Why manual review is the default

Auto-submit is the default for a reason. Chargeback evidence windows are tight — Stripe surfaces a 7–21 day deadline against the underlying Visa 30-day / Mastercard 45-day window. If the merchant forgets to respond, Stripe (or Shopify) auto-submits the form anyway, and the merchant loses by default. Auto-submit-by-default tools sell the “set it and forget it” promise: you don't have to remember anything; the AI handles it.

For a high-volume merchant with 50+ disputes/month, the math seems obvious. Spending 5 minutes per case = 4+ hours/week of high-skill ops time. Automating that away looks like pure win.

Read the 1-star reviews on every auto-submit tool — Chargeflow, ChargePay, Disputifier, Stripe Smart Disputes, Shopify's built-in form. The complaints concentrate in three patterns:

1. Premature submission before evidence is complete. Verbatim from a Chargeflow App Store review: “Chargeflow submitted my chargeback before their own 5-day evidence window closed, so my proof was never included and I lost the case.” Stripe disputes can only be submitted once. Premature = lost.
2. Submission on a customer who's actually right. The customer ordered the wrong size, you said you'd send a replacement, the replacement was delayed, the customer disputed. Auto-submit fights it; you would have refunded. The dispute now counts against your chargeback ratio (gates Visa VDMP / MATCH programs) and you've antagonized a customer.
3. Submission on a case where you don't have the evidence. The Shopify fulfillment didn't sync, the carrier webhook didn't fire, the Zendesk ticket lived in a closed-merged thread the AI couldn't see. The submitted packet is missing the one piece that would have won. Single-submission = lost.

A merchant who reads each case for 30 seconds before approving would have caught all three.

Stripe charges a $15 (US) per-dispute receipt fee, non-refundable, regardless of outcome. If auto-submit loses a case it shouldn't have submitted, you're out:

• The original transaction amount (the chargeback pulled the funds)
• The $15 receipt fee
• The dispute counts toward your chargeback ratio (1.5% threshold triggers Visa VDMP / Mastercard MATCH placement)
• The customer relationship (refund-then-fight is the worst-of-both)

The expected value of “30 seconds of merchant review before submission” against this downside is enormous. The only counter-argument is that the merchant doesn't actually do the review. Guarded automation does not remove that downside; it makes the acceptance criteria explicit and auditable.

Settings drift. Toggles get flipped. New employees don't know what they're doing. That is why new Victava accounts start on hold: the merchant reviews each response before anything is submitted.

Merchants who deliberately choose automation can enable Auto-Pilot. Eligible disputes must clear the merchant's minimum score, amount cap, waiting period, deadline rule, and packet validation. Inquiries, missing scores, incomplete packets, and specified high-risk cases do not auto-submit. Every decision is written to the audit trail.

Auto-Pilot can be paused or turned off at any time. Manual review remains available for every merchant and for any case outside those configured limits.

Human approval fails when the workflow is slow. We've optimized:

1. Case brief preview. We render a single-page summary: dispute reason → top 3 evidence items → predicted win/loss → recommended action. Reading takes ~30 seconds.
2. One-click approval. If the brief looks right, click Approve. If not, the Edit button opens a section-level editor; the Skip button accepts the dispute (1-click); the Add Files button lets you attach a custom screenshot or PDF.
3. Reminder cadence. 72h, 48h, 24h email reminders before the Stripe deadline. The 72h reminder is when most approvals happen.
4. Bulk approval. For high-volume merchants, the dispute list view supports multi-select with batch-approve — you scan the brief inline, click checkboxes for the obvious-win cases, hit Approve N selected.
5. Audit trail. Every approval records who clicked it, when, and what was submitted. If a dispute is lost and you want to investigate why, the trail tells you exactly what evidence went to Stripe.

For a typical SMB Shopify merchant with 1–10 disputes/month, total time spent in our dashboard is under 5 minutes/week. The approval gate isn't the bottleneck.

Some merchants will batch-approve without reading. That's their choice; we can't force engagement. Two design decisions mitigate it:

• Click-to-approve, not link-to-approve. The Approve button is in the dispute detail page, not in the email. Reading the email and clicking through requires opening the case at minimum.
• Risk-flagged disputes block bulk approval. If our scorer flags a dispute as low-confidence-win or refund-overlap, bulk approval is disabled for that case — the merchant has to open it explicitly. This is the highest-leverage quality gate.

For merchants who genuinely want to approve every case in 2 seconds without reading: the predicted win-rate is still surfaced, the refund-overlap warnings still fire, and the bulk approval still checks the cases. Their floor is higher than auto-submit even if they don't use the tool perfectly.

Keep Auto-Pilot off if:

• You have 50+ disputes/month and the math on review-time-vs-recovery doesn't close even at 30 sec/case.
• Your evidence sources or refund policy change frequently.
• Your team wants a person to check customer context before every filing.

For a median DTC Shopify merchant with 1–10 disputes/month, manual review is often the clearest starting point. Higher-volume teams can enable Auto-Pilot gradually with conservative limits and use the decision log to verify its behavior.