Tier 01
Pilot
200 products a month
$119/ month
$0.60 per product · 60 a day
- Your own sign-in
- Unlimited dry runs — validation never costs a fetch
- Email support
- Extra products at $0.60
For a first catalogue, or a supplier you are still testing.
Pricing
Every row you submit is a fetch on one paid API and a rewrite on another. So there is no unlimited plan here. There is a rate that falls as your catalogue grows — and the same software on every tier.
Per finished product on the Catalogue plan. $0.60 at the entry tier.
A price on a supplier page is logged, then dropped. Only your column ships.
Row 12 of your sheet is row 12 of the CSV. Never renumbered, never shifted.
The import CSV, the rows still without a photo, and two worksheets: what failed, and why a product is a draft.
01 Plans
There is no crippled tier. Every plan runs the identical pipeline — the same validation, the same extraction, the same rewrite, the same files back. What changes between them is how many products a month you can put through, and what each one costs you.
Tier 01
200 products a month
$119/ month
$0.60 per product · 60 a day
For a first catalogue, or a supplier you are still testing.
Tier 02
800 products a month
$399/ month
$0.50 per product · 200 a day
A full store build, or a catalogue you refresh every month.
Tier 03
3,000 products a month
$999/ month
$0.33 per product · 600 a day
Agencies running several store builds at once. Above 3,000 a month, ask.
Tier 04 · one-off
Send the sheet. We run it, work the failures, and send the files back. No account, no learning curve, no month. This is the one an agency buys when the store build is due and nobody has three weeks.
Invoiced monthly, cancel by telling us. Prices in USD, exclusive of any tax we are obliged to add. Your monthly allowance is what you are billed against; the daily ceiling is what the app enforces on your account, so a runaway paste cannot spend a month in an afternoon.
02 Estimator
Move the slider to the number of products you actually have. The hours are costed at ten minutes a product and $30 an hour — our assumptions, stated so you can disagree with them.
| Products | Plan | You pay | Per product | Hours you keep | By hand, at $30/hr |
|---|---|---|---|---|---|
| 50 | Pilot | $119 / mo | $2.38 | 8 hrs | $250 |
| 100 | Pilot | $119 / mo | $1.19 | 17 hrs | $500 |
| 200 | Pilot | $119 / mo | $0.60 | 33 hrs | $1,000 |
| 500 | Studio | $399 / mo | $0.80 | 83 hrs | $2,500 |
| 800 | Studio | $399 / mo | $0.50 | 133 hrs | $4,000 |
| 1,500 | Catalogue | $999 / mo | $0.67 | 250 hrs | $7,500 |
| 3,000 | Catalogue | $999 / mo | $0.33 | 500 hrs | $15,000 |
A plan costs the same whether you fill it or not, so the per-product figure only reaches the headline rate when you use the allowance. Under 200 a month, ask about a single one-off batch instead.
products
Arrow keys move 25 at a time; Page Up and Page Down move faster.
800 a month, 200 a day.
Per month, invoiced. $0.80 per product at this volume.
At ten minutes a product — two working weeks.
Your own hours at $30. A VA would be about $800.
Machine time is roughly five to eight products a minute — the rate limiter on the rewrite API is the ceiling, not the number of links.
03 The alternatives
Nobody is choosing between us and nothing. They are choosing between us, a fortnight of their own evenings, and a person they pay by the hour. Here is that comparison with the assumptions on the table — and below it, the honest note on where the other two win.
| 500 products | You, by hand | A VA | Catlist, Studio |
|---|---|---|---|
| Time | |||
| Hours you personally spend | 83 hrs | ~8 hrs briefing and spot-checking | ~3 hrs validating and reviewing flags |
| Elapsed before the store can open | 8 weeks at two hours an evening | 3–4 weeks | An afternoon |
| Money | |||
| Cash out of pocket | $0 | $800 | $399 (the plan covers 800) |
| Your hours, costed at $30 | $2,500 | $240 | $90 |
| Real total | $2,500 | $1,040 | $489 |
| What the output is like | |||
| Consistency across 500 rows | Drifts. Row 400 is not written like row 4. | Drifts between people and across weeks. | The same prompt on every row, in one pass. |
| A price you are not sure about | You squint at it and move on. | They guess, or they ask you and wait. | Refused before it parses. The row ships as a draft. |
| A product nobody can read | You notice it. Eventually. | They notice it, and they can phone the supplier. | Listed by row number in a separate file, with the reason. |
| Doing it again next quarter | The same 83 hours. | Cheaper — they have learnt your catalogue. | The same afternoon. |
Twenty products. Or your own products, that you photographed, that nobody else has copy for. There is no source page to read, and no machine is going to know the thing better than the person who made it. Write those twenty yourself and they will be better.
Anything needing a login, a phone call, a supplier portal, or a judgement about which of two near-identical products your sheet meant. A VA also gets better at your catalogue every week; software does not. If your bottleneck is decisions rather than typing, hire the person.
The sheet is long, the sources are ordinary product pages, and a wrong row is expensive enough that you would rather it were flagged than guessed. That is the whole case. It is not that we are cleverer than your VA — it is that we do the same thing to row 900 as we did to row 1, and we say so when we could not.
04 On every plan
Not a feature list with holes cut in it for the cheap tier. Everything below runs for everybody — the only difference between plans is the number of rows.
Your sheet row number rides all the way through and is never renumbered. Both CSVs come back sorted by it, so row 12 of the output is row 12 of your sheet.
Check rows replays the whole batch — pairing, prices, SKUs, duplicates, misaligned columns — without a single fetch. Free, and as often as you like.
The cell is validated as you typed it, before a number is parsed out of it. A barcode, a date, a range or 129 kr 50 is refused rather than quietly becoming a price.
SKU and barcode come from your columns or they ship empty. An invented barcode is a permanent problem in a Shopify catalogue and a Merchant Center risk.
Structured data first — JSON-LD, then Open Graph, then microdata. It works on sites nobody wrote a rule for. Amazon has its own adapter and collapses by ASIN.
A free plain request first; a residential proxy only if that was blocked; a full browser render only if that was too. You are not paying for a render on a page that did not need one.
Candidates upgraded to the best resolution the CDN serves, junk dropped, duplicates merged, every URL verified to load, and a vision pass to say which ones are actually the product.
Links that failed or scraped thin are retried against a cleaner site for the same product, confirmed by GTIN or a vision check — after the main batch, so it never slows the good rows.
The import CSV; the products that scraped fine but have no photo, importable once you paste an image URL; the dead links with row numbers and reasons; and every draft with the rule it broke.
Optional per-run switches, also on every plan: premium copy for the product body, and sharpening of genuinely small photos. Photos put on a plain white background cost nothing extra. A photo the AI redraws as a clean studio shot — one you pick, or every weak or watermarked photo if you switch that on — costs half a credit at Fast speed or a quarter at Economy, and only when the redraw passes our checks and you keep it.
05 Getting in
There is no sign-up button and there is not going to be one this quarter. Every run spends real money on two paid APIs, so an account gets set up for a named company with a size on it — which also means we can size your batch and quote it properly before you pay for a month you did not need.
How many products, and two or three of the supplier links. That is genuinely enough for us to know whether they read cleanly.
We put your own links through and send you back what comes out. If your suppliers do not read well, we would rather you found out from us than from an invoice.
One company, a sign-in for each person, a daily ceiling set to your plan. Signing in lasts 30 days at a time. We can turn it off, raise it, or add your VA the same day.
06 Questions
They are logged so you can see what the site said, and then dropped. A price on a supplier page is very often another currency, a sale price that expires, or the price of a related product in the sidebar — and shipping one of those is worse than shipping nothing.
The only price that reaches your CSV is the one in your own column, and it is checked
as typed before any number is pulled out of it. A barcode, a date, 15 - 20
or 129 kr 50 are all refused outright rather than silently becoming
4006381333931, 20240501, 1520 and
12950. Two numbers in one cell is never a price, because which one is the
price is unknowable.
A row with no usable price still imports — as an unpublished draft, tagged
MISSING-PRICE. Nothing ever goes on sale at 0.00.
The fetch escalates on its own: a free plain request, then a residential proxy if that was blocked, then a full browser render if that was blocked too. Most sites stop at the first or second step. Separately, links that failed or came back thin get a second chance against a different site selling the same product, confirmed by matching GTIN or a vision check before anything is swapped in.
When a page still will not read, the row is not silently dropped and it is never invented. It lands in the failed-links file with its row number, everything you pasted for it, and the actual reason.
And it costs you nothing: only a finished product with a photo uses a credit. A link that will not read, or a product with no usable photo, is not charged.
You do. It is your CSV, your product copy, your store. We claim nothing in it, we do not resell it. Your current batch, your edits to it and your uploaded sheets are kept on our server so a restart or a redeploy does not lose your work; they are replaced when you start a new list. Your account also records how many products it used.
Fix it before you import — that is rather the point of handing you a CSV instead of pushing straight into your store. Nothing is live until you import it, and nothing with a missing price or a missing image imports as anything but a draft.
The rewrite is instructed to use only facts that are on the source page and never to
name the retailer it read. When the vision pass is unsure what the product actually is,
or the title and the description point at two different kinds of thing, the row is
tagged NEEDS_REVIEW — filter that tag in your Shopify admin and you have
your review queue.
It will still get things wrong sometimes. It is a language model reading a supplier's marketing copy. What it will not do is invent a specification that was not on the page.
Yes — Amazon has its own adapter rather than going through the generic path, and it collapses by ASIN, so the same product pasted three times with different slugs and tracking parameters becomes one product and the two drops are reported with the row they duplicated.
Amazon is also the most likely of any source to need the expensive fetch tier, which is part of why the pricing is per row rather than flat.
Mechanically: we gather the candidates, upgrade each to the largest version the supplier's CDN will serve, throw out logos, badges, swatches and tracking pixels, merge the same picture at different sizes, and verify every URL actually loads. Images that turn up on two or more products in one batch are store banners, not products, and are stripped. The URLs go in the CSV and Shopify re-hosts them free on import. Optionally we sharpen genuinely small ones. A photo is only redrawn by AI when you ask for it — by ticking it on the review screen, or with a batch switch for weak or watermarked photos — because redrawing can change small details. A redraw is checked against the original before it is kept, you see every one before you import, and you can undo it.
Legally: that is between you and your supplier, and we are not going to pretend otherwise. Finding a photo does not license it. Plenty of distributors grant image rights to their resellers in the dealer agreement — plenty do not, and marketplace listing photos usually belong to whoever shot them. Check before you publish. We give you the file; we cannot give you a licence to it.
No, and that is deliberate. You get a CSV in Shopify's standard product-import layout — one row per product plus a row per extra image — and you import it yourself. It means you can read the whole batch before anything touches your store, and fix a hundred rows in a spreadsheet in the time an API integration would take to fix one.
No. One credit is one finished product with a photo. A link that will not read, or a product that came back with no usable photo, costs nothing — our failures are not your bill.
Before a run starts, the app checks that your credits and your daily ceiling would cover every row if every one finished, so a batch is never stopped halfway. You are then charged only for the ones that did. The daily counter rolls over at midnight UTC, and Check rows finds dead links without spending anything.
Yes, with a top-up. When a batch would not fit in what is left, the app says so before anything is spent, and we add extra products the same day at your plan's own rate — $0.60, $0.50 or $0.33 — invoiced with the month. Top-ups never expire. There is no penalty rate, because a punitive overage just makes people split one sheet across two accounts, and splitting the sheet breaks the one thing that matters most here: one run, one row numbering.
The daily ceiling on your account is a different thing. It is a burst limit that stops a runaway paste from spending a month of API budget in an afternoon, and we will raise it for a day if you ask.
You choose, for each batch and each time you send photos. Both make the same photo and run the same checks; only the wait and the price differ.
On Studio, a product at $0.50 with three photos redrawn costs $1.25 at Fast or $0.88 at Economy. Only photos that pass the checks are charged, and undoing one gives its credit back.
That is what step two of getting access is. Send two or three of your own supplier links with your request and we put them through and send back exactly what comes out — the CSV, and the honest note if your suppliers are hard to read. It costs you nothing and it costs us a few cents, which is a much better trade than a refund conversation three weeks later.
07 Request access
Two or three of your supplier links and a rough count is enough. We will run yours, send you back what comes out, and quote the tier that actually fits — which is sometimes a one-off batch rather than a plan.
Already have an account? Sign in. Questions first? [email protected].
We reply by email, usually within one business day, with a quote and, if you sent links, what they actually produced.