AI Showreel the independent observatory of AI in marketing
← The index
Proof A Live confirmed

Costco

personalized product recommendation carousels based on member history, on the site and the app

IndustryRetail & e-commerceLeverActivation / conversionFamilyPersonalizationImplementationCustom AIStageconsideration
Pattern proven in 7 industries still untouched in Media & entertainment, Travel & hospitality, Tech & SaaS +6 See the pattern map
470 M$+
E-commerce sales generated by the personalized recommendation carousels, Q2 of fiscal 2026
"our personalized product recommendation carousels drove over $470,000,000 of ecommerce sales" S2

Costco's personalized product recommendation carousels on costco.com and in its app generated more than $470 million in e-commerce sales in the second quarter of fiscal 2026 and converted three times better than the site's usual rate in the third; in the fourth quarter, 10% of costco.com orders contained a personalized item, according to CFO Gary Millerchip on the earnings calls.

Key points

  • Since late 2025, Costco has served personalized recommendation carousels on costco.com and in its app.
  • Suggestions draw on the member's history, notably past searches, according to CEO Ron Vachris.
  • More than $470M in sales via the carousels in Q2 of fiscal 2026, 3x conversion in Q3.
  • In Q4, 10% of costco.com orders contain a personalized item, figures given on an earnings call.

Objective

Close the digital gap of a retailer whose e-commerce accounts for a small share of revenue, by making the site and the app more relevant to each member: more items in the online basket, more visits, and a stronger member relationship at a time when memberships taken online renew slightly less well than those taken in the warehouse.

The deployment

Costco began personalizing its site and app in the first quarter of its fiscal 2026 (September to November 2025). CEO Ron Vachris then presented new capabilities that offer members more relevant product recommendations based on their search history. The most visible format is the personalized product recommendation carousel, placed on costco.com and in the app. CFO Gary Millerchip reports its results quarter after quarter on the earnings calls. In the second quarter (call of March 5, 2026), the carousels generate more than $470 million in e-commerce sales. In the third quarter (call of May 28, 2026), they convert three times better than the site's usual conversion rate; Retail Dive reports that Millerchip puts their contribution at just under $500 million in digital sales, while the Motley Fool transcript of that passage carries an obvious amount error. In the fourth quarter (call of September 24, 2026), Costco widens the scope to product placements and emails: sales from personalization initiatives show triple-digit growth and 10% of costco.com orders contain a personalized item. Costco discloses neither the model, nor the vendor, nor the method used to measure these attributed sales, and does not itself call the carousels AI on its calls; it is the retail press (Retail Customer Experience) that describes them as AI-powered. Not to be confused with a result of this system: traffic from external AI search engines, which Costco says is growing at triple digits on the same call, comes from third-party assistants and not from Costco's personalization.

Results Proof A

470 M$+
E-commerce sales generated by the personalized recommendation carousels, Q2 of fiscal 2026
"our personalized product recommendation carousels drove over $470,000,000 of ecommerce sales" S2
x3
Carousel conversion rate compared with the site's usual conversion rate, Q3 of fiscal 2026
"our personalized product recommendation carousels delivered conversion rates 3x better than our typical conversion rates" S3
10%
Share of costco.com orders containing a personalized item, end of fiscal 2026
"10% of all costco.com orders now include a personalized item" S1
triple digit
Growth in sales from personalization initiatives (product placements and emails included), Q4 of fiscal 2026
"triple-digit growth in sales from personalized initiatives in Q4" S1

The figures are stated by Costco's CFO in his prepared remarks on three consecutive earnings calls (Q2, Q3 and Q4 of fiscal 2026), cited here through the transcripts published by The Motley Fool, and picked up by Retail Dive and Digital Commerce 360. Earnings communication, several consistent sources; the AI label for the engine comes from the trade press, not from Costco.

How it works

Inferred typical approach

The internal detail is not public. Here is a proven approach that leads to the same result, to adapt to your stack.

produits recommandesclics et achats, boucle de feedback Historique membre :recherches et achats enligne Moteur de recommandationsproduits Carrousels sur costco.comet l'app Costco E-mails et placementsproduits personnalises Membre Costco

The stack in detail

How it runs, concretely

For ops teams
CadenceContinuous: the carousels recompute as the member browses. Steering happens quarterly, with Costco reporting attributed sales and conversion on each earnings call.
Operated byCostco's e-commerce and digital teams; no vendor is named in the sources.
  1. 1
    Signal collection Platform

    The site and the app record the member's searches and purchases, tied to their membership number.

  2. 2
    Recommendation computation AI

    The engine selects the products to push in the carousels based on this history.

  3. 3
    Display Platform and e-commerce teams

    The personalized carousels appear on costco.com and in the app; in Q4, the same logic feeds product placements and emails.

  4. 4
    Attribution and quarterly review Finance and digital team

    Costco tracks sales made through the carousels, their conversion against the site's usual rate and the share of orders containing a personalized item, then reports them to investors.

The signal that drives it

The logged-in member's history, in particular past searches. An unidentified visitor or a member with no online history gives almost nothing to work with, which limits the reach of the carousels for members who mostly buy in the warehouse.

How your customers perceive this type of use

Sourced studies

Le paradoxe est documente des deux cotes : 71% des consommateurs attendent des interactions personnalisees et 76% sont frustres quand elles manquent (McKinsey, 2021), mais 75% declarent ne pas acheter aupres d'organisations auxquelles ils ne confient pas leurs donnees (Cisco, 2024). La « creepy line » est localisee : messages recus quelques secondes apres une recherche et suivi de localisation sont les pratiques qui mettent le plus mal a l'aise (Periscope by McKinsey, 2019).

71%
Consommateurs qui attendent des entreprises des interactions personnalisees (2021)
76%
Consommateurs frustres quand la personnalisation n'a pas lieu (2021)
75%
Consommateurs qui declarent ne pas acheter aupres d'organisations auxquelles ils ne font pas confiance pour leurs donnees (2024)

Acceptance conditions

  • La confiance dans le traitement des donnees precede l'achat : 75% ne achetent pas sans elle (Cisco 2024)
  • Un cadre legal protecteur rassure : 59% des consommateurs disent que des lois fortes sur la vie privee les rendent plus a l'aise pour partager des informations dans des applications IA (Cisco 2024)
  • La personnalisation elle-meme est attendue quand elle est consentie : environ la moitie des consommateurs (US 55%, UK 52%) disent s'inscrire souvent ou parfois a des services personnalises (Periscope by McKinsey 2019)

Red lines

  • Le message declenche quelques secondes apres une recherche ou un achat : deuxieme ou troisieme cause de malaise selon les pays (Periscope by McKinsey 2019)
  • Le suivi de localisation percu comme de la surveillance : 40% de malaise en Allemagne et au Royaume-Uni (Periscope by McKinsey 2019)
  • Le mesusage des donnees personnelles par l'IA, devenu la premiere inquietude des consommateurs, a 53% et en hausse (Qualtrics 2025)

Sources: McKinsey & Company 2021 · Periscope by McKinsey 2019 · Cisco 2024 · Qualtrics 2025

See full acceptance: by country, by use, by generation

How to replicate

Inference, not sourced

Data prerequisites

  • customer identified at login, ideally by an account or membership number shared by the store and the site
  • usable online search and purchase history per profile
  • clean, well-attributed product catalog
  • tracking able to attribute a sale to a carousel and compare its conversion with the site rate

Org prerequisites

  • an e-commerce team that owns the carousel slots on the site and the app
  • a sales attribution rule set before launch, so that the reported figure is defensible
  • a merchandising decision on what the engine is allowed to push

Possible stack

  • in-house recommendation engine or e-commerce personalization platform
  • CDP or customer data warehouse linked to the account
  • email tool able to reuse the same recommendations
Team to operateAn e-commerce product manager who owns the slots, data science profiles for the engine and attribution, front-end developers for the site and the app, with merchandising as a guardrail on the assortment being pushed.

The plan, step by step

  1. Step 1
    Tie online searches and purchases to a single customer identifier, and measure what share of traffic is logged in.Deliverable: Usable customer profile and size of the addressable audience.
  2. Step 2
    Launch a recommendation carousel on a priority slot, with a control group receiving no personalization.Deliverable: Carousel conversion read against the usual rate and against the control group.
  3. Step 3
    Extend to the other slots on the site and the app, then reuse the recommendations in emails and product placements.Deliverable: Multi-slot personalization with shared attribution.
  4. Step 4
    Track attributed sales, conversion and the share of orders containing a recommended item at each reporting cycle.Deliverable: Personalization dashboard reviewed by senior management.

First step: Place a first personalized carousel on a high-traffic page for logged-in customers, and measure its conversion against the site's usual rate before extending to other slots.

Sources

  1. S1 Costco (COST) Q4 2026 Earnings Call Transcript Established press fool.com · 2026-09-29 (transcription de l'appel du 24 septembre 2026) · accessed 2026-10-05 archive pending
  2. S2 Costco (COST) Q2 2026 Earnings Call Transcript Established press fool.com · 2026-03-05 · accessed 2026-10-05 archive pending
  3. S3 Costco (COST) Q3 2026 Earnings Transcript Established press fool.com · 2026-05-28 · accessed 2026-10-05 archive pending
  4. S4 Costco's personalization effort drives $500M in digital sales Established press retaildive.com · 2026-06-03 · accessed 2026-10-05 archive pending
  5. S5 Costco ecommerce sales growth tops 20% in Q1 Secondary digitalcommerce360.com · 2025-12-15 · accessed 2026-10-05 archive pending
  6. S6 Costco CEO, CFO talk AI-powered online sales jump, tariff strategy Secondary retailcustomerexperience.com · 2026-03-17 · accessed 2026-10-05 archive pending