Handling disputes on delivered work, in numbers

On a Tuesday evening in Marseille, a €380 translation job for a commercial lease triggered WEVONE’s dispute pipeline over four omitted sub-clauses. The buyer claimed non-delivery; the provider claimed the sub-clauses were marked optional in the initial task specification. Funds sat in structural escrow, locked by a system that requires neither party to trust the other, but demands that both define what 'done' looks like.

Across 1,420 completed micro-missions logged during our Q3 pilot in France and Belgium, exactly 60 missions—4.2%—entered the dispute state. In traditional gig platforms, disputes are treated as customer service failures to be buried or hand-waived by offshore support reps. At WEVONE, every disputed euro is a telemetry signal revealing where human intent diverges from written agreements.

Here is what the numbers actually tell us about why work breaks down, how structured escrow resolves it, and where our automated pipelines hit their current limits.

The Anatomy of the 4.2 Percent

When a job stalls, platforms usually blame bad actors. Our data shows a different pattern. Out of the 60 disputed missions in our reference dataset:

• 58% (35 cases) stemmed from Scope Ambiguity. The client assumed a deliverable included sub-tasks (e.g., 'fix website bug' assumed to include mobile layout re-design) that were never explicitly detailed in the mission manifest.
• 24% (14 cases) were caused by Time-Window Slippage. The provider delivered the work, but missed an agreed deadline by more than 12 hours without issuing an automated delay request.
• 12% (7 cases) involved Technical Incompatibility. Files were delivered in formats or environments the buyer could not run or open.
• 6% (4 cases) were Outright Abandonment or Fraud. The buyer attempted to take deliverables off-platform before approving funds, or the provider uploaded placeholder files to trigger payout timers.

Malicious fraud represents less than a tenth of all platform friction. The dominant driver of disputes is administrative vagueness—two parties operating under different definitions of completion.

Where Proof Meets Scope: The Escrow Lock

When a user initiates a mission on WEVONE, funds do not transfer to the provider, nor do they stay in the buyer's bank account. They move into WEVONE’s transactional escrow ledger. The funds are bound by a programmatic release contract: payout occurs automatically 72 hours after work submission unless the buyer raises an explicit objection supported by evidence.

If an objection is raised, the 72-hour clock freezes. The transaction enters the Dispute State, triggering three mandatory protocol steps:

1. Deliverable Freeze: The system snapshots the exact files, text blocks, or time-stamped proof-of-work submitted prior to the dispute flag.
2. Context Log Assembly: Mia pulls the complete, immutable message history, task manifests, and milestone revisions associated with that transaction ID.
3. Evidence Requirement Window: Both parties are granted 24 hours to submit specific counter-evidence matched against the initial specification.

This architecture fundamentally alters user behavior. Because escrow funds are held neutral, neither side can hold money hostage or force a refund through silence. The incentive pivots immediately toward submitting structured proof.

Mia’s Parser vs. Emotional Arguments

In conventional dispute tickets, users write long, impassioned paragraphs about how hard they worked or how rude the counterparty was. Mia ignores emotional posturing entirely.

As WEVONE's underlying infrastructure, Mia’s context memory evaluates disputes by running the original job brief through a semantic diff against the submitted work and chat logs.

Consider a concrete example from our Lyon pilot: A local business paid €600 for a local SEO setup. The buyer opened a dispute claiming the provider 'did nothing.' Mia’s context parser audited the task record, identifying that the provider had uploaded five time-stamped screenshots of Google Business Profile verification, updated XML sitemaps, and verified indexation logs—all actions specified in the initial contract. The buyer’s counter-claim that 'traffic didn't double in three days' was cross-referenced with the contract scope, which made no performance guarantee on traffic metrics.

Mia auto-resolved the case in favor of the provider within 14 minutes of evidence submission, releasing €600 from escrow and charging the buyer a standard non-justified dispute processing fee.

Overall, of the 60 disputes in our pilot dataset:

• 82% (49 cases) were resolved by Mia using context memory and structured log parsing without human staff intervention.
• 11% (7 cases) were settled through mutual compromise during the 24-hour evidence window, where parties agreed to a split release (e.g., 80% payout for 80% completed scope).
• 7% (4 cases) required manual human review by WEVONE trust specialists.

The Real Cost of Friction and Current Limitations

We must be precise about what this system can and cannot do today. WEVONE is early, and automated arbitration is not magic.

Our current parser operates with high accuracy on digital deliverables (code, text, structured files, digital verification). It struggles significantly with subjective real-world local services. In a test case involving physical painting under the Mission universe, a buyer disputed a living room wall coat, claiming the finish was uneven. Mia cannot parse the physical texture of dried paint from a low-resolution smartphone photo. That case required human review, costing WEVONE €42 in operational overhead to settle a €150 job.

To address this limitation, we are currently testing structured photo-verification protocols for physical missions in our beta environments. Until those protocols mature, physical service disputes remain our largest operational bottleneck.

Furthermore, dispute outcomes directly feed our internal Contribution Score. A user who triggers unjustified disputes loses karma capital, which restricts their visibility and temporarily raises their transaction fee tier. This prevents bad-faith actors from using the dispute pipeline as an informal discounting tool.

By replacing arbitrary customer support with deterministic escrow triggers and clear context tracking, we convert vague grievances into measurable scope checks. Fraud shrinks, resolution times drop from days to minutes, and the cost of trust drops toward zero.