Fruits Inc 3

Current State

Focus

Clearer Gameplay Presentation + Production Ramp

Validation Question

Can Peggy’s Party’s basic moving-entity, capacity-limited idle-management loop produce stronger first-minute comprehension and engagement when reframed as a visible Fruits Inc farm-to-juice-to-jelly-to-pie value chain?

Setup

ready

Primary Reference

Fruits Inc

Reference Thesis

Use Fruits Inc’s readable farm-to-industry fantasy and product-value chain as the primary reference: four harvested fruits can be sold raw or processed successively into juice, jelly, and pie, with each transformation increasing value and the final pie being sale-only.

Open Questions

• Do clearer action cues and a more balanced farm, Juice factory and fleet progression improve opening comprehension and sustained production compared with fruitsinc004.zip?
• Which build(s) and date range does the Game Events export cover?
• What changed between0051 and0052, including analytics inclusion?
• Can the missing31 progress rows and popup visible/interact data be exported?

Next Actions

• Obtain full event export with build/date filters and popup interactions/timing, separate0051 recordings, and repair normalized experiment links before final verdict. Do not change gameplay based solely on this provisional snapshot.

Latest Findings

• Latest Poki Player Fit for fruitsinc0052.zip: average session 120.8s, p90 250.1s, median 50.0s, under-3m sessions 85.2%, Poki start rate 76.0%.
• 004: mean 113.02s, median61s, p90 220s, 3-minute reach13.4%; 0051:108.71s/50s/231.8s/14.2%; 0052:120.8s/50s/250.1s/14.8%. Small tail gains do not establish a statistically reliable win.
• Pasted opening events: Apple build311/433, dispatch255/311, collect215/255, Apricot build180/215, Juice build124/180, dispatch98/124, collect91/98.
• Collection exposed-to-collected: Apple215/231, Juice91/95. Insects entered159/168, treatment completed146/159, subsequent collection124/146. Customization confirmed110/119, subsequent Juice dispatch98/110.
• Loading completes1108/1437 (77.1%); left328/1437 (22.8%). Loading is a separate issue; event and Player Fit denominators cannot be pooled.
• Event snapshot contains20 of51 listed progress rows, no exact version/date filters, and unresolved start/complete/left counts. Category totals are not established unique-session counts.
• Latest recordings import mixes002/003/004/0051 and lists no0052. Only aggregate recording metadata was accessible through exposed MCP; no new video review or build-specific annotation conclusions were made.
• Experiment mapping documented, but normalized associations remain inconsistent:0051/0052 unassigned; old002/003/004 and mixed import linked to latest. Exposed MCP has no record-level retagging tool.
• Prioritize reviewing the first construction interaction and the transition from two farms to Juice. High collection/feature completion among exposed players makes a broad feature rework a weaker first choice, but selection effects and incomplete aggregation prevent causal conclusions.

Updated

2026-10-17T02:11:05.342664Z

Active Hypotheses

A visible Fruits Inc production chain will give Peggy’s Party’s movement, waiting, processing, collection, and upgrades clearer causality and a stronger aspirational goal, leading players to understand the first action pipeline faster and remain engaged longer than in the party fantasy.

Peggy’s Party’s build-first onboarding still held a 41-second median and did not reverse repeated sub-three-minute sessions. The implementation spine is functional, but the party activities did not make processing, throughput, or upgrades compelling enough. Fruit batches visibly gaining value through successive facilities should make travel and station progression self-explanatory while preserving the already-built systems.

Meaningful variation after the opening—a brief farm rescue, visible product identity, or earlier useful fleet growth—will encourage continued production and improve sustained engagement beyond the two-farm opening; separate builds will identify which addition earns inclusion.

003 increased median session from 42s to 60s and average from 119.24s to 127.98s, while three-minute reach stayed 15.2%. This supports testing post-opening variety, but is not causal proof and may include sample noise.

Clearer actionable guidance, visible truck parking and a progression ramp that establishes Apple, Apricot and Juice before farm upgrades will reduce early confusion and production bottlenecks, helping more players continue beyond the opening.

The 004 recordings prompted presentation and pacing changes. This experiment evaluates the combined revision; it cannot isolate the contribution of any individual change.

Latest Experiment

Clearer Gameplay Presentation + Production Ramp

draft

Do clearer action cues and a more balanced farm, Juice factory and fleet progression improve opening comprehension and sustained production compared with fruitsinc004.zip?

Covers fruitsinc0051.zip and fruitsinc0052.zip. Túlio confirms identical gameplay;0052 adds analytics only. Treat session differences as sample variation, not evidence of a gameplay intervention between these builds. Game Events are expected from0052, pending explicit export filter verification. Follow-up proposal focuses on first construction and Apricot-to-Juice comprehension.

Expected Result

Look for improved opening task completion and fewer confused or blocked interactions in recordings, alongside improved median session duration and three-minute reach relative to 004. Report mean and tail duration as supporting measures. No numeric success threshold was preregistered; preliminary results are not a final verdict.

Prototype Scope

Distinct farm/factory construction lots and visible truck parking; animated actionable hints with safe screen placement, regular cutoff at level 10 and first-time insect/customization exceptions; fleet then factory then farm upgrade hint priority; construction/upgrade indicators and construction/disinfection celebrations; clearer insect treatment presentation and truck-clearance interactions; Apple → Apricot → Juice opening economy; rebalanced farm/Juice upgrades, factory parking-first upgrades and faster fleet capacity ramp; waypoint-based truck routing and portrait adjustments. Landscape HUD queue clearance is a follow-up build revision and should be identified separately in evidence.

Non-Goals

• Popup redesign, deferred to a later pass
• Ads and rewarded upgrade offers
• Jelly and Pie economy rebalance
• Changing the original adaptation or insect/customization experiments or reassigning their historical evidence
• Attributing a combined experiment outcome to a single change
• Assuming the new analytics implementation shipped in an earlier test build

Evidence Capture

next pass

Assess first Apple construction and Apricot-to-Juice transition. Authored construction-popup affordable indicator and Brooke message panel checkpointed on codex/fruits-inc; runtime integration proposed, not implemented. Peggy work remains isolated.

secondary comparator

fruitsinc003.zip; historical context only

known code checkpoints

{'version': '0.05', 'commit': '0af5060', 'scope': 'Presentation, progression, guidance and waypoint routing'} · {'version': '0.05.1', 'commit': '6d7bc4f', 'scope': 'Additional landscape truck queue clearance from HUD'}

new evidence status

Both 0051 and 0052 Player Fit summaries are completed and available. User confirms both belong to this experiment; compare separately. Their normalized records currently have experiment_id null.

association audit

At review, normalized 002/003/004 records and newest mixed recording import were linked to this experiment, while 0051/0052 were unassigned. This evidence plan records correct membership, but exposed MCP tools cannot edit Player Fit associations or extract/import version-filtered recording subsets. Those record-level repairs remain pending.

primary baseline

player_fit_id · mean_seconds · experiment_id · three_minute_reach_percent · p90_seconds · build · median_seconds

recordings

Latest 41-row mixed import contains versions 002, 003, 004 and 0051; no 0052 version is listed. Do not attribute mixed summary tags or duration to 0051. Older source imports remain historical; version-filtered 0051 recordings are required for qualitative assessment.

analytics

Túlio confirms 0052 added analytics only; gameplay is identical to0051. Supplied Game Events are expected to be from0052; exact export filters and missing rows still need verification.

build mapping

{'player_fit_id': 'b7442b1c-eed9-46ac-b952-ea9a03f738ee', 'record_experiment_id': None, 'poki_test_id': 'ac4a2465-936b-43ed-bf82-a973a953eb8c', 'build': 'fruitsinc0052.zip', 'poki_version_id': 'cab54041-f262-4ac0-b605-c2ab19b31e6a'} · {'player_fit_id': 'ce77a160-3517-4154-9a18-57aecda9009a', 'record_experiment_id': None, 'poki_test_id': 'd5846f59-25d3-4e86-ad7e-cea750f3fa52', 'build': 'fruitsinc0051.zip', 'poki_version_id': '34ce28d1-b6c5-4bdc-8e14-93b55c792943'}

metrics

Loading completion separately from gameplay retention · Session mean, median, p90 and three-minute reach · Opening Apple/Apricot/Juice progression and first Juice collection · Confusion around dispatch, collection, upgrade availability, insects and customization · Congestion and return to production

Latest Analysis

0051/0052 preliminary read: opening still weak; feature completion comparatively high

0052 improves mean and p90 versus 004 but both new builds have lower median (50 vs 61 seconds). No clear sustained-engagement win. Provisional event completions prioritize investigation of initial Apple build and Apricot-to-Juice transition; they are not final abandonment estimates.

Findings

• 004: mean 113.02s, median61s, p90 220s, 3-minute reach13.4%; 0051:108.71s/50s/231.8s/14.2%; 0052:120.8s/50s/250.1s/14.8%. Small tail gains do not establish a statistically reliable win.
• Pasted opening events: Apple build311/433, dispatch255/311, collect215/255, Apricot build180/215, Juice build124/180, dispatch98/124, collect91/98.
• Collection exposed-to-collected: Apple215/231, Juice91/95. Insects entered159/168, treatment completed146/159, subsequent collection124/146. Customization confirmed110/119, subsequent Juice dispatch98/110.
• Loading completes1108/1437 (77.1%); left328/1437 (22.8%). Loading is a separate issue; event and Player Fit denominators cannot be pooled.
• Event snapshot contains20 of51 listed progress rows, no exact version/date filters, and unresolved start/complete/left counts. Category totals are not established unique-session counts.
• Latest recordings import mixes002/003/004/0051 and lists no0052. Only aggregate recording metadata was accessible through exposed MCP; no new video review or build-specific annotation conclusions were made.
• Experiment mapping documented, but normalized associations remain inconsistent:0051/0052 unassigned; old002/003/004 and mixed import linked to latest. Exposed MCP has no record-level retagging tool.

Interpretation

Prioritize reviewing the first construction interaction and the transition from two farms to Juice. High collection/feature completion among exposed players makes a broad feature rework a weaker first choice, but selection effects and incomplete aggregation prevent causal conclusions.

Open Questions

• Which build(s) and date range does the Game Events export cover?
• What changed between0051 and0052, including analytics inclusion?
• Can the missing31 progress rows and popup visible/interact data be exported?

Next Action

Obtain full event export with build/date filters and popup interactions/timing, separate0051 recordings, and repair normalized experiment links before final verdict. Do not change gameplay based solely on this provisional snapshot.

Latest Player Fit

fruitsinc0052.zip

Imported: 2026-10-16T23:58:34.989644Z · Captured: 2026-10-16T22:04:01Z

Latest Poki Player Fit for fruitsinc0052.zip: average session 120.8s, p90 250.1s, median 50.0s, under-3m sessions 85.2%, Poki start rate 76.0%.

No live performance evidence imported yet.

Latest Decision

None