Under the hood

The ranking doesn't just look at what you type. It looks at what customers actually buy.

When a customer searches for something in chat, Aisly doesn't just match keywords against your catalog: it also weighs what real visitors to your store have actually added to their cart. Below is the real formula, and a simulation of what it can do once behavioral data starts coming in.

What's below is a simulation, not the result of an actual Aisly customer. Text similarity is computed on the real vectors of our demo catalog; the "true" conversion rates are a declared scenario assumption, chosen to show what the engine is capable of — not a result already achieved by someone. Your store, your traffic, and your customers determine the real numbers.
+65.9%
more add-to-carts after 25 simulated days, adaptive ranking vs. a static ranking based on text similarity alone
Day 7
when the product with the best simulated conversion rate reaches 1st place, starting from 8th
€600
price of the promoted product: the cheapest in its range — text similarity alone would have left it stuck mid-ranking

The ranking corrects itself

25 simulated days · 40 visitors/day · top 4 positions shown each day

Position of "Collection Snowboard: Hydrogen"
By pure text similarity it started 8th out of 12
1st 3rd 5th 7th 9th d.1 d.13 d.19 d.25
Cumulative add-to-carts
Adaptive engine vs. static ranking (similarity only)
150 100 50 0 d.1 d.13 d.19 d.25
adaptive (146) static (88)

No rewrite for this page

It's the same SQL function running in production, not a simplified version

score = similarity × ( 1 + (add_to_carts + 1) / (impressions + 5) )
similarity is the real cosine similarity between the product's embedding and the search's (pgvector). Bayesian smoothing — a weak prior, 1 cart add per 5 "imaginary" impressions — keeps a product with no history from sitting at zero, and one with very little data from dominating on statistical noise. It runs in sql/002_personalized_ranking.sql, the same function that powers the advanced analytics dashboard where you see what actually converts.

What we assumed for the scenario

For every product: similarity is real. The "true" add-to-cart rate is a declared assumption — not observed data — chosen to represent a common e-commerce phenomenon: a product that's mid-pack by semantic similarity can convert far more (or less) than the text alone would suggest.

ProductReal similarityAssumed conversion rate
The Multi-managed Snowboard0.94504.0%
The Multi-location Snowboard0.93873.5%
The 3p Fulfilled Snowboard0.92051.0%
The Out of Stock Snowboard0.90770.5%
The Collection Snowboard: Oxygen0.90303.0%
The Compare at Price Snowboard0.90002.0%
The Inventory Not Tracked Snowboard0.89232.0%
The Collection Snowboard: Hydrogen0.89199.0%
The Collection Snowboard: Liquid0.89063.0%
The Videographer Snowboard0.83621.5%
Selling Plans Ski Wax0.79520.5%
Gift Card0.69410.2%

Day-by-day log

What the engine would have shown in 1st place that day, and the cumulative count under both scenarios

Day1st place (adaptive)Hydrogen rankAdds/day adaptiveAdds/day staticCum. adaptiveCum. static
1The Multi-managed Snowboard85353
2The Collection Snowboard: Oxygen4871310
3The Collection Snowboard: Liquid3221512
4The Multi-location Snowboard3762218
5The Multi-managed Snowboard2622820
6The Multi-managed Snowboard2613421
7The Collection Snowboard: Hydrogen1964327
8The Multi-managed Snowboard2344631
9The Multi-managed Snowboard21165737
10The Collection Snowboard: Hydrogen1716438
11The Collection Snowboard: Hydrogen1847242
12The Collection Snowboard: Hydrogen1327544
13The Collection Snowboard: Hydrogen1838347
14The Collection Snowboard: Hydrogen1638950
15The Collection Snowboard: Hydrogen1839753
16The Collection Snowboard: Hydrogen15610259
17The Multi-managed Snowboard26110860
18The Multi-managed Snowboard22711067
19The Multi-managed Snowboard25311570
20The Multi-managed Snowboard23211872
21The Multi-managed Snowboard23212174
22The Multi-managed Snowboard25312677
23The Multi-managed Snowboard27513382
24The Multi-managed Snowboard26313985
25The Multi-managed Snowboard27314688

What this shows, and what it doesn't. It shows that the real scoring function, without any modification, shifts the ranking toward what actually converts as behavioral events come in — with genuine similarity and conversion assumptions stated in full above.

It doesn't show a result for your store: traffic, catalog, and actual customer behavior vary, and none of the numbers above are a performance promise.

Demo catalog aisly-test · 13 embedded products · simulation seed 42 · reference query: similarity to "The Complete Snowboard"

The ranking learns
from every cart.

14 days free. No credit card required. Cancel anytime.

Create your free account →