New reportWhy Black Friday is a retention event
Hello Retail

Recommendations with Hello Retail MCP

Ecaterina Capatina·November 4, 2026·3 min read
A demo store basket page with a Goes well with your basket recommendation box showing running socks, a soft flask and a headlamp

With Hello Retail, your AI assistant can set up, fix and measure your product recommendations. Connect Claude or ChatGPT to Hello Retail, and each job becomes a single request. Here is what you can ask for.

What your assistant builds on

Every request draws on what Hello Retail already knows about your products and shoppers.

  • Product Intelligence. Analyzes 250M+ products and learns the global patterns between them, so even first-time visitors see relevant products.
  • Behavioral data. Learns how each shopper browses and picks up their affinities for brands, categories and price ranges.
  • Best-practice strategies. Proven algorithms for each page, from the homepage to the cart and the 404 page. Start from one, or combine them into your own strategy.
  • Analytics for every box. Views, clicks, conversions and revenue per box, so your assistant can tell you what works and what to fix.
  • Your merchandising rules. Pin or exclude products where a human hand is needed.

How it works

  • Connect once. Sign in to your Hello Retail account from Claude or ChatGPT. No keys to copy.
  • Ask in your own words. Your assistant can create boxes, change settings, placement and design, and report how each box performs.
  • Nothing goes live by surprise. Every new box and every change is saved as a draft. You publish it in My Hello Retail.

Build a new recommendation in one request

Add recommendations to the cart page that match what’s in the basket.

  • Your assistant picks a best-practice strategy for cart pages.
  • It creates the box and places it under the basket, as a draft.

Then keep going in the same chat:

Add a title that shows how much is left for free delivery.

The title now counts down to free delivery as the basket changes.

Works on any page. Name the page, say where the box goes, and change the number of products, the layout or the devices by asking.

hello-retail MCP

Add a recommendation box to our cart page with products that go with what’s in the basket.

  • Found best-practice strategies for cart pagesrecoms_listBestPracticeAlgorithms
  • Created a cart box from the best fitrecoms_create
  • Placed it under the basketrecoms_updatePlacement

Your cart box is ready as a draft. Publish it in My Hello Retail when you’re ready.

Cart pageDraft ready for review
Nordtrail

Your basket

Fjell Hiking Boots€169
Subtotal €169Checkout

Goes well with your order

New

Wool Hiking Socks

€25

Waterproofing Spray

€15

Trekking Poles

€79

Illustrative examples for a demo store, in Claude or ChatGPT.

Ask how your recommendations are doing

Which recommendation box is underperforming?

  • Your assistant compares views, clicks, conversions and revenue across your boxes.
  • It finds the weak one and tells you why, for example clicks that dropped three weeks ago.
  • It suggests a fix and can prepare it as a draft.

Run Hello Retail Search too? One request gives you a monthly report on both, with quick fixes that could lift revenue.

hello-retail MCP

Which recommendation box is underperforming?

  • Compared click-through rates across your boxes
  • Checked the cart box’s daily trend
  • Found a best-practice strategy for cart pages

The cart box. Its clicks dropped three weeks ago. I can switch it to a best-practice strategy as a draft for you to review.

Click-through rate · last 6 weeks

Cart boxOther boxes
6 weeks ago3 weeks agoThis week

Illustrative numbers for a demo store, in Claude or ChatGPT.

More things you can ask for

Four everyday jobs, one request each.

hello-retail MCP

Our “bought together” box is gone from the product page since the redesign.

  • Read the placement
  • Logged the API calls
  • Fixed the placement

A box disappeared

Your assistant finds out whether the script isn’t firing or the strategy returns no products, and updates the placement so the box shows up again.

hello-retail MCP

Apply our new design to all recommendation boxes, but keep the old one serving until I’ve checked.

  • Copied the design
  • Edited the copy
  • Switched all boxes

Give every box the new look

One request restyles every box. The old design keeps serving until you have checked the new one and published it.

hello-retail MCP

Send recommendation clicks to GA4 with event category “recoms”.

  • Read GA4 settings
  • Turned on click events

Track clicks in Google Analytics

Your assistant updates the Google Analytics settings, so recommendation clicks show up in GA4 under the category you chose.

hello-retail MCP

Add a recommendation slider to our new outdoor landing page.

  • Found a strategy
  • Created a draft slider
  • Placed it on the page

Recommendations on a new landing page

Your assistant creates a draft box with a best-practice strategy and the right placement, ready for you to check and publish.

Illustrative examples for a demo store, in Claude or ChatGPT.

Key takeaways

  • Describe it, and it’s built. Name the page and what you want, and your assistant sets up a box from a best-practice strategy.
  • Fixes and restyles are one request. A missing box, a new design for every box, GA4 tracking: Ask for it in your own words.
  • The numbers are in the same chat. Ask which box underperforms and get the reason and a suggested fix.
  • You stay in control. Every change is a draft until you publish it in My Hello Retail.

How to get started

MCPAI for EcommerceProduct RecommendationsAutomation

Keep reading