Kaizan Wants to Hear a Client Leaving Before They Say It

Churn almost never arrives as a phone call. It arrives as a reply that used to take an hour and now takes three days. As a stakeholder who quietly stops being copied in. As a meeting that keeps slipping to next week, then next quarter, then off the calendar. By the time a client actually tells you they are leaving, the decision is months old. Kaizan, a London startup that just raised £2.5m, is built on the belief that all of that is legible - if you have something patient enough to read it.

The company calls itself "client super intelligence for client service teams." Strip the marketing and the job is narrower and more interesting than that: pull together every scrap of communication flowing between a firm and its clients - emails, calls, chat, CRM notes, project tools - and use AI to score how healthy each relationship actually is. Not how healthy the account manager says it is on the Friday pipeline call. How healthy it is in the evidence.

On 11 May 2026, Kaizan announced a £2.5m seed round led by Pembroke VCT, with Velocity Capital and a group of angel investors. The money is earmarked for two things founders always say and rarely mean equally: product, and America. Kaizan means the second one. The category it plays in - client success, account management, professional services software - is enormous, unglamorous, and mostly still run on gut feeling and spreadsheets.

01 / The insightThe data already exists. Nobody reads it together.

The frustration at the heart of Kaizan is one anyone who has run an account will recognise. The information that predicts a client leaving is not hidden. It is sitting in plain sight, spread across ten tools that never talk to each other. The warning signs are in the inbox of a junior who did not think to flag them, in a call recording nobody re-listens to, in the gap between the last three emails. No single person sees the whole picture, so no single person acts.

Founders Glen Calvert and Pravin Paratey started the company around 2021 to close that gap. Calvert, the CEO, had spent years in adtech and agency-adjacent businesses where client relationships are the product. Paratey, the CTO, brought the engineering. In its first raise back in 2022, Kaizan was pitched as a "Siri for client teams" - a phrase that now feels quaint. The 2026 version is less about asking questions and more about a system that watches on its own and speaks up when something is off.

A quiet client is not a happy client. It is often a client already talking to your competitor.The premise behind Kaizan's health scoring

02 / The frameworkCARE, or how to turn a feeling into a number

The engine underneath is a model Kaizan calls CARE. It is an acronym, and like all acronyms it is slightly forced, but the underlying idea holds up. It breaks the vague question "is this client OK?" into four things you can actually measure and track over time.

What makes this more than a dashboard is what Kaizan bolts on top: AI "helpers" that do not just report the score but act on it. A helper can draft the reply to a cooling client, assemble the quarterly business review deck from the quarter's actual conversations, or flag that a decision-maker has gone silent. The pitch is that an account manager watching five clients can suddenly watch fifty, because the software does the reading and hands over only what needs a human.

There is also a quieter, more technical bet. Kaizan connects to tools like Claude and ChatGPT via MCP, so a team member can ask about an account in plain language - "why is Acme unhappy?" - and get an answer cited back to the original emails and calls. The receipts matter. A health score no one trusts is a health score no one uses.

03 / The rivalKaizan listens. Dock builds the room.

To understand where Kaizan is pointed, it helps to look at the company most often mentioned in the same breath. Dock, based in the US, is chasing the same prize - keeping and growing client relationships - from the opposite direction. Where Kaizan is invisible, sitting behind the scenes and inferring health from signals, Dock is the thing everyone can see: shared digital workspaces, deal rooms, onboarding portals and client hubs where vendor and customer collaborate in the open.

Neither approach is obviously right, which is what makes the contrast worth watching. Dock's advantage is honesty of data: if a client logs into the portal and opens the proposal three times, you know it, no inference required. Its weakness is the oldest problem in software - getting people to actually use the thing. Kaizan's advantage is that it demands nothing of the client; the client just carries on emailing and talking, and the system reads along. Its weakness is that inference is inference. A cheerful email can hide a bad decision. A silent client might just be on holiday.

One is a sensor. One is a surface. Watching which one wins tells you where B2B relationships are actually headed.The Kaizan vs Dock question

04 / The wagerWhy retention is the whole game

There is an unglamorous truth under all of this: keeping a client is cheaper than winning one, usually by a wide margin. Every service business knows it. Almost none of them run their retention with the same rigour they bring to sales. Sales has a pipeline, a forecast, a CRM everyone updates because their commission depends on it. Retention has a customer success manager, a gut feeling, and a quarterly panic when a big logo gives notice.

Kaizan's real wager is that AI has finally crossed the line where retention can be made systematic - where "which of my clients is quietly slipping?" becomes a question you answer on a Monday morning instead of discovering in a cancellation email. The 21%+ average revenue growth per client the company cites is the headline version of that claim. Investors clearly bought some of it. "Kaizan addresses key operational challenges faced in service-led organisations," said Alicia Taylor, investment director at Pembroke Investment Managers, "using advanced communication and sentiment analysis to deliver a single, unified view of client health."

It is worth being sober about the limits. A sentiment score is a guess dressed as a metric, and the failure mode of tools like this is the confident false positive - a client flagged as at-risk who was fine, or worse, the calm score on an account that walks anyway. There is also the plainer discomfort of a system that reads every client email in the building. Kaizan is not alone in that; it is simply honest that reading is the product. The teams that get value from it will be the ones that treat the score as a prompt to go talk to a human, not a substitute for doing so.

05 / What to watchThe next twelve months

The US expansion is the real test. The American market for this software is deeper and far more crowded - Dock is only one name on a long list, and the incumbents in CRM and customer success are not standing still. A £2.5m seed does not buy a land war; it buys a sharp point of the wedge. Kaizan's is that it does not ask the client to change anything. Whether that is enough of an edge to pull share from tools people are already paying for is the question the next twelve months will answer.

For anyone running client relationships, the more useful takeaway is smaller and does not require buying anything. The signals Kaizan reads are ones you can watch by hand: how fast a client replies now versus three months ago, how many people at the account you actually know, whether the last few conversations felt warmer or cooler. Kaizan's bet is that you are too busy to track fifty of those at once. It is probably right. That is the whole business.