How Ribble Cycles uses Noibu to make its site tell them when it's broken
By connecting Noibu to the tools its developers already use, Ribble now catches and ranks revenue-impacting issues as they happen. Fixes shipped in just 11 weeks protected an estimated £35k–£100k a year in sales.
Estimated annual sales protected by fixes shipped in 11 weeks
Noibu-surfaced issues fixed since connecting Noibu to their AI tools
Issues returned when the team asks what's broken, ranked by what's costing orders
Releases logged to Noibu automatically from GitHub in 11 weeks
“We used to find out something was broken from a dip in sales. Now the site tells us first, and tells us what it's costing.”
— Matthew Lawson, Chief Digital Officer, Ribble Cycles
About Ribble
The challenge: great data, capped by human attention
Ribble Cycles has used Noibu since 2020. In year one it delivered a 42.89x ROI and more than $2.5M in annualized revenue saved. By 2026, Ribble's site was, in Chief Digital Officer Matthew Lawson's words, already clean and optimized.
But the value had a ceiling. Noibu was a dashboard, and someone had to log in to use it. They had to know which tab to trust, and that the big number isn't a bug count. That knowledge lived with two people on the team.
Then Ribble connected Noibu's MCP, which lets AI tools like Claude pull Noibu's data directly (here is how that connection works), and the ceiling came off.
“Before the MCP, Noibu's value was capped by how often a human opened it. Now it isn't capped at all.”
— Matthew Lawson, Chief Digital Officer, Ribble Cycles
The shift: from somewhere you go to something that comes to you
That connection changed where Noibu lives. Ribble's developers already work in Claude Code and GitHub, so Noibu now answers them there, in plain English.
“Noibu already knows what's broken on your site. The MCP means you don't have to go and ask it any more.”
— Matthew Lawson, Chief Digital Officer, Ribble Cycles
Day to day, that looks like this:
- Ranked answers, not raw counts. Ask “what's broken” and the team gets back the nine issues worth acting on, not a raw count of 7,600. They're ranked by what's costing orders, with checkout at the top.
- Evidence for every release. Each merge posts a release marker to Noibu automatically, so “did this morning's release break anything?” has a before and after.
- Monitoring that follows up. After a big change, Noibu keeps watching and flags anything new that appears.
What Ribble built on top
Ribble layered four things onto Noibu's data and its AI connection.
Four layers on top of Noibu
| Layer | What it does |
|---|---|
| Labels | 18 custom attributes and nine funnel events on every session, so a session answers a question instead of just showing an error |
| Markers | A GitHub Action posts a release to Noibu on every merge, so there's always a before and after |
| Recipe | A written triage skill that asks Noibu the same three questions in the same order, with guardrails for every way the data has misled them before |
| Reference code | Shown to customers at the point of failure, so a phone call becomes a session replay |
“Noibu gave us the data and the MCP. We gave it context, release markers and a routine.”
— Matthew Lawson, Chief Digital Officer, Ribble Cycles
What it caught
The best catch never threw an error
A price on Ribble's site showed the wrong currency. The page looked perfect, and no error fired. An error alert alone would never have flagged it.
Because every Ribble session carries custom labels, the team found every affected customer in a single search. Matt calls it the shift from “did it error” to “was it right.”
The fix went live on September 23. Since then, the site has shown the correct currency on nearly 9,000 visits, roughly 300 a day. More than 5,000 of those were UK shoppers, who have gone on to place 58 orders worth £21k.
The issues a daily check can't see
Ribble's triage runs two passes every time. The first is a 24-hour scan for anything a release just broke. The second is a 30-day sweep ranked by damage to the funnel, not by how often an error fires. A handful of sessions a day never clears a daily bar, but each one might be a lost order.
On August 6, 2026, the sweep surfaced five priority issues where the daily scan caught one. Three shared a single root cause. One affected page converted at 4.3% against a 39.3% baseline. Another converted 0 of 59 sessions while erroring on the buy button (how to work through a backlog like this).
The results
Between June 22 and September 9, 2026, Ribble shipped 33 fixes for issues that Noibu surfaced, with all 404 releases in that window logged automatically. Those fixes protected an estimated £35k to £100k in annual sales.
The most conservative estimate comes from three checkout blockers alone, where shoppers clearly wanted to buy and couldn't:
Three checkout blockers, before the fix
Source: Noibu session data for Ribble Cycles, June 22 to September 9, 2026.
Another 16 fixes were verified shopper-impacting problems that Noibu couldn't put a figure on, because too few shoppers hit them in the window. They included Stripe payment-method errors, a billing address error at the delivery step, and a Cloudflare 403 on checkout.
This builds on six years of results. In the first year alone, Noibu saved Ribble more than $2.5M in annualized revenue across 29 shopper-impacting issues, a 42.89x ROI. Read the original Ribble case study.
Noibu for the whole business
Noibu isn't just a developer tool at Ribble anymore. When a customer calls to say they couldn't pay, they read out the short code shown on their screen, and the team is watching their session in seconds. Trading reviews Noibu's priority issues alongside the sales numbers, so a problem is on the table before it shows up in the sales report.
For Ribble, that's the real shift. Problems surface the day they appear, and the people who can act on them see them before the sales report does.
Frequently asked questions
What is Noibu MCP?
How does Ribble Cycles decide which site issues to fix first?
Can Noibu find problems that don't throw an error?
How do release markers help an ecommerce team?
How much revenue did Ribble Cycles protect with Noibu?
Try it on your store
See how Noibu's AI agents find what's breaking checkout, read about the Bug Resolution agent, or get a free website audit to see what's costing your store orders.
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.
