Benchmarks / Monthly reports / September 2026
Agentic commerce benchmark report: September 2026
The September 2026 agentic commerce benchmark report is Ottom's monthly summary of how AI shopping agents fared when they tried to discover, evaluate, compare and buy on online stores, by retail category. Across 251 stores measured at every stage, the median agentic readiness score was 55 out of 100.
The month in brief
All stores
251 stores measured at every stage from 3 July 2026 to 1 October 2026. Median agentic readiness score 55 out of 100, with the middle half of stores between 52 and 57. 4% of them let an AI shopping agent reach the payment step.
| Compared with | Median score | Reaching payment |
|---|---|---|
| 30 days earlier | Not comparable: there is not yet enough published history to compare. | |
| 3 months earlier | Not comparable: there is not yet enough published history to compare. | |
| 6 months earlier | Not comparable: there is not yet enough published history to compare. | |
| 12 months earlier | Not comparable: there is not yet enough published history to compare. | |
Most widespread stalls
- Products have no GTIN or MPN to match against: found on 65% of stores scanned.
- Size or colour can only be chosen by hovering: found on 58% of stores scanned.
- No sitemap for agents to crawl: found on 51% of stores scanned.
The live all stores benchmark, with its trend.
Apparel
60 stores measured at every stage from 3 July 2026 to 1 October 2026. Median agentic readiness score 55 out of 100, with the middle half of stores between 52 and 58. 13% of them let an AI shopping agent reach the payment step.
| Compared with | Median score | Reaching payment |
|---|---|---|
| 30 days earlier | Not comparable: there is not yet enough published history to compare. | |
| 3 months earlier | Not comparable: there is not yet enough published history to compare. | |
| 6 months earlier | Not comparable: there is not yet enough published history to compare. | |
| 12 months earlier | Not comparable: there is not yet enough published history to compare. | |
Most widespread stalls
- Products have no GTIN or MPN to match against: found on 70% of stores scanned.
- Size or colour can only be chosen by hovering: found on 62% of stores scanned.
- No sitemap for agents to crawl: found on 38% of stores scanned.
The live apparel benchmark, with its trend.
Beauty and personal care
26 stores measured at every stage from 3 July 2026 to 1 October 2026. Median agentic readiness score 55 out of 100, with the middle half of stores between 52 and 57. 0% of them let an AI shopping agent reach the payment step.
| Compared with | Median score | Reaching payment |
|---|---|---|
| 30 days earlier | Not comparable: there is not yet enough published history to compare. | |
| 3 months earlier | Not comparable: there is not yet enough published history to compare. | |
| 6 months earlier | Not comparable: there is not yet enough published history to compare. | |
| 12 months earlier | Not comparable: there is not yet enough published history to compare. | |
Most widespread stalls
- No sitemap for agents to crawl: found on 65% of stores scanned.
- Size or colour can only be chosen by hovering: found on 62% of stores scanned.
- Products have no GTIN or MPN to match against: found on 62% of stores scanned.
The live beauty and personal care benchmark, with its trend.
Food and drink
30 stores measured at every stage from 3 July 2026 to 1 October 2026. Median agentic readiness score 57 out of 100, with the middle half of stores between 52 and 61. 3% of them let an AI shopping agent reach the payment step.
| Compared with | Median score | Reaching payment |
|---|---|---|
| 30 days earlier | Not comparable: there is not yet enough published history to compare. | |
| 3 months earlier | Not comparable: there is not yet enough published history to compare. | |
| 6 months earlier | Not comparable: there is not yet enough published history to compare. | |
| 12 months earlier | Not comparable: there is not yet enough published history to compare. | |
Most widespread stalls
- Products have no GTIN or MPN to match against: found on 87% of stores scanned.
- Size or colour can only be chosen by hovering: found on 43% of stores scanned.
- No sitemap for agents to crawl: found on 30% of stores scanned.
The live food and drink benchmark, with its trend.
Health and wellness
24 stores measured at every stage from 3 July 2026 to 1 October 2026. Median agentic readiness score 57 out of 100, with the middle half of stores between 52 and 57. 0% of them let an AI shopping agent reach the payment step.
| Compared with | Median score | Reaching payment |
|---|---|---|
| 30 days earlier | Not comparable: there is not yet enough published history to compare. | |
| 3 months earlier | Not comparable: there is not yet enough published history to compare. | |
| 6 months earlier | Not comparable: there is not yet enough published history to compare. | |
| 12 months earlier | Not comparable: there is not yet enough published history to compare. | |
Most widespread stalls
- Size or colour can only be chosen by hovering: found on 67% of stores scanned.
- Products have no GTIN or MPN to match against: found on 54% of stores scanned.
- No sitemap for agents to crawl: found on 38% of stores scanned.
The live health and wellness benchmark, with its trend.
Home and furniture
23 stores measured at every stage from 3 July 2026 to 1 October 2026. Median agentic readiness score 52 out of 100, with the middle half of stores between 52 and 57. 0% of them let an AI shopping agent reach the payment step.
| Compared with | Median score | Reaching payment |
|---|---|---|
| 30 days earlier | Not comparable: there is not yet enough published history to compare. | |
| 3 months earlier | Not comparable: there is not yet enough published history to compare. | |
| 6 months earlier | Not comparable: there is not yet enough published history to compare. | |
| 12 months earlier | Not comparable: there is not yet enough published history to compare. | |
Most widespread stalls
- Products have no GTIN or MPN to match against: found on 74% of stores scanned.
- No sitemap for agents to crawl: found on 74% of stores scanned.
- Size or colour can only be chosen by hovering: found on 70% of stores scanned.
The live home and furniture benchmark, with its trend.
Jewellery and accessories
33 stores measured at every stage from 3 July 2026 to 1 October 2026. Median agentic readiness score 52 out of 100, with the middle half of stores between 52 and 55. 0% of them let an AI shopping agent reach the payment step.
| Compared with | Median score | Reaching payment |
|---|---|---|
| 30 days earlier | Not comparable: there is not yet enough published history to compare. | |
| 3 months earlier | Not comparable: there is not yet enough published history to compare. | |
| 6 months earlier | Not comparable: there is not yet enough published history to compare. | |
| 12 months earlier | Not comparable: there is not yet enough published history to compare. | |
Most widespread stalls
- No sitemap for agents to crawl: found on 85% of stores scanned.
- Products have no GTIN or MPN to match against: found on 67% of stores scanned.
- Size or colour can only be chosen by hovering: found on 64% of stores scanned.
The live jewellery and accessories benchmark, with its trend.
Comparisons over 30 days, 3 months, 6 months, 12 months are not in this issue: there is not yet enough published history under the current methodology to compare against.
Still collecting data, so not in this issue: Electronics, Outdoor and sports.
Citing this report
The figures are published under CC BY 4.0. Cite them as: Ottom, Agentic commerce benchmark report: September 2026, https://ottom.io/benchmarks/reports/2026-09. How every figure is measured is on the methodology page.
Get the monthly benchmark report
One email a month: the median agentic readiness score by category, the stages where AI agents stall, and how each has moved over the last month, quarter, half year and year.