Verintra / Feed optimizer

What grade does your feed get?

Paste the feed URL your store already publishes. In about two minutes you get a letter grade for every product and a ranked list of what to fix first.

  • No login
  • No card
  • Nothing written back to your feed

Or try full Pro for 14 days, no card

Slim fit chino, sand, 32Five findings, four of them one click away
F41
title
Chino 32Slim Fit Chino Trousers, Sand, W32
description
41 charactersCotton twill with a touch of stretch, tapered through the leg.
gtin
emptytaken from your catalog, never invented
google_product_category
emptyApparel > Clothing > Pants
color, size
emptySand · 32
Sample data · source feed unchanged

Feed visibility

How much of yours is lit?

Readable data gets discovered.

From URL to grade in two minutes.

The scan asks nothing of you. Paste a URL, read the result. If the fetch fails, you know at once, and you know why. No endless spinner.

0s

You paste the feed URL

The same URL your ad channels already fetch. Nothing to upload.

URLpush ingest
8s

The feed is fetched and parsed

Whatever format your platform produces, read once and normalised into one shape.

XMLCSVJSONhttps only
45s

Every product is scored

Two engines read each item, and their findings combine into one letter grade.

30+ checksA to F
~2m

The report is ready

How the catalog spreads across the grades, then the fixes, ordered by the points each one recovers.

grade per productranked fixes

How the catalog splits, A to F

The shape of the problem before any of the detail. Two hundred F products and two thousand C products need completely different work.

412A
498B
231C
96D
47F

Verintra score

Every product gets one.

30+ checks. One letter.

Two scores, measured once.

Two engines read every product. One asks whether the data is complete, the other whether a model can understand it. A feed can be valid and still say nothing.

Accepted and understood
Understood, but rejected
Neither
Accepted, and says nothing
Verintra Score: rejectedaccepted
Sample catalog of 40 products
DimensionFor the channelFor the model
Required fields: price, availability, linknot weighted
Structured data on the product pagenot weighted
Description and semantic richness
Images and visual richness
Title

Weighted the same by both:

GTIN, or brand and MPNCategory specificityAttribute completeness

Thirteen fix groups, ranked by what they recover.

Findings collapse into groups, not thousands of rows. Each says how many products it touches and how many points it lifts, so the order of work is decided for you.

Every group opens the right tool with the affected products already filtered, and every bulk change can be reverted.

  1. 01
    Missing identifiersGTIN, MPN or brand absent, so the product cannot be matched
    214 products
    +9
  2. 02
    Short or vague titlesNo size, colour or material, so no query matches it
    168 products
    +6
  3. 03
    No product categoryUnmapped to the Google taxonomy, so bidding and ranking both suffer
    97 products
    +3
  4. 04
    Unusable imagesBelow the minimum size, or the URL does not resolve
    31 products
    +2
  5. 05
    Broken sale dates and pricesA sale window that has closed, or a price the landing page contradicts
    12 products
    +1

Five of the thirteen shown. Sample data.

You wrote the redactions.

Missing data is still a message.

The product you submitted

title
Wireless
gtin
brand
Acme Audio
color
material
product_type
Electronics

Grounded in your catalog, or it does not ship.

A model that invents a material is worse than an empty field: the empty field fails a check, the invention passes one. So generation is constrained, not trusted.

Titles and descriptions

Written from the fields you have

brand Halden product_type trail running shoe material Gore-Tex feature waterproof size_type wide

Halden Trail Runner GTX, Waterproof, Wide Fit

Hover a word to see the attribute it came from. Every word traces back to a field already on the product, and the result is written to a separate field for you to compare.

Category matching

The model never sees a wrong answer

The official taxonomyevery category Google publishes
5,500
A local shortlistmatched on words already on the product, no model
12
One categoryranked by the model, checked back against the shortlist
1

A category that was never offered cannot come back.

Patterns get a rule, not a spreadsheet.

If the category is chinos and the title has no size, append the size. One rule, applied to every product that matches, re-evaluated on every sync.

11operators
10actions
37fields
IF
THEN

Three sample products.

One feed, every destination that reads it.

Slim fit chino, sand, 32Slim Fit Chino Trousers, Sand, W32
A

33 channels28 presets plus custom formats: XML, CSV, JSON, JSON-LD and ACP. Every plan publishes to every destination, the AI ones included.

  • Google Shoppingxml
  • Microsoft Advertisingxml
  • Facebook / Metaxml
  • ChatGPT / OpenAI ACPjson
  • Perplexity Shoppingxml
  • TikTok Shopcsv
  • Pinterestxml
  • Idealocsv
  • Merchant CenterContent API v2.1, with disapproval reasons pulled in and grouped, so a rejection becomes a fix group instead of an email.
  • Meta catalogReal OAuth and batch updates. The same corrected data goes to catalog ads without a second spreadsheet.
  • GA4, 13 reportsGrades next to revenue. Every product shows its funnel beside its grade, split into Hero, Performer, Underperformer and Invisible.
  • A/B titlesTwo titles for the same product, measured against each other, so the better one is decided by the data and not by taste.
  • Size-run protectionWhen a size run breaks, the product is held back before it wastes clicks on sizes that are not there.

What the scan will not do.

write access

It does not touch your feed

The scan reads. It has no write path back to your store, by design, so there is nothing to undo afterwards.

sign up

It does not need an account

The grade comes first. You decide afterwards whether it was worth your two minutes.

OAuth

It does not connect to your ad accounts

Nothing is requested from Google or Meta to produce the grade. The feed on its own is enough.

guesses

It does not invent a single value

A missing field is reported missing. Nothing is filled in to make the number look better than the catalog is.

The free scan reads up to 2,000 products and opens two of the fix groups in full. The rest are counted and named, and unlock on a plan.

Before you paste the URL.

Where do I find my feed URL?

Wherever you set up Google Shopping. It is the same URL Merchant Center fetches. If you have never set that up, the app carries short guides for sixteen store platforms, from Shopify and WooCommerce to ikas, Ticimax and Ideasoft, each pointing at where that platform keeps the address.

Does it work with Shopify and WooCommerce?

Verintra works with any feed URL, whatever produced it, so both are fine. To be precise about what that means: there is no app to install on either platform. We read the feed they publish, which is why the platform does not really matter.

Can I upload an XML or CSV file instead?

Not at the moment. Ingestion is by URL, or by pushing data to us. A file you upload once goes stale the same day, and a stale feed is the problem rather than the fix.

What if my server blocks you?

Some hosts block requests from datacenter addresses or from any user agent they do not recognise. Our fetcher identifies itself as VerintraFeedFetcher, so allowing that is usually enough. The error message tells you when this is the cause.

See your own catalog in this view.

Fourteen days of full Pro, no card. Or grade the feed first, read only.

Start the 14-day trial