ChatGPT

All figures measured.

#ProductRecommended in
1 Klaviyo 3 of 3
2 Mailchimp 3 of 3
3 Omnisend 3 of 3
4 Sender 3 of 3
5 Shopify Email 3 of 3

Scanned 12 October 2026. 3 samples from ChatGPT, round 9. 5 products recommended. Methodology v2026-09-29.1.

Perplexity

All figures measured.

#ProductRecommended in
1 Omnisend 3 of 3†
2 Shopify Messaging 3 of 3†
3 Abandoned Cart Pro 2 of 3†
4 Customer.io 2 of 3†
5 Mailchimp 1 of 3†

† Corrected 12 October 2026: incomplete_round. Some of this round's samples had failed when it was first published. They were collected afterwards with the same questions and settings, and the figures marked † were recomputed. How corrections work

Scanned 12 October 2026. 3 samples from Perplexity, round 9. 5 products recommended. Methodology v2026-09-29.1.

Counts, not percentages: with three samples a percentage would imply a precision the sample size cannot carry. “2 of 3” means the assistant recommended that product in two of its three answers to this question.

Sources the assistants cited

  • techradar.com 4
  • shopify.com 3
  • apps.shopify.com 2
  • popupsmart.com 2
  • omnisend.com 2
  • australianonlinenews.com.au 1
  • canaryusers.ai 1
  • madlemmings.com 1
  • behance.net 1
  • logowik.com 1
  • fiverr.com 1
  • libautech.com 1

Scanned 12 October 2026. 6 samples across ChatGPT and Perplexity, round 9. Methodology v2026-09-29.1.