A merchant installed Discount Prime, upgraded to the Prime plan within minutes, then opened a support chat asking for a feature by a name we did not use. The strangest part: the feature was, at that exact moment, uploading to our servers.
A support request that stopped us cold
It started as an ordinary day and an unusual message. A new merchant had installed Discount Prime and, unusually, gone straight to the Prime plan before doing anything else. Minutes later they were in support with a single, precise question: "Where do I find the Advanced product option?"
At first we were genuinely confused. Advanced product option is not a phrase we use anywhere in the app. We asked what they meant. They explained that they had been searching for exactly this capability, had read our help documentation, upgraded to Prime specifically to get it, and now could not find it in the interface.
The request was so specific that it stopped us. This was not a shopper poking around. This was someone who knew precisely what they wanted, why they wanted it, and that we had it, more clearly than most people who have used the app for months.
The feature was literally still uploading
Here is the part that felt almost staged. As that conversation was happening, our engineering team was mid-deployment. The feature the merchant was describing was inside the exact version being pushed to the servers.
We had published the help documentation ahead of the final release, so that when users arrived they would not be confused about a new option. What we never imagined was that a feature could have a customer before it finished loading. We told the merchant it would be live in about an hour. Exactly one hour later, they came back, confirmed they could see it, and thanked us.
So what was the feature?
The capability they wanted is the Variant Metafield Rule: the ability to choose which products a discount applies to by writing a rule against your own variant metafields, instead of hand-picking products one by one. You write a condition like "any variant where custom.material is leather," and every matching variant is included automatically, including ones you add later. It lives in the product-scope selector under the advanced, rule-based group and works within our Tiered Unit, Wholesale/B2B, and Dropshipping pricing campaigns. The full walkthrough is in our help center guide.
It was originally requested by a large B2B customer who needed their Shopify pricing driven by the structured data in their ERP, and we shipped it quickly because it solves a real, hard problem. We think it is one of the clearest points of difference between Discount Prime and the other discount apps on the market. We wrote about the design and the B2B thinking behind it in metafield-based targeting for B2B campaigns.
The question that actually kept me up
As the person responsible for product, the resolved ticket was satisfying. But a different question stayed with me: how did this merchant find us, and how did they understand a genuinely complex feature so precisely?
After thinking about it, I had a theory, and I tested it. We run a Discount Prime AI Assistant as a custom GPT inside ChatGPT. I asked it, in the merchant's own words, "where is the Advanced product option in Discount Prime?" It offered a few guesses, and one of them was, essentially, "you may mean this feature." It then handed back a link into our help center. The very first help-center search under that title landed on the documentation for this exact feature, with the full explanation and usage. Bingo.
The merchant had almost certainly done the same thing. They did not find us by scrolling the Shopify App Store. An LLM mapped their fuzzy need, "advanced product option," onto the precise thing we had built, and pointed them straight at it.
Customers are arriving through LLMs, and they already know what they want
Lately, customer behavior has been genuinely surprising me. Merchants are discovering us not through App Store search, but through ChatGPT, Gemini, Perplexity, and the rest. And they are not arriving vague. They understand new features quickly, and they map their own requirements onto the exact capability we built, often before a human on our side has said a word.
A few things compounded to make this work: using AI in our own product development and in how we write feature content, the speed at which we ship, and, crucially, presenting information in a clear, standardized way that an LLM can actually read and reason over. That last part turns out to matter enormously.
Why this feature matters for B2B and wholesale merchants
There is a reason this capability attracts this particular customer. Wholesale and B2B merchants on Shopify have historically had to reach for Shopify Plus to get the pricing behavior they need. By adding Wholesale / B2B Pricing and Dropshipping Pricing, and letting prices and discounts apply to selling price, profit margin, or, for dropshippers, product cost, we have been able to give non-Plus stores a level of control that genuinely helps them. It is also a narrowly targeted area. As far as we can tell, nobody has shipped a version that works cleanly at high product volumes the way merchants actually need. It is not that the feature is hard to build; it is that this was the lens we chose to build through. If you run your catalog from an ERP, we go into that in more depth in our guide to ERP integrations for Shopify B2B.
The message in a bottle, opened by a machine
I once wrote that we used to sit where we live and work, write our product information, seal it in a bottle, and throw it into the ocean of the internet, hoping the right person would find it and come to us. I told this story before about an American hat company that found us through a DevCraft Solutions blog post on Gemini.
These days, LLMs open those bottles before any human does, and they are surprisingly good at handing them to the exact people looking for what is inside. The volume through this channel is still small compared to traditional channels, but it is already remarkable, because of who arrives: a customer who, the moment they open the app, already knows what they want and where to find it.
What it means for how we market
This is changing how I think about marketing spend. A well-placed Shopify App Store ad targeting these keywords carries a suggested cost per click from Shopify's own bidding of around $30 to $50. You can win clicks at lower bids, but once you account for drop-off through the funnel, abandonment, and the lifetime value of each merchant, the economics of finding a customer through traditional advertising have shifted hard.
I cannot claim we have figured it out. But my daily observations of how our own customers behave keep surprising me, and they point in a clear direction: the content you write for machines to read is becoming as important as the content you write for people. When an LLM can cleanly understand your product and documentation, it will route the best-fit customers to you, sometimes before you have even finished shipping the thing they came for.
The Variant Metafield Rule is available on the Prime plan in Discount Prime, inside Tiered Unit, Wholesale / B2B, and Dropshipping pricing campaigns.
Related on Discount Prime: Metafield-based targeting for B2B · ERP integrations for Shopify B2B · Selling in the AI shopping era
