
• Test voice features (synthesis quality, latency, NSFW behavior)
• Test mobile apps if available
• If paywalled features matter, upgrade to Premium for at least one month

Session 5+ (1-2 hours): Receipts

• Document exact hours spent, messages sent, dollars spent
• Screenshot key interactions for the review
• Cancel or downgrade subscription if no longer needed

3. Scoring system

Every app is scored 0-10 in five categories. Final score is a weighted average:

• Dialogue (30%): Character consistency, conversational depth, memory, handling of edge cases.
• NSFW capability (25%): Uncensored quality, image generation, voice features. SFW-only apps score 0 here.
• Privacy (15%): Encryption, deletion policy, third-party sharing, transparency reports.
• Customization (15%): Depth of character creation, personality controls, kink/preference settings.
• Value (15%): Real monthly cost (including tokens/extras) vs feature delivery.

Affiliate program: 0% weight. An app's commission rate has no influence on its score.

4. What the scores mean

• 9.0-10: Best in class. Genuinely worth your time and money. Editor's pick territory.
• 8.0-8.9: Very good. Some weaknesses but strong overall delivery.
• 7.0-7.9: Good with notable gaps. Worth considering if specific strengths match your needs.
• 6.0-6.9: Mediocre. Better options exist unless one specific feature is decisive.
• 5.0-5.9: Poor. Significant problems. Not recommended.
• Below 5.0: Avoid. Either misleading marketing, broken product, or harmful practices.

5. Transparency requirements

Every Mr.Pimp review publishes:

• Exact hours tested — visible in the article header.
• Exact messages sent — visible in the article header.
• Exact dollars spent — visible in the article header.
• Test log — dates of original test and all re-tests, shown in sidebar.
• Verdict bullets — three quick takes (positive + warnings) for skimmability.
• Pros and cons — explicit, specific, no hedging.
• Hot take — the editor's straight-talk paragraph that summarizes the verdict in plain language.

6. Re-testing cadence

Reviews are re-verified every 3-6 months. We check:

• Pricing changes (token costs, subscription tier prices)
• Feature additions or removals
• Companion library changes
• Privacy policy changes
• Major UX changes

If significant changes warrant rescoring, we re-test fully. Re-test count is shown on every review ("3× re-verified").

7. Corrections and disputes

If you find an error in a review, email corrections@mr-pimp.com. We respond to all credible correction requests within 7 days. Major corrections are flagged with a public changelog entry on the affected review.

If you are an app vendor: we accept correction requests on factual matters (pricing, features, etc.) immediately. We do not accept requests to alter editorial opinion, score, or framing.

8. What can change a score

Score changes between editions happen because of:

• Product changes: features added or removed, pricing changes, paywall additions or removals.
• New competitive context: if a new app raises the bar in a category, existing apps may be rescored relative to the new ceiling.
• Methodology refinements: we occasionally adjust weightings or add subcategories. When this happens, all affected reviews are rescored and changelog entries note the methodology shift.

Score changes are never driven by affiliate program changes.

9. Affiliate disclosure

Outgoing links on Mr.Pimp routed through /go/{vendor}/ are affiliate links when the vendor has a program. We earn a commission when readers sign up via these links. The commission has no influence on rankings.

Apps without affiliate programs are linked using /go/{vendor}/ redirects too, but those redirects are not monetized.

The rel="sponsored noopener" HTML attribute is applied to affiliate links per Google's outbound link guidelines.