What are shopping agents actually doing on your store?
Agents guess codes, learn from rejections and blur into your analytics. Benson spots the ones it recognises and blocks the guessing.
14 days of everything in Own. No card. Then Free, unless you pick a plan.
- Block code guessing by agents
- Agent detection on every plan
- Private codes never published
Agent visits
A visible actor signal starts the trail.
Reads your public offers
The agent reaches the offers you publish.
Tries codes at the cart
Attempts become part of the cart context.
Useful shopping, or repeated code guessing?
Your newest customer doesn't have a face.
An AI agent lands on a product page, adds to cart and starts trying codes. SAVE10. WELCOME. FREESHIP. Every rejection message teaches it something.
In your analytics it's just another session. In your margin, it's a discount nobody approved.
You could block every bot. You'd also turn away agents shopping on a real customer's behalf, and the orders they bring.
You could ignore it. That works until an agent finds the code you emailed to fifty people.
Or you could give agents a front door.
Benson recognises the agents it can see, blocks their code guessing with per-visit caps and publishes the offers you're happy for them to use. A shopper with one code isn't touched, and observe mode shows it all before any limit goes live.
Useful shoppers, or costly guessers?
01 · The problem
Agents keep guessing codes.
They cycle the coupon field and learn from every rejection.
The fix
Block the guessing.
Cap code attempts per visit for agents and automation. A shopper with one code is left alone.
02 · The problem
Your real offer is invisible to machines.
A promo in a banner or email gives an agent nothing to read.
The fix
Publish the offers you approve.
A feed of public offers from live campaigns. Private codes never appear.
03 · The problem
Machine traffic blurs your data.
Useful shopping and costly guessing look the same in analytics.
The fix
See machine visits on their own.
Recognised agents, split from shoppers and coupon extensions.
No storefront system sees every agent; fetch-only and unsigned automation can slip past. Illustrations use sample data.
See what agents are doing on your store today.
14 days of everything in Own. No card. Then Free, unless you pick a plan.
Let it through, or limit it?
Something's off
Block it.
Repeated attempts cost you. Cap them, and reject automation codes where supported.
It's helping
Keep the feed.
Agents reach approved offers without guessing. Refine the policy text.
Too early
Keep watching.
Thin signal. Stay in observe mode; quiet isn't proof no agent visited.
Three steps, in order.
See the traffic.
Known agent visits, code attempts and policy blocks.
Publish your offers.
Turn on the feed for active public codes only.
Switch on the caps.
Observe first, then cap code attempts.
The Cloudflare detector is still coming, so visits that never reach the browser can be missed.
Questions
No. It sees browser-visible signals; fetch-only traffic can bypass the script. The Cloudflare detector for custom stores is still coming.
Give shopping agents a clear offer.
14 days of everything in Own. No card. Then Free, unless you pick a plan.