Lessons from submitting 500k images from Shopify stores to Google for indexing
We open up Pandora’s box for Shopify merchants
I launched Image Sitemap for Shopify in December 2017. To date we have submitted over 500k images to Google Search on behalf of our customers—with those customers seeing an average of 648% increase in indexed images. I started with a hypothesis about helping merchants reliably improve their Google image indexing. This hypothesis came about due to some limitations with the Shopify platform along with the massive amount of effort needed to manage large image sitemaps (or those greater than 2,000 images and 30,000 lines of XML — some of our customers’ image sitemaps contain more than 500k lines of XML). It seems that my hypothesis is on the way to being strongly validated.
By default, Shopify only includes a single image per product in their standard image sitemap. The Image Sitemap for Shopify app submits every single image to Google and includes important product metadata as well. We also resubmit all data to Google on every catalog change. Google seems to like what we’re doing, so I wanted to share our findings so far.
- Across our customers, we are seeing an average of 648% improvement in image indexing from the day our app is installed.
- Our largest customer, Contemporary Furniture Warehouse, has over 80k images in their catalog — in their first two weeks with the app they saw a 252% increase in indexed images in Google.
- Jessica Rose went from 0% indexed images in Google to 76% indexed in just three weeks. Jessica’s shop has been on Shopify for over three years!
- Our best performing customer, Parks Project, is reliably averaging 90%+ images indexed.
- We helped Obey Clothing expose a massive Googlebot error with their EU store.
- Baa Baa saw a 2000% increase in indexed images in 10 days.
- I Do Forever saw a 320% increase in indexed images in one month.
Apart from the obvious task of submitting every product image to Google, regardless of catalog size—there seem to be some interesting findings when looking at the following measures:
- File Name Similarity
- Site Traffic
- Image Uniqueness & Duplicate Content
1. File Name Similarity
File name similarity seems to be a substantial predictor of image indexing in Google. Google likes to see image file names that are very similar to both the product title and the product handle (or url pathname) . A correlation above 0.40 seems to be ideal. We calculate this correlation using Dice’s Coefficient.
File name: foo-bar-pocket-tshirt-striped-red-front.jpg
Product title: Foo Bar Pocket T-Shirt — Striped
Product handle: /products/foo-bar-pocket-tshirt-striped-red
Our customers whose average correlation is above 0.40 are indexing 40% more images in Google than those whose correlation is below 0.40.
2. Site Traffic
Site traffic also seems to be a substantial predictor of image indexing in Google—this is pretty obvious, but it’s worth discussing regardless. Our customers whose global Alexa ranking is inside the top 2million are indexing 28% more images in Google than those whose global ranking is outside the top 2million. Our customers whose global Alexa ranking is inside the top 1million are indexing 41% more images in Google than those whose global ranking is outside the top 2million.
3. Image Uniqueness & Duplicate Content
Image uniqueness is an opaque thing to measure, but it seems that Google is starting to take this more seriously.
Our analysis here is only anecdotal, but it seems that merchants whose images are very unique, highly detailed, and distinguishable are much more reliably indexed. For example, something like a white T-shirt on a light colored background with very little detail on the T-shirt seems to be demoted by Google.
This concept is particularly important for merchants who are selling products that are not unique—like a retailer who carries many brand names. For example, a shop that sells Zippo lighters and uses product imagery provided by Zippo without shooting their own unique product photography—Google does not seem to like this one bit.
One theory is that Google considers the images to be duplicate content. The images from the authoritative source seem to take precedence in every scenario.
It is still early days for our app, however the results to date were too exciting not to share. If you know anyone using Shopify, please recommend that they consider Image Sitemap for Shopify, available in the Shopify app store with plans starting at $4 / month.