Walmart has never published how it weights its ranking factors, and the specific numbers you have read in Walmart SEO articles do not come from Walmart. There is no 85% attribute fill floor. There is no 40/30/20 split between content, performance and price. There is no “Luminate ranking engine.” We are going to show you where each of those came from, what Walmart does publish, and why the difference matters when you are deciding who to trust with your listings.
This is an uncomfortable article for us to write, because our own service is built on knowing how Walmart search works. It is easier to sell a number. It is more useful to tell you the truth.
The three claims, and what is actually true
Claim 1: “Walmart’s ranking runs on Luminate”
What is true: Walmart Luminate is a data analytics product suite for suppliers, announced by Walmart Data Ventures and launched globally on 16 May 2024. Walmart’s own announcement describes it as a “comprehensive data analytics product suite” covering Shopper Behavior, Channel Performance and Customer Perception. It was later renamed Scintilla.
It is a dashboard you buy insights from. It does not rank anything. Anyone describing Luminate as the engine that scores your listings has never opened it.
Claim 2: “Polaris weights item completeness at 40%, performance at 30%, price at 20%”
What is true: Polaris is a semantic search engine that Walmart’s @WalmartLabs division built in 2012, using technology from its acquisition of the startup Kosmix. It was real, it was significant at the time, and it is now fourteen years old.
Walmart has never published a weighting for it or for anything that replaced it. The 40/30/20 figures appear to have been invented once and copied forward ever since. Search for them and you will find a dozen agency blogs asserting them and not one primary source.
Claim 3: “You need roughly 85% attribute fill to be indexed”
What is true: Walmart’s own Listing Quality Dashboard documentation names three components of the Listing Quality score, and then says this, word for word:
The three categories in the breakdown section are not equally evaluated and the algorithm can be different for all categories of products.
It also notes that a component’s impact “may differ per item because the model analyzes a granular view of customer drivers to purchase similar items.”
Read that again. Walmart is telling you directly that there is no universal weighting and no universal threshold, because scoring is category-specific and item-specific. An 85% number that applies to every seller in every category is not a simplification of Walmart’s system. It contradicts it.
What Walmart does publish
Three things, and they are more useful than the invented numbers.
1. The three Listing Quality components. Content & Discoverability, Offer, and Ratings & Reviews. Named by Walmart, visible in your own dashboard, scored per item.
2. That scoring is category-specific. Quoted above. This is the single most actionable fact Walmart publishes about ranking, and almost nobody repeats it, because “it depends on your category” does not make a good headline. It does make a good audit.
3. How the catalog reads your listing. In May 2025, Walmart Global Tech published a technical post on using LLMs to manage product catalogs. It describes a two-step system: “One LLM identifies relevant attributes from product descriptions and images,” and a second LLM, “trained on human-validated data, verifies the accuracy of the extracted attributes.” Walmart then states that “for attributes where LLM accuracy is above a certain threshold (typically 90-95+% depending on the attribute), we can confidently ingest the extracted values into our catalog.”
That 90-95% figure is Walmart’s confidence in its own model, not a fill rate you have to hit. We are flagging that explicitly because it is exactly the kind of number that gets lifted out of context and turned into the next fake threshold.
Why this matters for your listings
The practical consequence is not that optimization is impossible. It is that the shape of the work changes.
If you believe there is an 85% floor, you chase a number. You fill attributes to hit a percentage, including ones that are irrelevant to how people shop your category, and you stop when the bar turns green.
If you accept what Walmart actually says, you do something better: you look at what the top listings in your category have filled in, and you match or beat them on the attributes that category actually uses. Some categories are attribute-heavy. Some are almost entirely decided on offer and reviews. Walmart is telling you the weighting differs. Your competitors’ live listings are the closest thing to a readout of what it is.
And because the catalog pipeline extracts attributes from your descriptions and images, the quality of your natural-language content is not separate from your structured data. Thin, keyword-stuffed copy gives that pipeline less to extract. Specific, complete copy gives it more.
How to check a Walmart ranking claim in thirty seconds
- Ask where the number came from. A real one traces to walmart.com, corporate.walmart.com, marketplacelearn.walmart.com, tech.walmart.com, or a named Walmart executive on an earnings call. Nothing else is a primary source.
- Search the exact number in quotes. If every result is an agency blog and none is Walmart, it was invented.
- Check whether it varies by category. Walmart says its algorithm does. Any claim that applies one fixed number to every product is claiming to know something Walmart says is not fixed.
- Watch for repurposed figures. The 90-95% above is a real Walmart number attached to a completely different thing. Real number, wrong subject, is the most convincing kind of wrong.
The honest version of what an audit can tell you
We can tell you which attributes in your category are empty on your listing and filled on the listings beating you. We can tell you where your title, description and images give the catalog pipeline less to work with than they could. We can tell you whether your offer loses the comparison on price, shipping speed or stock. We can rank those fixes by expected impact and explain the reasoning.
We cannot tell you that content is 40% of your score, because Walmart has not said that and neither has anyone else who actually knows.
If a provider gives you a precise weighting, you have learned something valuable about the provider.
Frequently asked questions
Does Walmart publish its ranking algorithm?
No. Walmart publishes the three components of the Listing Quality score, Content & Discoverability, Offer, and Ratings & Reviews, and states in its own documentation that the three “are not equally evaluated and the algorithm can be different for all categories of products.” It has never published weightings, thresholds, or a named ranking engine.
Is there an 85% attribute fill requirement on Walmart?
No. Walmart publishes no attribute fill threshold, and its own documentation says scoring differs by product category and even by item. Attribute completeness genuinely matters, because Walmart’s catalog pipeline extracts attributes from your descriptions and images, but there is no universal percentage to hit. Compare your attribute coverage against the top listings in your specific category instead.
What is Walmart Luminate?
Walmart Luminate is a data analytics product suite that Walmart Data Ventures sells to suppliers, launched globally on 16 May 2024 and later renamed Scintilla. It provides shopper behavior, channel performance and customer perception insights. It is not a search ranking engine, and it does not score your listings.