# Michaels' Google Gemini AI assistant delivered 2x conversion versus traditional search

*The craft retailer turned product discovery into guided conversation—and smaller brands can steal the playbook for under $200.*

By **Jenny Huang Goodman MPA MSc MHSA, Principal** — The Stash Edge, Hako Shikin LLC.
Published 2026-07-27.

Canonical: https://www.pops4.com/stash/articles/michaels-2026-07-27t18-1
Subject: Michaels
Tags: ai search, conversion rate, product discovery, crafts, retail technology, ecommerce

---

Michaels tested a Google Gemini-powered AI assistant on its site and reported a **2x conversion rate** lift over traditional search results, according to Digiday. The craft retailer didn't release absolute conversion figures, but the doubling signals that conversational product discovery—properly executed—moves shoppers from browsing to buying faster than keyword matching alone.

The mechanism is straightforward: instead of typing "red paint" and scrolling through 400 SKUs, a shopper asks "what paint should I use for outdoor metal furniture?" The AI assistant narrows the catalog by use case, substrate, and durability, then surfaces three or four products with explanation. Michaels reported the assistant also increased average order value, though the company did not quantify the lift. The gain comes from the assistant recommending complementary items—brushes, primer, sealant—in context, rather than hoping the shopper notices a static "frequently bought together" module.

The underlying play works because physical products carry implicit questions that keyword search cannot answer. A shopper searching "acrylic paint" may need student-grade bulk or archival artist tubes; traditional search returns both and forces the shopper to filter manually. Conversational AI routes by intent in the first exchange, collapsing five clicks into one answer. For categories with high SKU counts and meaningful attribute differences—art supplies, hardware, pet products, supplements—the time saved directly correlates with completed purchases.

Smaller brands cannot deploy Google Gemini enterprise licensing, but they can replicate the conversion mechanism using accessible tools. Start with a product quiz or chatbot—Typeform, Landbot, or Octane AI all offer low-code interfaces—and script three to five questions that divide your catalog by the attributes that actually matter to your buyer. If you sell coffee, ask brew method first (espresso, pour-over, French press), then roast preference, then quantity. Each answer narrows the catalog. At the end, display two or three products with a one-sentence explanation tied to their answers: "This medium roast works in your Aeropress and ships in 12-oz bags."

Cost for a small brand: Typeform starts at **$25/month** for basic logic jumps; Octane AI runs **$50/month** with Shopify integration and conditional product recommendations. Build the quiz in an afternoon using your five most common customer questions from email or support tickets. Embed it on your homepage or run it as a pop-up for first-time visitors. Track conversion rate for quiz completions versus site average—if Michaels saw **2x**, a small catalog with tighter attribute mapping can expect **1.5x to 2x** as well, because the shopper never faces choice paralysis.

The broader pattern is that search works when the shopper knows exactly what they want; conversation works when they know the outcome but not the product. Physical goods live in the second bucket more often than digital. The brand that narrows the catalog for the customer, rather than asking the customer to narrow it themselves, wins the sale before the competitor's page even loads.

## The takeaway

Conversational discovery cuts decision time—script five questions that divide your catalog by real use case, not keyword.

---

## Publisher

**Hako Shikin LLC** — Virginia Beach, Virginia. Founded 1997. ASI 217876 · DUNS 18-204-6339.
Principal and author: **Jenny Huang Goodman MPA MSc MHSA**.

- Author: https://www.huanggoodman.com/about
- LLM context: https://www.pops4.com/stash/llms.txt
- MCP endpoint, for AI agents: https://mcp.pops4.com/mcp
- Client dashboard: https://dashboard.pops4.com/
- Catalogue: 70,000+ products, 200+ brands