"The biggest time sink in eBay listing isn't typing. It's stopping to decide what category, which specifics, and whether the AI got it right. Voice and image together reduce that decision load — not by deciding for you, but by giving you better starting material."
— Folder Lister development notesWhat the camera can see
AI image analysis scans a product photo and extracts structured data: brand logos, material textures, colour, shape, pattern, and sometimes model numbers. For a vintage ceramic vase, the camera might confidently identify "blue and white", "hand-painted", "ceramic" and "Delft style".
These are real, measurable attributes. eBay's own listing tools already do this — their image-based listing feature can suggest a draft title, category and item specifics from a single photo.
But confidence levels vary. Colour detection is typically 85–95% accurate. Material identification drops to 65–80% — the camera sees texture, not composition. And pattern or style attribution can be as low as 55–70%, because visual similarity doesn't prove authenticity.
Notice the confidence gap. The camera is reasonably sure about colour and shape, less sure about material, and almost guessing about condition. A chip on the bottom, a hairline crack on the inside, a repaired handle — these are invisible to a standard product photo.
What only the seller knows
The seller is holding the item. They can feel the weight, check the bottom for maker's marks, measure the height, and notice the small chip that's hidden in the photo. This information is critical for buyer confidence and eBay search visibility, but it lives entirely in the seller's head.
Typing it out — opening fields, finding the right category dropdown, entering measurements — is where most listing time actually goes. Speaking the same information takes a fraction of the time.
A real voice note might sound like:
"Good condition, small chip on the base but barely visible. Made in Holland, I can see the mark on the bottom. Probably 1960s or 70s. About 18 centimetres tall. I'd ask 24.95."
In 8 seconds of speaking, the seller has provided condition details, country of origin, era estimate, dimensions and price — five fields that image analysis would either miss entirely or guess at with low confidence.
"Image analysis tells you what something looks like. Voice input tells you what it actually is. The listing needs both — and so does the buyer."
— Folder Lister development notesThe merge: two sources, one draft
When the AI receives both the image analysis results and the voice transcript, it can cross-reference the two. The image says "ceramic" — the voice says "made in Holland" — together they point to a Delft pottery category with much higher confidence than either source alone.
The result is a structured draft with:
- Category suggestion — informed by both visual style and spoken context
- Item specifics — each tagged with its source (image or voice) and a confidence percentage
- Three description versions — so the seller can pick the tone that fits
Three description modes
| Mode | What it does | Best for |
|---|---|---|
| RAW | Your own spoken words, cleaned up. No AI rewriting. | Sellers who want authenticity |
| SEO | Rewritten for eBay search — keyword-rich, structured, professional. Follows your prompt instructions. | Maximising search visibility |
| Factual | Brief bullet-point format. Dimensions, condition, material. Nothing decorative. | Repeat stock, quick turnaround |
The SEO version can be steered with a seller prompt — "keep it short, condition first" or "mention the era prominently" — so the AI follows your instructions rather than a generic template.
Confidence-based review: you decide what to trust
Every extracted specific has a confidence percentage attached. Material at 72%, era at 28%, colour at 91%. A threshold slider lets you set your comfort level — anything below the threshold stays unchecked by default.
This is fundamentally different from tools that auto-fill everything and hope the seller notices mistakes. The review-before-apply model means:
- High-confidence fields (colour, dimensions, price) are ready to accept
- Medium-confidence fields (material, brand) get a second look
- Low-confidence guesses (era, origin from image alone) stay off until confirmed by voice
After a voice note, those low-confidence fields jump from 28% to 70%+ because the seller — who is holding the item — has confirmed or corrected them. The checkbox turns on. That's the system working as designed.
"AI should assist or serve. It should not decide. The moment a listing tool auto-fills 'excellent condition' on an item the seller hasn't described yet, it has crossed the line from assistance to liability."
— Folder Lister design principleMore complete listings rank higher
eBay's own documentation confirms that complete item specifics significantly improve search visibility. Listings with more filled specifics appear in more filtered searches — and filtered searches are where serious buyers shop.
The voice + image approach helps fill more fields than either source alone:
| Source | Fields it fills well | Fields it struggles with |
|---|---|---|
| Image only | Colour, shape, type, visible brand | Condition, origin, era, dimensions, price |
| Voice only | Condition, origin, era, price, flaws | Colour accuracy, category, pattern ID |
| Image + Voice | All of the above, cross-validated | Authenticity, exact age, hidden defects |
The free eBay Listing Score Scanner checks your title, item specifics, images, and trust signals against eBay's live taxonomy — and shows exactly where points are lost.
Tips for better voice notes
- Develop a mental template: condition, origin, era, dimensions, price, flaws. Same order every time. This keeps your voice notes consistent and the AI extractions reliable.
- Speak the details the photo can't show: bottom markings, weight, functionality ("tested, works"), repairs, missing parts. These are the fields that separate a good listing from a great one.
- Don't describe what's visible: the AI already sees the colour and shape. Use your voice time for the invisible attributes.
- Use a description prompt: "keep it factual" or "mention condition first" steers the SEO description toward your preferred style.
- Clean, well-lit photos matter more than quantity: a sharp, evenly lit photo on a neutral background gives the AI better data than ten blurry shots. eBay's zoom feature requires 1600px+ resolution.
"The sellers who scale aren't working harder — they're systematising smarter. A consistent 8-second voice note per item replaces 90 seconds of typing and produces more complete listings."
— Based on Folder Lister user workflowsFAQ
Can I use voice notes to create eBay listings?
Yes. Folder Lister lets you record short voice notes per item — price, condition, origin, flaws — while AI image analysis handles the visual attributes. The two sources merge into a reviewable draft with category, item specifics and description.
Does image analysis work for vintage or used items?
Image analysis can identify brand, material, colour and shape with reasonable accuracy. But condition, provenance and era are areas where the AI guesses — voice input from the seller produces far more reliable results for these fields.
How accurate is AI-generated item specifics for eBay?
Visual attributes like colour and material score 70–90% confidence. Non-visual fields like era and condition are 30–60% from images alone. Voice input pushes these above 80%. Folder Lister shows confidence per field so you decide what to trust.
How many item specifics should an eBay listing have?
Fill every required field and as many recommended fields as possible. eBay confirmed that complete specifics significantly boost visibility in filtered search. The free listing scanner checks your specifics against the live eBay taxonomy.
Is AI listing software safe for my eBay account?
Folder Lister uses eBay's official API and never auto-publishes. Every listing goes through seller review before anything reaches eBay. The review-before-apply model means the AI suggests, but you decide what gets submitted.
What is the difference between RAW, SEO and Factual descriptions?
RAW is your own spoken words cleaned up into readable text. SEO is rewritten for eBay search — keyword-rich and structured. Factual is brief bullet-points with only dimensions, condition and material. You choose which fits each listing.
Try voice + image listing yourself
The live demo on the homepage lets you record a voice note and see how the AI merges it with image analysis — no download needed. For the full workflow with your own items, download Folder Lister free.