Lead Horse helps ecommerce teams troubleshoot product feed issues before another failed upload.
We help identify whether the problem is file format, field names, JSONL structure, gzip compression, SFTP upload, image URLs, variants, eligibility, or product validation.
A successful transfer is not the same as an accepted feed.
Your file can transfer successfully and still fail later. The upload layer only confirms the file arrived — the feed layer needs it to be readable, valid, structured, and mapped correctly.
That is why feed troubleshooting needs more than "try uploading it again."
Most failures trace back to a handful of recurring issues. Here is where we look first.
The upload path may expect JSONL.GZ, CSV.GZ, or another specific format. A plain JSON array can fail if the path expects one product object per line.
Q&A, variant dictionaries, reviews, metadata, geo pricing, and availability may need to stay real JSON objects or arrays. Saved as broken strings, they fail or lose context.
Image URLs need to be direct, accessible, and tied to the right product or variant. Additional images should prove the product claim, not just fill space.
Search, ads, and checkout eligibility need to be set intentionally. Checkout should stay off unless operations, policies, fulfillment, and order handling are ready.
If variants are not grouped correctly, colors and sizes can look like unrelated products. If color data is vague, shopper expectations end up wrong.
Even a valid file can fail if the processor requires a specific remote directory, filename pattern, or marker file. A successful transfer does not mean the file was processed.
File-feed ingestion and API-style feeds often expect different field names for the same data. If the names do not match what the ingestion path expects, the file fails.
The data is the same. The schema is not. We map your fields to whatever the ingestion path actually expects.
A product feed can transfer through SFTP and still fail product ingestion. We check whether the file is deliverable and whether the products inside it are usable.
Is it the right format, compressed correctly, named correctly, and uploaded to the right place?
Are the fields valid, complete, useful, and structured for ChatGPT Ads?
We help identify exactly where the failure is — not just that something is wrong.
Describe what happened and paste the error — we will tell you where it is breaking.
A clear path from "it keeps failing" to a validated file that is ready for the next upload.
Describe the upload method, file format, error message, and where the process is breaking.
We look for format issues, schema mismatches, malformed fields, wrong field names, product problems, and upload process gaps.
We explain whether the issue is likely in the file, the schema, the upload path, the naming pattern, or the account setup.
We can create clean CSV, JSON, JSONL, and JSONL.GZ versions as needed.
We check row count, gzip integrity, JSONL parsing, required fields, booleans, arrays, URLs, and product scope.
Use us for a quick diagnosis or a full rebuild — scoped to whatever is actually breaking your feed.
You do not need to upload a file to get started. Tell us what happened, what error you saw, what format you used, and what product or category you are trying to launch. We will help identify whether the issue is format, schema, fields, images, SFTP, validation, or account setup.
We will review what you sent and reply within two business days with where the feed is likely breaking and how to fix it.
Describe what happened, paste the error, and tell us what you are trying to launch. We will identify whether the issue is format, schema, fields, images, SFTP, validation, or account setup.