Collect Product Images for E-Commerce Catalog Builds

TL;DR — The Google Images Scraper returns product image URLs with dimensions, source pages, and domains as JSON. Query by SKU or product name, filter by size, and fill catalog image gaps with documented provenance.

Why catalog image gaps hurt conversion

A product without a photo is a product that doesn't sell — and supplier feeds routinely ship SKUs with missing or broken image URLs. Google's image index is the fallback pool: manufacturers, retailers, and reviewers have already photographed almost everything.

The job-to-be-done is a candidate list per SKU: image URL, dimensions for quality filtering, and the source page so someone can confirm the match and the rights.

How does this compare to the alternatives?

Photo shoots are right for hero SKUs. For the long tail of a catalog, an indexed image with a source trail is proportionate.

How to collect product images in 4 steps

How do I query per SKU?

import os
from apify_client import ApifyClient

client = ApifyClient(os.environ["APIFY_TOKEN"])
skus = ["AquaPure AP-200 filter", "Trailblazer TB-40 backpack"]
run = client.actor("thirdwatch/google-images-scraper").call(
    run_input={"queries": skus, "maxResults": 30, "country": "us", "language": "en"}
)
rows = list(client.dataset(run["defaultDatasetId"]).iterate_items())

How do I keep only usable images?

import pandas as pd
df = pd.DataFrame(rows).drop_duplicates("imageUrl")
good = df[(df.width >= 600) & (df.height >= 600)]
print(good.groupby("query").size())

Dimensions filter out thumbnails before you spend review time.

How do I shortlist per SKU?

picks = (good.sort_values("position")
         .groupby("query").head(5)
         [["query", "imageUrl", "sourcePageUrl", "domain"]])

Top-position records from manufacturer or major-retailer domains are your first review queue — sourcePageUrl is where the reviewer confirms the match.

How do I record provenance?

Keep the full record set next to your catalog import: imageUrl, sourcePageUrl, domain, query. When a rights question lands later, the trail answers where every image was found.

Sample output

{"imageUrl": "https://mfg-cdn.example.com/ap200/front.jpg",
 "width": 1500, "height": 1500,
 "sourcePageUrl": "https://manufacturer.example.com/ap-200",
 "domain": "manufacturer.example.com",
 "query": "AquaPure AP-200 filter", "position": 1}

Common pitfalls

A returned image may show a different variant than your SKU — always confirm on sourcePageUrl before importing. Hotlinked URLs rot, so download promptly and keep the source trail. The same product image replicates across retail domains — dedupe by file hash, not URL. The actor supplies candidates with provenance; rights clearance for reuse is your call.

Related use cases

• Build an image dataset for computer vision
• Bulk image business data pipeline
• Guide to scraping e-commerce data
• Blog hub