Analytics · Holiday 2026
Some of your Black Friday visitors aren’t people.
Written forHeads of ecommerce, conversion teams and digital analytics leads
5 min read
Some AI assistants now open your product pages in a real browser, on a shopper’s behalf. They check the price and the stock, then leave. Your analytics records each one as a visit with no purchase. So your conversion rate can fall even when your shoppers haven’t changed.
A visit with no purchase
It arrives at 2:14 am. It spends five minutes on fourteen product pages, checks the stock in a size 10, reads the returns policy and leaves. It adds nothing to the cart.
Your analytics records it as a visit with no purchase. Add one for every shopper who sent an AI assistant to browse for them this season, and Monday’s conversion rate is lower than it should be.
Watch your traffic come in
Each dot is one visit. Your dashboard counts them all the same way.
- Sessions counted
- 0
- People
- 0
- AI assistants
- 0
What is your real conversion rate?
Enter your own Cyber Weekend numbers, Thursday to Monday. You choose the share of sessions that came from AI assistants. It varies a lot from one retailer to the next, so start with a guess. We can measure the real figure from your data.
Your guess. We can measure the real figure from your data.
Want the real split? We separate your Cyber Weekend sessions into people, AI assistants and uncertain visits, for free.
Book your free traffic checkWhere the wrong number costs money
Split tests. When a test sends visitors to different page URLs or from different sources, one version can draw far more AI visits than the other. Its conversion rate looks lower, and you keep the weaker page.
Audiences. AI visits that leave without buying land in your retargeting and lookalike audiences. Those audiences then teach your ad platforms to find more visitors like them.
You can’t improve a conversion rate that counts machines as shoppers.
Count three groups, not one
Don’t block AI assistants. Many of the shoppers behind them arrive ready to buy. Count the visits separately instead.
The rate to manage
Judge your team and your tests on this number.
Demand you can see early
Which products AI assistants check, how often, and which ones they skip.
The gray area, kept apart
Visits that could be either. Keeping them separate keeps the other two counts clean.
Then give AI assistants credit for the sales they start. Many shoppers come back from an AI assistant’s answer already decided. When a shopper clicks through from that answer and comes back later, signed in or on another device, Syntasa can link the visits. Then fewer of those sales end up as direct traffic.
What you get from Syntasa
Syntasa Customer Intelligence runs inside your own data warehouse. For this problem, it does two jobs:
- Sorts every session into people, AI assistants or uncertain, using the visit data you already keep, such as your analytics export or server logs. It reports conversion for each group.
- Links a shopper’s visits across sessions and devices, so you can see more of the sales AI assistants helped start.
Your current analytics tool stays as it is. Your data stays in your systems. Nothing leaves unless you approve where it goes, and every transfer is logged.
Customer Intelligence is Syntasa’s platform for customer data. The Customer Insights Agent and the Decisioning Agent run on it.
Your first step: book 30 minutes and bring your Cyber Weekend numbers. We’ll show you how we would split your traffic. You leave with an estimate of your real conversion rate and a plan to measure it on your data.
Find out how much of your holiday traffic could actually buy.
On the call, we estimate your real conversion rate from the numbers you bring. Then we agree a plan to split your Cyber Weekend sessions into people, AI assistants and uncertain visits on your own data, before your January review.
- No change to your analytics setup.
- Nothing leaves your systems unless you approve it.
Questions we get
Bot filters remove known crawlers. Some AI assistants announce themselves and are easy to separate. Others open pages in a real browser, much as a person does, and most filters let them through. You may not want to block either kind, because both bring you customers.
No. Let them in and count them separately. Then make sure your product pages give them the facts they look for.
No. Syntasa works alongside it, inside your own data, and adds the three counts.
Related: make sure AI assistants recommend your products, and find out what the holidays brought you.
John Bizimana
Director of Enterprise GTM at Syntasa. You can reach him atjohn.bizimana@syntasa.com.