How LiLi Tile grew add to cart 500% in 30 days with Noibu
LiLi Tile, a handmade cement and clay tile brand on Shopify, replaced PostHog and its other tracking tools with Noibu and fed Noibu's prioritized session data into Owl, its in-house AI system. Four days later a rebuilt site was live, cutting load time from 3.5 seconds to under 1. Over the next 30 days, add to cart rose 500%.
increase in add to cart in 30 days
increase in conversion rate
from connecting Noibu to a rebuilt site
At a glance
TL;DR — key outcomes from LiLi Tile’s use of Noibu
- LiLi Tile replaced PostHog and its other tracking and heat map tools with Noibu, then connected Noibu data into Owl, the AI system its team built in-house.
- Noibu surfaced that site speed was driving drop-off. Four days later, a rebuilt site was live.
- Over the following 30 days, add to cart rose 500% and conversion rate rose 200%.
- Every change was approved by a person before it shipped.
A tile company that wanted to be a technology company
LiLi Tile designs handmade cement, clay and terracotta tile using a process that has barely changed in over a century. The company sells direct to homeowners, designers and architects, and runs hundreds of thousands of sessions through its site every week.
Mory Katan is LiLi Tile’s CTO. He oversees the team that implements changes to the site, the team that reviews analytics and advertising, and the decision about what technology the company brings in. As he puts it, his mission is “to make us the most technologically forward tile company on the market.”
The problem: a beautiful site that wasn’t converting
LiLi Tile’s site looked the way a design brand’s site should. It just wasn’t turning visits into orders. “Beauty only goes so far in terms of web functionality,” Mory says. “If it doesn’t have conversion, you’re missing a big element. You’re just showing a nice picture.”
The team was running PostHog alongside several other tracking tools, heat maps included. The data was being captured. At the site’s volume, none of it was usable.
“There are probably ten solutions on the market for heat maps. But once that heat map is logged, how is it logged, and how is that data able to be extracted? I could see the heat maps and click through them. But clicking through 100,000 sessions a week is not scalable.”
— Mory Katan, CTO, LiLi Tile
Mory had tried other tools before. The limitation was never capture. It was whether anything could be done with what was captured, fast enough to matter.
Why Noibu: data an agent can act on
Mory had considered building deeper analysis into the stack LiLi Tile already had. Two things changed his mind: the capability already existed, and it connected to the AI models his team was already running.
“What made Noibu stand out was that they already had those capabilities, along with a connection to other AI models. If you connect it to a system like Claude, it’s very good at accessing and assessing the data quickly.”
— Mory Katan, CTO, LiLi Tile
LiLi Tile retired its other tracking tools and centralized on Noibu. The change wasn’t access to more data, it was that the data now arrived prioritized. “Rather than having to review 100,000 sessions and see where the heat maps align, Noibu can layer that data and feed actionable insights to Owl,” Mory says.
Owl is the system Mory’s team built in house. It joins Noibu session data with Shopify, the company’s CRM, inventory and warehouse data, then acts on what it finds.
Thursday to Sunday
LiLi Tile integrated Noibu on a Thursday afternoon and connected it directly into Owl, with one instruction: ingest the data, learn from it, and build site updates aimed at improving speed and conversion.
“Thursday night started. By Saturday night, I was already doing reviews. Sunday night I deployed a site — four days — that took site speed from 3.5 seconds to 1 second and under.”
— Mory Katan, CTO, LiLi Tile
The building moved fast. The approval stayed human. “A human was there to approve it, make sure everything looked good, and send it out,” Mory says. Nothing reached customers without him seeing it first.
What the loop looks like in practice
One example shows the full cycle. Noibu surfaced that a fifth of users were getting stuck in a single section of a page — reaching it, stalling, going no further.
“Noibu was flagging, hey, 20% of users are getting caught up in this section of the page. We identified that and said, okay, let’s dig deeper. Let’s move the section up and make an auto-folding section here. Now those 20% of people weren’t dropping off, they were actually getting to where they were looking.”
— Mory Katan, CTO, LiLi Tile
Found, built, approved, shipped, measured, then repeated, with each change informing the next. At LiLi Tile’s volume, that compounding is the whole point. “It’s like a war zone if you can’t make sense of it,” Mory says.
The results
Site load time, before and after the rebuild
Source: LiLi Tile. Load time has since come down to roughly half a second.
Over the 30 days that followed, add to cart rose 500% and conversion rate rose 200%. Load time has since come down again, to roughly half a second, with further conversion gains behind it. (More on why site speed drives Shopify sales.)
Before committing, Mory had asked a fair question: whether a tool like this caps out once the obvious problems are solved. It hasn’t. “If you’re constantly updating and improving, there’s no reason to unplug it,” Mory says.
You don’t need to build an Owl
LiLi Tile is unusual in having built its own AI system. The loop behind the results is not unusual, it’s how Noibu works by default.
Noibu’s own agents run the same cycle: find what’s broken, slow, confusing or leaking revenue, or growth opportunities, build the recommended change, hand it back for approval, ship it once approved, then measure the impact against baseline. Teams can run that inside the Noibu console, or connect Noibu to the AI tools they already use, which is the route Mory took.
What LiLi Tile proves isn’t that you need to build something. It’s that once site data is prioritized rather than merely collected, a very small team can move at a pace that used to require a much larger one.
“I think we were trying to piecemeal a lot of the data from other sources. It’s just made it easier that Noibu organized it for me.”
— Mory Katan, CTO, LiLi Tile
Frequently asked questions
How did LiLi Tile grow add to cart 500%?
LiLi Tile centralized its tracking on Noibu and connected Noibu's prioritized session data to Owl, the AI system its team built in house. Noibu surfaced that site speed was driving drop-off, and four days later a rebuilt site was live with load time cut from 3.5 seconds to under 1. Over the following 30 days, add to cart rose 500% and conversion rate rose 200%.
What tools did LiLi Tile replace with Noibu?
LiLi Tile retired PostHog and its other tracking and heat map tools and centralized on Noibu. The data was already being captured before, but at hundreds of thousands of sessions a week it wasn't usable. Noibu layers that data and turns it into prioritized, actionable insights.
What is Noibu's Performance agent?
Noibu's Performance agent finds what is making a store slow, such as Core Web Vitals and load time problems, using real shopper sessions. It builds the recommended change, hands it back for approval and measures the impact against baseline. Teams can run it in the Noibu console or through the Noibu MCP in AI tools like Claude.
Do Noibu's AI agents change the site on their own?
No. Every change at LiLi Tile was approved by a person before it shipped, and Mory Katan reviewed each one before it reached customers. Noibu's own agents work the same way: they find the issue and build the change, and a person on the merchant's team approves it.
Do you need to build your own AI system to get results like LiLi Tile?
No. LiLi Tile is unusual in having built its own AI system, Owl. Noibu's agents run the same find, build, approve, ship and measure loop by default, either in the Noibu console or connected to the AI tools a team already uses.
Try it on your store
Read how Noibu’s Performance agent finds what’s slowing a store down, or get a free website audit to see what’s holding your site back.
About Noibu
Noibu is AI for ecommerce operations, the always-on optimization system that keeps online stores running at their best. A team of supervised AI agents finds what's broken, slow, or leaking revenue, does the work, and hands results back for human approval. Trusted by retailers like Free People, Guess, Liquid I.V., and Ariat.
