The Invisible 30%

Sophie runs customer success at a London SaaS company. Her job title says "Head of Customer Success." Her job description mentions client relationships, retention strategy, and account growth.

Her actual week, 35-40% of the time, consists of: checking Mixpanel for usage trends across 420 accounts. Opening Intercom to read support tickets for tone (is this client frustrated or just asking a question?). Scanning Stripe for failed payments. Reviewing HubSpot for approaching renewals. Reading Typeform NPS responses. Noting which accounts have gone quiet. And then attending a 90-minute Monday meeting where she and three colleagues report on those 420 accounts from memory.

Nobody hired Sophie to check dashboards. Nobody hired her to read between the lines of support tickets looking for early signs of unhappiness. Nobody hired her to attend a weekly meeting whose primary activity is remembering things that happened across five disconnected tools in the preceding seven days.

They hired her to build relationships, reduce churn, and grow accounts. But somebody has to do the monitoring. The five tools don't talk to each other. There's no aggregated view. No alerts. No system that says "this account is drifting." So Sophie does it. Manually. In the gaps between her actual job.

She has never logged this work. It doesn't appear on any timesheet (because who would log "spent 20 minutes checking five dashboards for signals of client unhappiness"?). If you asked her "how do you spend your week?" she'd say "managing client relationships." She wouldn't describe the dashboard surveillance, the cross-tool synthesis, the mental scorekeeping across 420 accounts that makes the relationship management possible. That's not a job. It's the gap between the job and the tools. And the gap has a cost she's never calculated.

Every business has work that nobody named. Every Blueprint we've published found it sitting in the middle of the operation, consuming the most capable person's time, invisible to everyone including the person doing it.

At the Manchester software consultancy, James spent 1.5 days per month-end cross-referencing Teamwork, SharePoint, and Xero to calculate revenue recognition. His job title said "Finance Director." One and a half days of every month said "human data bridge between three systems that refuse to acknowledge each other's existence."

At the Edinburgh physio clinic, Fiona spent 10 hours per week tracking referrals across email, NHS Gateway, and fax. Her job title said "Practice Manager." Her actual function, for 10 hours of every week, was "intake channel monitor, spreadsheet updater, and NHS reporting engine, who also sometimes managed the practice (when she had time, which was Fridays, if nothing else went wrong, which it usually did)."

At the Toronto contracting company, Ray spent his mornings calling down a list of subcontractors. His job title said "Project Manager." His actual function was "human matching algorithm with 18 years of training data stored in one brain, operating without documentation, backup, or version control."

At the London SaaS company, Sophie spends 35-40% of her time synthesising signals across five tools to determine which of 420 accounts might be unhappy. Her job title says "Head of Customer Success." Her actual function, for a third of every week, is "client health monitoring system, running on biological hardware, with a weekly reporting cadence of once per Monday and a data retention rate of whatever she can remember."

This is the work nobody named. Not in any job description. Not on any task list. Not in timesheets. Not discussed in performance reviews. Not assigned a budget line.

But it consumes 30-40% of the most capable person's time. In every business. Across every industry we've looked at.

The big idea behind everything AdAI does is four words: your team is the software. This is what those four words mean. Your team is the monitoring system. The integration layer. The matching engine. The quality checker. The compliance tracker. Work that should be done by systems is being done by people. Expensively. Invisibly. Without anyone ever putting a name on it, let alone a cost.

Three reasons this work never gets named, never gets counted, and never appears on anyone's radar until someone takes the time to look for it (which, given that the people who would look for it are busy doing it, tends not to happen).

It's embedded in "real" work. Sophie doesn't check Mixpanel and then do client success. Checking Mixpanel IS part of doing client success, in her mind. The monitoring and the job have merged. She's never separated them because they've always been intertwined. James doesn't cross-reference three systems and then do finance. The cross-referencing IS the finance work. The bridge and the destination have become the same activity. Asking "how much time do you spend on cross-referencing?" is asking someone to separate water from tea. They don't experience them as distinct.

Nobody tracks 3-minute tasks. Checking a dashboard takes 3 minutes. Reading a support ticket for tone takes 2 minutes. Scanning an NPS response takes 1 minute. Nobody logs these. Nobody would log these. They're too small to notice individually. But 3 minutes per account, 15 accounts checked per day, 5 days per week: 225 minutes. 3.75 hours. Per CSM. Times four CSMs: 15 hours per week of dashboard checking alone. None of it tracked. None of it visible in any reporting. Fifteen hours of human monitoring that doesn't exist on any timesheet, budget, or workload assessment (because the individual units are 3 minutes long, and nobody has ever suggested logging 3-minute tasks, because that would itself be a task, and we'd need a system for that, and the system would probably be a spreadsheet someone updates when they have time, and we're back where we started).

The person doing it considers it "part of the job." Sophie considers dashboard monitoring part of her job. She's right, in the sense that somebody has to do it. She's also, quite specifically, wrong, in the sense that the work only exists because five systems don't aggregate their signals. If they did, she'd have a health score updating daily instead of a mental synthesis updating whenever she gets around to checking. She'd spend 3.75 hours per week on client strategy instead of client surveillance.

The word businesses use for people who do this well is "dedicated." It's actually "compensating." Dedicated sounds like a virtue. Compensating describes a system that failed to be built. Karen was dedicated. Fiona was dedicated. Dave was dedicated. Sophie is dedicated. They were all compensating for systems that don't exist, doing work that nobody named, absorbing costs that nobody counted.

A framework for making the unnamed work visible. Four steps. The results will be uncomfortable in proportion to how capable your team is (because capable teams compensate most effectively, which means their invisible work is the most expensive).

Step 1: Shadow your most capable person for one day. Not their calendar. Their actual activity. Every dashboard check. Every cross-reference. Every "let me just look at this." Every follow-up that wasn't triggered by a system alert but by their memory or instinct.

Step 2: Categorise every activity. Named: tasks that appear in their job description, are tracked on timesheets, or get discussed in performance reviews. Unnamed: everything they do to bridge, monitor, check, cross-reference, and synthesise across systems that don't connect.

Step 3: Calculate the percentage. Unnamed minutes divided by total minutes. Most owners expect 10-15%. The actual number, when measured, is 30-40%. Every time.

Step 4: Multiply. Unnamed percentage times annual salary equals the annual cost of the work nobody named.

Sophie: 35% of £55,000 = £19,250 per year in unnamed monitoring work. Per CSM. Times four: £77,000 per year. For the entire CS team to do work that a £280/month health scoring agent now handles.

James: 35% of month-end time at £45 per hour = £6,480 per year in unnamed cross-referencing. For work that a £220/month daily data sync eliminated.

Fiona: 10 hours per week of unnamed referral tracking. £26,000 per year. For work that a £140/month monitoring agent reduced to 2 hours per week.

The numbers are always larger than expected. Because nobody has ever counted 3-minute tasks before. And 3-minute tasks, accumulated across a week, a month, a year, and a team, are the most expensive line item that has never appeared on any budget.

The moment you name the invisible work, you can decide what to do about it.

Sophie's team was spending £77,000 per year monitoring client health across five tools. The agent that aggregated health signals costs £280 per month (£3,360 per year). The team's monitoring time dropped by 60-70%, freeing roughly £50,000 per year in capacity. Capacity that went back into client relationships, account growth, and the work that actually prevents churn (which, as it happens, is the work that "Head of Customer Success" describes and that Sophie had insufficient time to do while she was busy being a human monitoring system).

James was spending £6,480 per year cross-referencing three systems. The agent costs £220 per month. His month-end went from 1.5 days to 2 hours. The recovered time went into financial analysis the MD had been asking for and James had never had time to produce.

Fiona was spending £26,000 per year on unnamed referral tracking. The agent costs £140 per month. She got 8 hours per week back for patient flow, staff development, and the practice management she was actually hired to do.

In every case, naming the invisible work was the first step. Calculating it was the second. Fixing it was the third. The fix was never a new hire (you'd need to find someone willing to be a full-time dashboard checker, which is not a job posting that attracts talent). The fix was a system that does the bridging, monitoring, and cross-referencing that humans had been absorbing invisibly for years. The most expensive line item nobody had seen, replaced by the cheapest line item on the software budget.

Your team is doing work nobody named. It doesn't appear on any job description, any timesheet, or any budget line. It's there. Every day. Consuming 30-40% of your most capable people's time.

Shadow one person for one day. Count the unnamed tasks. Calculate the percentage. Multiply by the salary.

If the number is above 20% (and it will be), the AI Workflow Diagnostic helps you identify which unnamed work has the highest cost and which systems should be talking to each other. Takes 10-15 minutes.

Or grab Unstuck. Thirty-nine stories. Ten characters. Ten industries. Same pattern. Karen, Fiona, Dave, Ray, Marcus, Pam, Rachel, Linda, James, Sophie. Your team is the software. It doesn't have to be.

by SP, CEO - Connect on LinkedIn
for the AdAI Ed. Team