They Fixed the Wrong Thing for Two Years

A Bristol HR tech firm had a retention problem. Thirty-one percent of new hires left within their first 12 months. Industry average for their size: 18-22%. The gap was obvious. The cause wasn't.

Over two years, the firm tried the standard responses. Salary benchmarking against market rates (the salaries were competitive; the problem wasn't compensation). An employee engagement survey (scores were middling; the survey identified "onboarding experience" as a concern, but the firm's response was to add a welcome lunch and a better first-day schedule, which addressed the symptom at the wrong resolution). A culture initiative (values workshop, team-building days; the values were fine, the culture was fine, the people leaving weren't leaving because of culture). Manager training on retention conversations (the managers had the conversations; the conversations didn't change anything because the conversations weren't the problem).

Each initiative was reasonable. Each addressed retention at the category level. None fixed the specific operational failure that was producing the turnover.

The specific failure: a 90-day onboarding process with 34 milestones, tracked in a spreadsheet by an operations manager who also ran facilities, vendors, and office operations. She checked the spreadsheet at weeks 1, 4, and 12. Between those check-ins, milestones drifted. By probation review, 29% were incomplete. Four of 17 hires reached probation without completing mandatory compliance training. The exit interviews said "I didn't feel set up properly." The firm heard "retention problem" and reached for retention-shaped solutions. The problem was a spreadsheet checked three times in 90 days.

This is, when you think about it, a pattern that costs businesses an extraordinary amount of time and money: treating a visible symptom with category-level solutions when the cause is a specific, fixable, operational bottleneck that nobody has identified because nobody looked at the right resolution.

The instinct is understandable. When the symptom is big (31% turnover, 18% churn, £640,000 in lost revenue), the response feels like it should be big too. A retention strategy. A client experience overhaul. A digital transformation programme. The size of the response matches the size of the problem. It feels proportionate.

The instinct is also, in most cases, expensive and wrong.

Tom at the Leeds logistics firm lost 25 clients last year. Eighteen were surprises. The firm's response before working with us: a client satisfaction survey (sent annually, which means the data was already 6-12 months old by the time anyone read it), account manager training on relationship building (the AMs were already good at relationships; the problem was signal visibility), and a quarterly business review overhaul (the QBRs were fine; the problem was the 11 months between them where nobody was aggregating health signals). Category-level retention initiatives. The actual bottleneck: four systems holding health signals that nobody viewed simultaneously. One agent. £260 per month. Surprise cancellations down 62% in the first quarter.

Sophie at the London SaaS company had 45 surprise cancellations. £414,000 in revenue. The firm had invested in a customer success team, built QBR processes, created escalation playbooks. The category was covered. The bottleneck was specific: five tools holding health signals, aggregated by four people at a weekly meeting from memory. One agent. £280 per month. Monday meeting: 90 minutes down to 20.

Marcus in Houston checked his bank balance Monday mornings and hoped. The category response would have been a CFO hire, a financial management system, or a consulting engagement to "fix the finance function." The actual bottleneck: three data sources not connected, producing no forward view. One agent. $240 per month. Sunday night anxiety became a Monday morning dashboard.

The pattern across every case: the business was working at the category level. The solution was at the bottleneck level. The category work was expensive, slow, and diffuse. The bottleneck fix was specific, fast, and measurable.

Three reasons the specific operational failure stays invisible while the category gets all the attention.

The symptom looks like a big problem, so people reach for big solutions. "31% turnover" sounds like a company-wide issue. It triggers company-wide responses: culture, compensation, leadership. Nobody's first instinct when told "your turnover is too high" is to ask "what does your onboarding spreadsheet look like?" The resolution of the diagnosis doesn't match the resolution of the symptom. The symptom is visible at 30,000 feet. The cause is visible at ground level. Most businesses diagnose from 30,000 feet and never descend.

Category solutions feel safer. A salary review is defensible. A culture initiative looks responsible. A training programme demonstrates commitment. A single operational fix to a spreadsheet tracking process feels, to the person proposing it, embarrassingly small relative to the problem. "We have a 31% turnover problem and the solution is to automate Claire's onboarding spreadsheet" is not a sentence that sounds proportionate. It is, however, accurate. And the fact that it doesn't sound proportionate is precisely why the fix gets overlooked for two years while the firm runs salary benchmarking exercises that change nothing.

Nobody connects the operational data to the strategic symptom. The HR director sees the turnover number. The operations manager sees the onboarding spreadsheet. The HR director doesn't examine the spreadsheet because onboarding tracking is an operational detail. The operations manager doesn't connect the spreadsheet to the turnover number because retention strategy isn't her domain. The problem sits in the gap between two people's responsibilities, visible to neither at the right resolution. The invisible 30% isn't just invisible work. It's invisible causation. The unnamed operational task that's producing the named strategic symptom.

A framework for finding the specific operational cause behind a category-level symptom. Three questions.

What do the people who experienced the problem say? Not the managers. Not the HR director. The people who left. The clients who cancelled. The invoices that went unpaid. Exit interviews at the Bristol firm said "I didn't feel set up properly." That's not a compensation problem. That's not a culture problem. That's an onboarding execution problem. The phrasing points at the resolution. Listen to the phrasing.

Where in the process does the first failure occur? Not the final failure (the resignation, the cancellation, the missed payment). The first failure. At the Bristol firm: the buddy introduction that didn't happen in week 1. The compliance training that stalled at day 14 and nobody chased until day 76. The IT setup that took 9 days. The first failure is usually operational, specific, and fixable. The final failure is the symptom that triggers category-level responses.

Can you describe the fix in one sentence? If the fix requires a paragraph, you're still at the category level. "Automate the onboarding milestone tracking so milestones are prompted, chased, and escalated automatically" is one sentence. "Implement a comprehensive employee experience strategy spanning culture, compensation, development, and retention" is a category. The one-sentence fix is the bottleneck. The paragraph is the detour.

Claire's fix: one agent, £180 per month. Milestone completion: 71% to 96%. Compliance gaps: eliminated. The turnover number won't be fully clear for another 12 months, but the exit interviews have already changed. "I felt supported from day one" has replaced "I was still figuring things out at month 3." The operational data is connecting to the strategic symptom in the direction the firm spent two years hoping it would.

The most expensive thing in most businesses isn't the problem. It's the two years spent solving the wrong version of the problem. The salary reviews that didn't change turnover. The client surveys that didn't prevent churn. The culture workshops that didn't fix what culture wasn't causing. The digital roadmaps that recommended 47 things when one thing was the bottleneck.

Every business in this series arrived at the same conclusion: they'd been working at the wrong resolution. The symptom was visible at the strategic level. The cause was hiding at the operational level. And the gap between those two levels was filled with expensive, well-intentioned work that felt proportionate to the problem and did nothing to solve it.

The fix, when someone finally identifies the actual bottleneck, is usually small, fast, and embarrassingly inexpensive relative to the category-level work that preceded it. £180 per month for onboarding tracking that a £75,000-per-year culture initiative couldn't touch. £260 per month for client health scoring that an annual satisfaction survey never surfaced. £220 per month for revenue recognition that a £12,000 consulting engagement would have added to a roadmap nobody implemented.

One bottleneck. One fix. The fix is always smaller than expected. The cost of not finding it is always larger.

This is the AdAI series. Every week we publish the mechanics behind a real AI agent build and the strategic lesson it contains. Free to read. No signup required.

Want the next pair in your inbox? Subscribe to AdAI News. One blueprint and one strategic framework, every Thursday morning.

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