TL;DR: Sellers on large consumer marketplaces lose real money not because they are wrong but because the dispute process is procedural, unforgiving, and staffed by support that stops replying. The winnable product is not arbitration — it is the boring layer that captures proof at shipment, tracks every dispute clock, and makes the case file assemble itself.
The evidence
PainHunt's E-commerce Platform domain shows 114 posts above the 10/15 threshold at an average score of 11.8 — the highest average in this batch — with intensity 7.7/10. Signal arrives mainly through Discourse and app store reviews, with a tail on Mastodon, GitHub, and Bluesky.
The failures described are procedural, not evidentiary. A package was misrouted — dispatched toward one city and delivered to another — with the buyer left holding no usable tracking information and roughly 150 EUR gone. The dispute system then failed on a technicality: a dispute was accidentally cancelled by the buyer, and the platform declined to help reopen it. In another account, after a failed transaction, there was simply no support response.
From the seller side, the language is harsher: support is described as unable to resolve seller issues, and the platform as facilitating fraud while leaving legitimate sellers financially exposed with inadequate protection. A separate strand in the same cluster involves platform app developers describing investigations opened without a formal complaint, repeated infringement accusations, and inconsistent enforcement — a governance complaint with the same shape as the dispute one.
The feature requests are correspondingly modest and telling: a dedicated seller support channel with faster response, third-party fraud protection and dispute resolution, third-party buyer protection, real-time package tracking with intervention alerts, and transparent, consistent enforcement processes. Note the word appearing twice: third-party. The users in this data have already concluded the platform will not fix this and are describing an external product.
Why now
Marketplace support scaled by deflection. Automated flows and self-service replaced humans across the industry, which works for the common cases and fails hard on the edge cases where the money actually is. A mis-clicked cancellation used to be reversible by a person; now it is a state transition with no path back.
At the same time, sellers concentrated their livelihoods on platforms they do not control. An independent seller has no leverage, no account manager, and no appeal beyond the form — and, crucially, no record of their own. Everything that happened to the order lives in the platform's system, presented through the platform's interface, on the platform's schedule.
Cross-border volume made the failure modes more common and less recoverable: more handoffs, more carriers, longer timelines, and dispute windows that expire while a package is still theoretically in transit. The gap between "what happened" and "what the seller can prove within the window" widened, and nothing on the platform side is closing it.
The wedge
Do not try to arbitrate. Be the seller's own record and their clock.
- Capture proof at the moment of shipment. Photos of the packed item and label, weight, carrier acceptance scan, and a timestamped hash — automatically, as part of the existing shipping flow. The single most common way a seller loses is having no contemporaneous evidence, and evidence gathered after the accusation is worth much less.
- Run a deadline clock on every open dispute. Ingest cases, track each platform's response windows, and escalate before they expire. The 150 EUR in this data was not lost to a bad judgment; it was lost to a process that closed.
- Assemble the case file in the platform's own format. Sellers are not short of facts, they are short of time to compile them into the shape the form wants. Generating that package in one click is the daily habit that makes the tool sticky.
- Publish the outcome data. Aggregate anonymized dispute outcomes by platform, category, and failure type. Sellers currently have no idea whether their loss rate is normal, and a credible benchmark is both a retention hook and the strongest possible marketing.
Start with one marketplace and one seller segment — the mid-volume independent seller who loses several hundred a month and has no operations staff. That seller can be reached in the same forums this data came from.
Risks and honest caveats
- You are building on someone else's rules and they change without notice. Dispute windows, evidence formats, and API access are all the platform's to alter. Design so a change degrades the product rather than breaking it, and never let the core value depend on an undocumented endpoint.
- Never promise an outcome. The temptation is to sell "win your disputes." You cannot control the decision, only the preparation. Selling an outcome you do not control creates refund liability and reputational damage the moment a customer loses a strong case.
- Access may be restricted. Some marketplaces are actively hostile to third-party tooling touching order and dispute data. Check terms first — a tool that violates them is not a business, and a seller account suspension caused by your product is worse than the problem you were solving.
- The market is loudest at its least profitable end. The sellers complaining hardest are often the smallest, with the least ability to pay. Price and target for the tier where a few hundred a month in losses justifies a subscription.
- Insurance is a different business. Reimbursing losses looks like the natural extension and turns you into an underwriter with capital requirements and adverse selection. Stay on the evidence side unless you intend to become an insurer deliberately.
How to validate this further
Browse the marketplace and commerce threads in the Pain Point Browser, then pressure-test the evidence-and-deadlines framing with the Idea Validator. Related reading: when the fulfillment app you depended on shuts down and protecting merchants from frozen funds.