Your Client Decided Weeks Before They Told You

Most account teams find out a client is unhappy at the moment the client says so out loud. By then the decision has usually been made. Not in the meeting you attended, but in an internal conversation a few weeks earlier that you were never invited to. The review call where you presented good numbers and everybody nodded was already after the fact. I have sat on both sides of that. At Jumia I was the client being pitched to and managed. At Meta and at Aleph I have been the partner trying to hold relationships across several markets at once. The pattern is the same everywhere. The signals were there. Nobody was reading them, because reading them was nobody's actual job. ## What account teams already know but cannot see Every client relationship generates evidence continuously. How quickly they reply. Whether the senior person still joins the call or sends a deputy. Whether they are asking about new work or only about invoices. How many people from their side are still in the thread. Whether the tone of written communication has changed. An experienced account director senses most of that. The problem is that sensing does not scale, does not survive a handover, and does not show up in a pipeline review. When the account director leaves, the sense leaves with them. That is the gap worth closing. Not replacing judgement. Making the inputs to judgement visible to more than one person. ## Three things that work **Account health scoring.** Combine the operational signals you already have: response times, meeting attendance seniority, support ticket volume and sentiment, scope changes, invoice payment behaviour. You are not trying to produce a perfect number. You are trying to rank accounts so the weekly review starts with the three that moved rather than the ones somebody remembered. **Renewal and expansion signal.** The same evidence points the other way too. Clients asking about capabilities outside current scope, new stakeholders appearing, volume creeping up. Most teams discover an expansion opportunity when the client eventually asks. Catching it earlier changes who framed the conversation. **The administrative load.** Meeting summaries, action tracking, review preparation, account plans that are actually current. This is the least glamorous application and often the one that returns the most time. An account manager spending a day a week on reporting is a day a week not spent with clients. ## The part that goes wrong Here is where I have to say the uncomfortable thing. The fastest way to get value from AI in client success is also the fastest way to create a problem, because the material is client confidential. Meeting transcripts. Commercial terms. Internal assessments of client stakeholders, which are sometimes blunt. Contract detail. When an account manager pastes that into a free tool to get a summary before a call, they are being efficient and creating exposure at the same time. I call it BROAI. Bring Your Own AI, after the Bring Your Own Device problem. In client facing teams it is close to universal, because these are exactly the people under the most time pressure. The answer is not a memo telling people to stop. It is an approved tool that does the job, so the shortcut and the sanctioned route are the same route. ## What good looks like Signals visible to the team rather than held in one person's instinct. An approved place to process client material. Somebody accountable for what the scores are used for, because a health score that nobody acts on is just another dashboard. SWITCH, the framework I use with clients, opens with Scan and Score for this reason. Know what you actually have before buying anything. A question for your next account review. When you lost your last significant client, how long before they told you did the evidence exist in your own systems? ## Frequently asked questions **What is customer health scoring?** A ranking of accounts by risk, built from operational signals the business already generates, so attention goes to the relationships that have changed rather than the ones somebody happened to mention. **Can AI predict which clients will churn?** It can flag which accounts have changed behaviour in ways that historically preceded loss. The judgement about what to do stays human. **What should a client success team automate first?** Meeting summaries and action tracking, usually. Lowest risk, fastest payback, and it returns time to the people who should be in front of clients. **Is it safe to put client meeting notes into AI tools?** Only into tools your organisation has approved and controls. Client confidential material in a free consumer account is an exposure regardless of how useful the summary is. **Does this replace account managers?** No. It removes the administrative work that stops them managing accounts, and makes the warning signs visible to more than one person. *Stanislaus Martins has led client and partner relationships across Sub Saharan Africa at Meta and Aleph, and now advises and trains enterprise teams on practical AI adoption. Formats are on the [speaking and training page](/speaking), or start a conversation on the [work with me page](/work-with-me).*