# HarborMug review decision: fix the seal first

Prioritize a lid-and-gasket improvement for HM-BLUE-001, the 500 ml stainless insulated mug. Leaks are the most frequent complaint and undermine its core function. First compare correctly assembled current lids against a supplier gasket/lid candidate; use those results to decide whether a product change or clearer assembly instructions address the failure. The next priorities are cupholder fit and gasket cleaning.

These are synthetic, deliberately constructed reviews, not a random customer sample; all figures describe this batch only. No product changes, customer outreach or performance tests have been executed.

## Pain summary

Quotes below are exact excerpts; the evidence file retains complete original text, including example markers.

| Rank / pain | Evidence / frequency | Type / root-cause hypothesis | Action | Validation | F × S × Fix × D = score |
|---|---|---|---|---|---|
| 1. Leaks when tilted | “lid leaks when tilted”; R01–R08; 8/30 (26.7%) | Function not met; gasket seating, seal compression or closure geometry may fail. None is established. | Inspect closure and gasket assembly; compare current lid with a supplier seal-system candidate. | Controlled tilt tests before and after cleaning/reassembly; compare normal-user and trained assembly. | 3 × 3 × 2 × 2 = **36** |
| 2. Does not fit cupholder | “too big; doesn't fit my cupholder”; R09–R14; 6/30 (20.0%) | Size/fit; base profile may exceed usable holder space. Vehicle details absent. | Add measured base/height fit guidance; evaluate a narrower-base successor only against measured target holders. | Physical fit matrix and shopper measurement-comprehension check. | 2 × 2 × 3 × 2 = **24** |
| 3. Gasket is hard to clean | “gasket is hard to clean”; R15–R19; 5/30 (16.7%) | Experience; inaccessible groove or awkward removal may impede cleaning. | Trial a supplier-approved removal/reassembly guide; request accessible seal geometry if needed. | Residue inspection, cleaning time, reassembly success and post-clean leak test. | 2 × 2 × 2 × 2 = **16** |
| 4. Paint chips after few uses | “paint chips after few uses”; R20–R23; 4/30 (13.3%) | Durability/life; coating adhesion or wear resistance may be weak. | Request coating process evidence and compare coated samples under repeated hand washing and handling. | Standardized wear cycles, photographs and chip counts versus current coating. | 2 × 2 × 2 × 1 = **8** |
| 5. Care restrictions are unclear | “care instructions unclear; hand-wash restriction was hidden”; R24–R26; 3/30 (10.0%) | Not as described; care disclosure may be hard to find. | Put hand-wash and microwave restrictions beside purchase information and on the insert. | Shopper recall and care-task comprehension; later track care-related complaints. | 1 × 2 × 3 × 1 = **6** |

Frequency bands are predeclared here: high = at least 25% of unique reviews (3); medium = more than 10% but below 25% (2); low = at most 10% (1). Severity: 3 = core function failure or plausible consequential damage; 2 = meaningful use restriction or premature deterioration; 1 = minor inconvenience. No injury is reported. Fixability: 3 = straightforward information change; 2 = supplier/sample iteration needed; 1 = major unresolved redesign. Differentiation: 2 = measurable useful benefit; 1 = basic expectation. Scores are judgment-based prioritization, not measured engineering facts.

Fit's fixability score applies to the immediate fit-guide action, not a base redesign. Cleaning assumes supplier review; gasket removability is not known. Leak severity reflects loss of containment, not evidence of a burn incident. Holding differentiation constant still leaves leaks first (F × S × Fix = 18 versus fit 12, cleaning 8, chips 8 and information 6). Information changes can proceed cheaply while seal development runs.

## Evidence and denominator rules

The source contains **32 rows**. Source rows 31 and 32 exactly repeat R01 and R09 respectively, across every original field. Excluding those two repeats leaves **30 distinct reviews from 30 distinct synthetic customer identifiers**. Original rows remain in `original_reviews.csv`; `review_evidence.csv` maps every row to its retained review and explicitly marks exclusions.

Identical complaint wording across different customers is not sufficient reason to merge their reviews. Removing the example-number suffix for comparison yields six wording groups, but those groups are not six independent buyers. The normalized text is retained as a separate field; it never overwrites original evidence. Same-customer near-duplicates would require manual comparison of SKU, text and chronology; none remains here. There are no follow-ups.

The primary denominator is all 30 retained reviews, including four positive desk-use reviews (R27–R30). There are 26 complaint-bearing reviews: leaks 8/26 (30.8%), fit 6/26 (23.1%), cleaning 5/26 (19.2%), chips 4/26 (15.4%) and unclear care 3/26 (11.5%). Ratings are 12 two-star, 14 three-star and four five-star reviews. The resulting 26/30 share of ratings at or below three stars is a sample description, not the store's bad-review rate.

Tagging allows overlapping themes. Gasket cleaning and unclear care both carry the broad cleaning theme: **8/30 (26.7%)** distinct reviews. They retain separate actionable pain labels, because difficult physical cleaning and missing care information require different fixes. Leak reviews also carry a seal-system tag; fit carries size-fit; chips carries durability. No review explicitly reports both leakage and difficult cleaning. Do not infer that overlap from a plausible shared gasket cause. Theme totals must be computed from unique review IDs rather than added as independent customers. Original `manual_pain` values are retained and the final labels reflect the text, including the positive reviews.

## Ordered improvement backlog

| Order | Change and owner role | Cost / cycle assumption | Intended outcome and release condition |
|---|---|---|---|
| 1 | Product/QC: compare current lid/gasket with supplier candidate; document assembly tolerances | Medium; one sample iteration, quote pending | Reduce tilt leakage; candidate must pass the seal gate below within the cost cap. |
| 2 | Operations: publish measured fit diagram and purchase guidance | Low; next content revision after measurement | Reduce mismatched cupholder purchases; users can correctly assess fit. |
| 3 | Product/operations: validate cleaning and reassembly guide, then consider accessible seal design | Low for guide; medium for geometry; supplier approval required | Easier cleaning without inducing leaks; pass residue and reassembly checks. |
| 4 | Supplier/QC: improve coating process or select alternate coating | Medium; sample iteration | Less early chipping; beat current control in matched wear checks. |
| 5 | Operations: disclose care limits prominently | Low; next content revision | Buyers recognize hand-wash-only care before purchase. |

These are five actions for five supported pains; no additional complaints are invented. Priorities describe business impact, while low-cost content edits may run concurrently. Structural lid, coating and base changes require sample approval; a supplier switch is allowed but should depend on demonstrated performance.

The current unit cost is **$8**, against a **$9 cap**, leaving **$1 per unit** for the combined product/accessory changes. Obtain a comparable quote including any added gasket or insert; do not allocate $1 separately to every fix. Tooling and development costs need separate approval and are unknown. The $24 price and $4 shipping cost do not establish net profit because other expenses are absent. Weekly order/refund totals cannot attribute returns or leakage to a particular batch and are not a validation baseline.

## First-change validation protocol

All quantities and gates below are proposed operating criteria, not completed measurements or formal certification standards.

1. **Diagnose before selecting the fix.** QC should inspect ten current units across available lots for damaged/missing gaskets, closure alignment and seating. Use cool water in controlled containment. Record fill level, lid position, leak location, orientation and assembly method. If trained correct assembly still leaks, prioritize hardware. If ordinary users fail while trained assembly passes, test a revised illustrated guide before committing to tooling. A guide cannot be declared sufficient merely because trained staff can seal the mug.
2. **Compare hardware.** Test 30 current and 30 candidate units, spread over three available lots per design. Use the same documented fill level, closure procedure and five-minute hold at 45°, 90° and inversion. Dry each exterior first; record drops and mass change on a 0.1 g resolution scale. A proposed unit pass is no visible droplets and no detectable loss at that resolution in every orientation. If the mug is not designed for inversion, record that limitation and restrict carrying claims accordingly; tilt failures still require resolution.
3. **Check repeatability and cleaning interaction.** Repeat the seal test after 20 supplier-approved hand-wash, drying and reassembly cycles. Have ten users unfamiliar with the design follow the draft guide without coaching. Require all ten to reassemble correctly and pass the leak check; record time and errors. Do not instruct removal until the supplier confirms the gasket is removable and approves the method. Compare visible residue after a fixed, identical food-soil and wash procedure; require no visible residue under the same inspection conditions. This is a cleaning-usability screen, not a microbiological safety claim.
4. **Release gate.** Require 30/30 candidate units to pass both initial and post-cycle seal checks, no worsening of cleaning/reassembly usability, and combined unit cost no more than $9. Record control results alongside candidate results. Any candidate failure pauses release for root-cause review and a new sample round. Zero failures in 30 does not prove a zero field failure rate; the approximate one-sided 95% upper bound remains 9.5% under independent representative sampling, which lot clustering may weaken.
5. **Verify with actual use after approval.** Keep variant and lot identifiers. Follow comparable current and improved cohorts at 7, 14 and 30 days after delivery. Ask at days 7 and 30: “After closing the lid, has liquid leaked when you tilted the mug?” (Yes / No / Have not tried), and “Could you clean and reassemble the lid using the guide?” (Yes / No / Have not tried). Record exposure and nonresponse. A minimum exploratory comparison is 50 tilt-exposed respondents per version; it supports a pilot decision, not a guaranteed detectable effect. Plan a larger test from the observed control rate and a pre-agreed meaningful reduction.

Field metrics: leakage respondents divided by respondents who actually tilted the mug; cleaning failures divided by respondents who tried cleaning/reassembly; leak-related returns divided by delivered units in the same matured cohort; ratings at or below three stars divided by all reviews received for that cohort. Also report response rate, absolute counts, all-cause returns and complaint overlaps. Compare matched 30-day exposure windows and uncertainty intervals. Lower response-based leakage alongside no worsening in returns or cleaning is directional evidence; an uncertain result extends observation rather than proving success. Any consequential leakage report triggers immediate QC review.

For fit, measure base diameter, diameter at holder contact height, handle clearance and overall dimensions on ten units. The supplied 9 × 9 × 19 cm envelope does not establish base diameter. Build a physical holder-fit matrix for five customer-relevant vehicle/holder combinations once those use cases are collected; publish only tested combinations. Five shopper testers should all correctly interpret the guide before release. For coating, compare ten current and ten candidate samples after 30 standardized hand-wash/handling cycles; require no visible chips in candidates and record control performance. For disclosure, five shopper testers should all identify both care restrictions before purchase; revise placement on any failure.

An optional [Rijoy](https://www.rijoy.ai/) validation loop can be evaluated for structured feedback and solved/unresolved segmentation. Confirm suitability before implementation. Any participation reward should be independent of rating or whether the issue was solved. Unresolved cases go to support and the next product review; solved cases may inform consented repeat-purchase follow-up. No account setup, rewards or outreach have been performed.

## Next-supplier selection requirements

| Must-have requirement | Acceptance evidence | Avoid |
|---|---|---|
| Reliable lid/gasket seal after repeated cleaning | Pass the comparative seal and cycle protocol; documented assembly tolerances | Untested “leakproof” claims or instructions used to conceal a persistent seal failure |
| Accessible cleaning path and reproducible reassembly | Supplier-approved method; user task and residue checks | A seal requiring unsupported removal or tools not supplied |
| Measured cupholder compatibility | Dimensioned base profile, tolerances and tested holder matrix | “Fits all cupholders” or substituting overall width for base diameter |
| Coating withstands ordinary hand washing | Matched wear-test photos and counts | Appearance-only sample approval |
| Care and materials are documented | Consistent packaging/PDP instructions; applicable supplier material documentation reviewed before purchase | Unverified dishwasher, microwave or food-contact claims |
| Total revised unit cost ≤ $9 | Comparable written quote including included accessories | Multiple uncosted improvements assumed to fit the same $1 headroom |

Receiving QC should check closure/gasket condition, run the seal screen, verify critical dimensions, inspect coating and verify insert presence. Before the first revised lot, QC should define lot sampling and rejection rules from supplier variability; the development sample is not a permanent lot-acceptance plan.

## Customer-facing English draft

No different target language is specified; English follows the US DTC brief. The following care/fit copy is ready for internal placement review using known product facts:

> **HarborMug 500 ml Insulated Mug**  
> Hand-wash only. Not microwave safe. Overall dimensions: 9 × 9 × 19 cm. Check your cupholder's usable width, depth and clearance before ordering. Cupholder compatibility varies.

The measured base diagram should accompany the final fit guide once available. A gasket removal/reseating tutorial remains conditional on supplier confirmation and user testing; no unsupported disassembly instructions are supplied. Desk-use comfort is a positive theme worth preserving in usability checks, but four synthetic reviews do not justify a public customer endorsement. Do not add “leakproof,” “easy-clean” or universal-fit claims before validation.

## Supporting deliverables

- `original_reviews.csv`: untouched source evidence, including repeated rows.
- `review_evidence.csv`: full original fields, source-row lineage, duplicate decisions, final pain types, overlapping themes and normalized comparison text.
- `pain_calculations.csv`: denominators, counts, percentages and score components for all five pains.
- `analysis_summary.json`: reconciliation totals and cost headroom.

The decision-changing gaps are the untested seal cause, unknown gasket-removal method, absent holder measurements and unquoted supplier changes. These prevent approval of a final hardware design or performance claim, but do not prevent starting seal diagnosis and correcting known care disclosure.