Entity SEO for DTC Brands: An Audit of 29 Stores

Entity SEO is the work of making search engines and AI assistants recognize your brand as one specific company and attach the right facts to it: name, products, founders, location, profiles. We read the entity records of the 29 brands in Shopify's own DTC showcase on 5 October 2026. Of the 27 we could load, 21 publish an Organization record, none links it to Wikidata, and one ships nine empty profile slots.

What is entity SEO, and why does it matter in AI answers?

It matters more in AI answers because the engine describes you in its own words instead of showing your page. Keyword SEO asks which page ranks for a phrase. Entity work asks something earlier: when someone types your brand name into ChatGPT or Gemini, which company does the engine think it means?

Being mentioned often does not help if the engine has picked the wrong company. At DeviLabs we score this as a separate criterion in every baseline. Our GEO measurement protocol lists five binary checks per answer: correct brand name, correct location, correct core services, correct people, and "not confused with another entity, group company, or country". On that last one the protocol is blunt: "C5 is the criterion no public framework includes, and in practice it fails more often than the other four combined."

We saw it on our own brand. In the protocol's field notes, Gemini did not resolve LuckyPaws, our own local business in Wassenaar, to the right company. A mention counter would have scored that answer as a win. That is why we run correction before amplification: fix who the engines think you are, then work on how often they name you.

What 29 DTC brands publish as their entity record

Most of them publish a partial record, and none connects it to Wikidata. We took the 29 brands in Shopify's own article on DTC brand examples and, on 5 October 2026, read each homepage the way a crawler that does not run JavaScript sees it. Two refused the request, leaving 27 of the 29.

Shopify's article does not link Hint, so we used drinkhint.com. For each readable homepage we recorded every Organization-type JSON-LD block (Organization and subtypes such as OnlineStore; schema.org Brand blocks are a different type and not counted), its name, whether it had a logo or founder, and every sameAs entry, counting blank ones separately. Then we looked every brand up on Wikidata, by the official website on its item and by its exact English name. The sample is a showcase list chosen by Shopify, so these are better-known brands than the average store.

What the table says, in counts:

• 21 of the 27 readable homepages publish Organization-type markup in their HTML. Six publish none before JavaScript runs, Liquid Death and Lovevery among them, though both have Wikidata items.
• 14 of those 21 list at least one real profile in sameAs. Seven list none.
• Six ship blank sameAs entries. Pepper's record has nine slots and all nine are empty.
• None of the 21 links to Wikidata or Wikipedia, including the brands that have an item.
• Three homepages (Blume, Blueland, Glamnetic) publish two to four separate Organization-type blocks for the same company.
• Hint's markup names the company "drinkhint.com", a store name left as a domain. Pulp Pantry's domain now presents the brand as Trashy Chips, and its markup carries no alternateName for the old name, so anyone following Shopify's list meets a name the record never connects to the old one.
• Only Feastables names a founder in its Organization markup. Glamnetic puts one on a separate Brand block.

Limits: one fetch per site from a cloud server, homepage only, and markup injected by JavaScript is not counted. A crawler that renders the page may see more. We refetched the six homepages without markup to make sure we had the real page and not a bot check; all six returned the full store page. We did not test what AI engines say about these brands, so the table describes the records, not their effect on answers.

How Shopify's Dawn theme ships blank sameAs

Shopify's Dawn theme builds its Organization block from admin settings. The name is your store name, and sameAs is a fixed list of nine social-link settings. Leave a setting empty and that slot still ships as an empty entry. If your store name is a domain or an old brand name, the markup calls your company that.

You can see it in the Dawn theme's header template: the name comes from the shop name, and the sameAs array lists Twitter, Facebook, Pinterest, Instagram, TikTok, Tumblr, Snapchat, YouTube and Vimeo links whether you filled them in or not. Pepper's page source declares a Dawn-based theme, and its nine sameAs entries are all empty. August also runs a Dawn-based theme and ships five blanks next to four real profiles.

Shopify's newer Horizon theme takes the opposite route: its Organization block has a name, a logo and a URL, and no sameAs at all. Other themes build the block their own way (Halfdays runs Debut and ships seven blanks), so check yours whichever theme you run.

Google's documentation describes sameAs as the URL of a page on another website with more information about your organization, such as a social or review profile. An empty entry is not a URL. It is unlikely to break anything, but it says nothing, and a record whose only real profile is Instagram gives an engine one corroborating link where it could have several.

The fix is small, and it never means opening accounts you do not use just to fill a slot:

1. Find the block. View your homepage source and search for "sameAs". Note any empty or null entries and how many Organization blocks the page carries.
2. Edit the theme. In Shopify admin, go to Online Store, Themes, Edit code. On Dawn the block sits in sections/header.liquid. Either wrap each sameAs line in a check that skips it when the setting is blank, or replace the array with a hand-written list of your real profiles. On Horizon, add a sameAs list to the same block.
3. Keep one block. If an SEO app adds its own Organization markup, turn that output off in the app or remove the theme's block, so one record remains.
4. Check the result. Paste your homepage URL into Google's Rich Results Test or the Schema Markup Validator and confirm one Organization node with no empty sameAs entries.

While you are there, check the name. Google's Organization markup guide says the markup can help it "disambiguate your organization in search results", which only works if the name in it is the name customers use.

Does every brand need a Wikidata item?

No. Some guides say yes: Searchbloom writes that "Every brand should have a Wikidata entity record", and Discovered Labs advises you to "Create a Wikidata entry even if you lack a Wikipedia article". Neither quotes Wikidata's own rule, which is the one that decides whether your item survives.

Wikidata's notability policy accepts an item that "can be described using serious and publicly available references", or one that fills a structural need in the database. It also says data outside those guidelines will not be accepted. A brand whose only sources are its own site and its own Instagram has nothing for an item to cite.

In our sample, 8 of the 29 brands have an item: six we found through the official-website statement on the item, and Glamnetic and Lovevery, whose items carry no official website at all. Even among Shopify's showcase brands, most have none, so a missing item is normal and is not the first thing to fix.

Wikidata is also a quick way to see how many other things share your name. On 5 October 2026 we counted the Wikidata items whose English label exactly matches the brand name:

If your brand is a common word, the name alone resolves to nothing. The description, the official website, the founder and the location are what separate you, so those facts must be identical everywhere they appear. If you do qualify for Wikidata, keep the item complete: official website, inception, founder, headquarters, and a link from your homepage sameAs back to it.

The entity record checklist for a DTC store

Work through these six steps in order, starting with the name, because every later step repeats it. Most are an edit to your own site or profiles. Wikidata is the only one that depends on other people, because it needs outside coverage before an item can stand. Recheck the record after every rename, new market or new profile.

1. Fix one canonical name and one sentence. Write what you are, where you are based and what you sell in one line. Use it on your About page, in the markup description, and in every social bio, marketplace storefront and wholesale listing.
2. Publish one Organization block on the homepage. Name, url, logo, description, sameAs, plus legalName and a real address if you have them. Google's guide says there are no required properties and asks for as many relevant ones as apply. A separate Brand block for your product line is fine; a second Organization block is a duplicate.
3. Clean sameAs. Only live profiles that are yours or describe you: Instagram, TikTok, LinkedIn, YouTube, your Amazon brand store, a review profile such as Trustpilot, and Wikidata if you have an item. No blanks, no dead accounts.
4. Put the facts in the HTML. Founding year, founder, city and product range on the About page, readable without JavaScript. Only one brand in our sample named a founder in its Organization markup.
5. Record renames. If you changed your name, say so on the About page with the year, add the old name as alternateName, and redirect the old domain. Pulp Pantry to Trashy Chips is the pattern: the old name still circulates in lists like Shopify's.
6. Decide on Wikidata honestly. Press coverage or a Wikipedia article: create or complete the item. Neither: spend the effort on coverage first.

Here is what step 2 looks like for a fictional linen store. Replace every value with your own and delete any line you cannot fill truthfully:

{
  "@context": "https://schema.org",
  "@type": "OnlineStore",
  "name": "Northfold",
  "alternateName": "Northfold Linen",
  "legalName": "Northfold Home B.V.",
  "url": "https://www.northfold.example",
  "logo": "https://www.northfold.example/logo.png",
  "description": "Northfold is a Dutch online store selling stonewashed linen bedding, made in Portugal.",
  "foundingDate": "2021",
  "founder": { "@type": "Person", "name": "Founder Name" },
  "address": { "@type": "PostalAddress", "addressLocality": "Utrecht", "addressCountry": "NL" },
  "sameAs": [
    "https://www.instagram.com/northfold",
    "https://www.tiktok.com/@northfold",
    "https://www.linkedin.com/company/northfold",
    "https://www.trustpilot.com/review/northfold.example"
  ]
}

How to test whether AI engines know who you are

Ask the engines who you are, and score what they say. Five identity questions, run in ChatGPT, Gemini, Perplexity and Google AI Mode, show whether the engines describe the right company. Score every answer against the same five checks we use in client baselines, and log what each engine got wrong.

For prompt 4 you need the other things with your name. Search your exact brand name on wikidata.org, note any company, band or place that comes up, and use the closest one.

Use fresh, logged-out sessions, so no chat history or personalization colors the answer, and name the exact product mode you tested. Run each prompt three times per engine and score the majority result; if the three runs disagree, log that as volatility instead of averaging it, which is how our protocol scores it. Score each answer yes or no on brand name, location, core products, people, and "not confused with another entity". Prompt 3 checks a fact beyond those five: write the price the engine gives next to your live price. In our own field notes, ChatGPT returned prices for LuckyPaws that did not match the published rates. Our measurement protocol has the full method and a capture template; how to measure AI visibility explains why one run is never enough.

Then route each failure to its cause. A wrong company means a disambiguation problem: the steps above. A right company with wrong facts means some source says the wrong thing: find it, correct it on your own pages first, then ask the outside source to update.

What entity SEO will not do for you

It will not make AI engines recommend you. Our GEO claims audit rated "schema markup increases AI citations" false on the controlled evidence, and added that "schema aids entity disambiguation, rich results, and knowledge graph association." Everything in this post helps engines describe the right company. None of it makes them rank or recommend you.

Being named in a "best linen bedding" answer depends on what independent sources say about you. As we put it in why ChatGPT recommends your competitor: "Your product page is a claim. A Reddit thread is a witness." Entity work makes sure the witnesses are talking about the right company. Getting them to talk is the GEO work that comes after.

Your next step: open your homepage source today, search for "sameAs", and count the empty or null entries. Then run the five prompts and write down which company each engine describes.