Migrating from CardDealerPro to AI Card Vault — a 1,000-card stress test

Editor's note: "R." is a composite — the migration walkthrough below traces the workflow of a real CDP-to-AICV migration session, with the specific dealer details, numbers, and timeline anonymized and stylized. The migration tool itself, the import CSV behavior, and the field-mapping mechanics are all as described.

Two of the dealers I trade with most use CardDealerPro. Both have been on it for over a year. One of them — call him R., he runs a Pokemon-focused store from Pennsylvania — agreed to let me migrate his entire 1,043-card active inventory to AI Card Vault as a stress test, on a Sunday in January 2026, with him watching over my shoulder.

This is what happened. Nothing dressed up.

The starting state

R.'s CardDealerPro account:

• 1,043 active cards across two batches
• All on eBay (his only channel)
• About 65% Pokemon, 30% sports, 5% Magic
• Roughly $14,800 in total list value
• Custom title templates configured (the feature he was paying for the Majors tier to use)
• Custom fields: condition_detail, source_show, my_cost — three he'd set up over a year of refinement

He'd been on CDP for 14 months. The thing pushing him to migrate: he'd added Whatnot two months earlier, was managing the Whatnot product list manually, and was tired of double-selling cards across eBay and Whatnot.

Step 1 — Export from CDP

CDP's export flow is straightforward. Settings → Export → "All Inventory" → CSV. About 8 seconds to generate, 4 MB file, opens in Excel.

What's in the CDP export:

• Card title, description, price, condition, quantity
• eBay listing ID and current status
• Custom fields (condition_detail, source_show, my_cost — all preserved)
• Images as URL references (not embedded; they live on CDP's CDN)
• Batch name + creation date
• Marketplace status per card (eBay: live / sold / draft / paused)

What's NOT in the CDP export:

• The card's actual catalog identity (set + number + parallel). CDP exports their display title but doesn't always export the structured card spec. The vault's importer has to re-identify each card to populate the structured fields.
• Cost basis (R.'s my_cost field captured it, but CDP doesn't have a native cost-basis field).
• Cross-channel state (R. wasn't using anything but eBay, but if he had been, that wouldn't transfer cleanly).

So the export is good but incomplete. The migration tool has to do real work.

Step 2 — Import into AI Card Vault

The vault's CDP importer is built specifically for this CSV format. (We built it after my own first CDP migration — see the manifesto for why I had to do this in the first place.)

The flow:

1. Upload the CDP export CSV.
2. The importer detects the CDP schema and maps fields automatically.
3. For each card, the importer:
   • Pulls the image from CDP's CDN
   • Re-runs the card through the identify pipeline to get structured spec data
   • Preserves the custom title, description, and price
   • Maps CDP's batch into a vault batch
   • Maps custom fields directly (no transformation needed)
4. Surfaces a review queue: cards where the AI's identification differs from CDP's stored title get flagged for manual review.

For R.'s 1,043 cards:

• 1,028 imported cleanly (98.6%)
• 15 went into the review queue. Of those:
  • 9 were correct (AI's read agreed with CDP) — bulk-approved.
  • 4 had different parallel/foil designations. R. reviewed each, picked one.
  • 2 were Japanese promos CDP had categorized as "Unknown" — vault correctly identified both.

Time elapsed: 38 minutes total (28 min for the import pipeline, 10 min for R.'s manual review).

Step 3 — Verify the listing state

For active eBay listings, the vault doesn't republish on import — that would risk creating duplicate listings if eBay still has CDP's listings live. Instead it pulls each card's current eBay state via the eBay Inventory API and merges:

• Cards live on eBay → marked Listed in the vault, mapped to the existing eBay listing.
• Cards sold but not yet shipped → marked Sold, queued for shipping.
• Cards in draft on CDP → marked Identified in the vault, queued for review.
• Cards paused → marked Paused (the vault preserves this state).

This is the critical migration step. If the importer republished everything, R. would have had 1,043 duplicate listings on eBay within minutes — a worse outcome than not migrating at all.

The verification took the vault about 6 minutes and didn't require any action from R.

Step 4 — Disconnect CDP

Once the verification was clean, R. signed into CDP and disconnected his eBay account from CDP's app. The vault's eBay connection is on his own seller account directly; CDP's was a separate OAuth grant.

(He kept his CDP account active for another two weeks "just in case." Both because he wanted to verify the sold-not-shipped queue cleared correctly and because his CDP subscription was paid up. Practical decision.)

What was faster

For R., immediately after the migration:

• Bulk re-listing. R. had 87 cards he'd been meaning to re-list for weeks (paused due to old photos he wanted to redo). In the vault, he batch-uploaded new photos from Google Drive, the AI re-identified, and 87 cards went back on eBay in about 12 minutes. CDP's equivalent flow had been a per-card slog because of feature-gating.
• Flagging cards for Whatnot. R. tagged 200 cards for an upcoming Whatnot show in about 4 minutes (a vault batch operation). His prior CDP workflow was a manual product-list build that took him 90+ minutes per show. (Note: the vault's direct Whatnot integration is on the Phase 5 roadmap. Today, tagged cards export to CSV for hand-upload to Whatnot — still faster than the CDP flow he was doing, but not yet the API-direct sync we're building toward.)
• Custom title templates. R. had been paying Majors tier ($19/mo) for this on CDP. In the vault, custom templates are available on the $10 Sleeve tier. He saved $108/year on this one feature alone, immediately.

What was different / required adjustment

Not everything was a straight win. R.'s honest feedback:

• The card list view is sortable differently. CDP's default sort is recent-first; the vault's default is batch-then-position. R. had to spend an evening getting used to it.
• The shipping label flow is different. CDP's was deeply integrated with eBay's label purchase. The vault's uses PrintNode (Dymo 30334) more directly. R. had to set up PrintNode (took 8 minutes) and learn the new label-print keyboard shortcut.
• The "AI credits" mental model became "imports." R. was used to thinking in CDP's AI credit budget. He had to adjust to the vault's import budget (similar mechanic, different naming). After two weeks he didn't notice.
• CDP's chat-support model. R. liked having a support chat. The vault is founder-email for now (he writes me; I reply same-day). The trade-off was acceptable to him; might not be for everyone.

What broke

Two things, both recoverable:

1. One Japanese card mis-identified. Card #847 — a 2002 Japanese Promo from a CoroCoro insert — had been listed on CDP with the right title but wrong year (CDP had it as 2003). The vault correctly identified it as 2002. The eBay listing carried CDP's wrong year; R. had to re-publish that one listing manually with the corrected year. Five-minute fix.
2. Image URLs from CDP's CDN started 404ing after CDP cancellation. When R. eventually canceled his CDP account three weeks later, CDP's CDN began returning 404s on his image URLs. The vault had downloaded the images on import, but they were stored as CDN references in the migrated data, not as vault-hosted images. The vault's migration tool now copies images to the vault's S3 immediately on import (we fixed this within a week of R.'s migration). If you migrate today, this doesn't happen.

The 60-day report

Sixty days after migration, R.'s data:

• 1,043 cards migrated → still ~900 active (143 sold)
• Sell-through unchanged on eBay (~12% per month)
• Whatnot revenue added in the 60 days: meaningful (R. ran 4 shows in the period). Per-show prep time dropped from 90+ minutes to under 10 minutes once he was flagging cards in the vault.
• Subscription cost differential: $19/mo CDP Majors → $30/mo Vault Binder. R. moved up a tier because his import volume warranted it. The Binder tier was still less than CDP's All-Star ($59) for a similar feature surface, and saved him CDP's per-channel upgrade pressure.

What I'd tell anyone considering the switch

It's a real Sunday afternoon. Not a "kick off the migration and finish next week" project. Sit down for 90 minutes — 30 min of import, 30 min of review, 30 min of getting used to the new UI. By dinner, you're done.

Three things to budget for:

• Image migration. The vault now hosts all migrated images on day one. Confirm this before disconnecting CDP.
• Custom field mapping. Bring a list of your CDP custom fields. The vault maps most of them automatically; verify each one.
• Shipping flow re-learn. Maybe an evening, maybe less. If you use PrintNode + Dymo, the keyboard shortcuts are different.

Honest tradeoffs

The vault isn't strictly better than CDP at everything. Where CDP currently has an edge:

• Bigger catalog (20M+ vs the vault's ~10M-card coverage). For obscure sports parallels and Magic foreign-language, this matters.
• Established team and chat support. Vault is one-person founder email.
• Mature mobile-app polish. The vault's mobile experience is web-first; CDP has a native app feel for some workflows.

If your business is "I list 200 cards of common sports parallels per week on eBay only and chat with support twice a month," CDP is probably fine for you. Don't migrate.

If your business is "I want multi-channel, I'm tired of feature gates, my Japanese Pokemon doesn't get respect from the catalog, and my Whatnot stream is bleeding hours of pre-show prep" — the math has changed.

— Jamie

The CDP-to-AICV migration importer is live for any incoming customer. The 14-day trial lets you run it before committing to a paid plan.