Planning case studies — the disciplines, worked through
Five short case studies showing what the planning disciplines on this site look like when a real operation has to use them — what the situation was, what the data actually showed, what the team did, and what it cost or saved.
The two sites that took it in turns
The situation. A two-site operation — one centre in Scotland, one in the South of England — planned both sites against a single national shrinkage assumption. One site or the other kept missing service in weeks the plan said were comfortable, and the volume forecast took the blame every time. The volumes, when checked, were fine.
What the data showed. Two years of shrinkage, pulled by component and by site, told a different story. The sites had genuinely different shapes: school holidays fell in different weeks, big local festivals landed in different months, and each site had local days that were not official bank holidays but that staff wanted to observe anyway. The national average was roughly right over the year and wrong at each site almost every week — structurally over-covered in one place and short in the other, on a rotation you could set a calendar by.
What the planning team did. They built a separate shrinkage profile per site and then turned the difference into a lever rather than a nuisance. Leave was limited in one location in the weeks it was ramped up in the other; on bank holidays one site could close while the other opened; and when the inevitable winter sickness bug hit one site, offline activity was rebalanced at the other to hold cover.
The outcome. Total shrinkage barely moved — but more staff got the time off they actually wanted, sickness and attrition edged down, and the unexplained service misses stopped. Nobody had put two-site flexibility in the expansion business case. It worked anyway.
The discipline: shrinkage is a forecast, not a number — decompose it, profile it locally, and start the capacity plan there. Deeper: Shrinkage: the planner’s hardest input · shrinkage calculator.
400,000 emails at four o’clock on a Friday
The situation. During the pandemic, a well-meaning comms team sent roughly 400,000 customers an email explaining how to contact the operation — at 4pm on a Friday, without telling the contact centre. A locked-down population with little better to do than scroll their phones read it, remembered they had a reason to call, and acted on it at once.
What the data showed. The queue went from zero to over two hundred in seconds — not a drift, a cliff. That shape mattered: organic demand does not arrive as a vertical line. A step-change that sharp, on a Friday evening, with one dominant contact reason, pointed to a single external trigger rather than a forecasting miss.
What the planning team did. The honest answer is that the staffing levers mostly failed — persuading agents to stay late at 4pm on a Friday is a lever that exists on paper. What worked was diagnosis: identifying the mailing as the root cause within the hour, sizing the affected population, and switching effort to managing the tail into the following week rather than fighting a battle the evening had already lost.
The outcome. The weekend was rough and the recovery was planned rather than improvised. The lasting fix was structural: marketing and comms activity now flowed into the planning calendar before it was sent, and the event was written into a surge playbook — graded by impact and duration — so the next self-inflicted spike would be detected from the send list, not from the queue.
The discipline: you cannot forecast the individual surge, but you can playbook the class — and detection beats heroics. Deeper: The demand-surge playbook · Planning for a marketing campaign.
The surge that never came
The situation. The same disciplines can fail in the opposite direction. An operation’s analysis — persuasive at the time, less so in hindsight — said a major volume spike was imminent. The team did everything the textbook asks: offline activity postponed, staff borrowed from other departments, extra support lined up, everyone briefed.
What the data showed. Nothing. The calls never materialised. The early indicators the team had read as the front edge of a surge turned out to be ordinary volatility wearing dramatic clothes — and every cost of the response was already paid before that became clear. Borrowed staff sat idle, postponed work queued up to be repaid, and the planning team wore the egg.
What the planning team did. Rather than burying the episode, they treated it as the other half of the same lesson as every genuine surge: a playbook needs to grade events by impact and duration, with numerical triggers for when to prepare and when to actually fire the levers. Preparing is cheap; firing is not. The revised playbook separated the two explicitly — line up the response at the first sign, act only when the variance persists — and was reviewed monthly for its first year, quarterly after.
The outcome. The next few scares were handled at the “prepare” stage and cost almost nothing; the next real event was caught no later than the false alarm had been. Premature action on noise pays its full cost for nothing, every time — slightly-late action on a confirmed signal is targeted and usually cheaper.
The discipline: diagnosis before action — sometimes the hardest thing to do is nothing. Deeper: Real-time: the discipline of doing nothing · Noise vs signal · Real-time playbooks.
The green dashboard and the queue it was hiding
The situation. A regulated operation was comfortably hitting its aggregate service level — low eighties, green every week — and on that basis nobody was worried. Then Consumer Duty arrived, with its expectation that outcomes be evidenced by segment, and the planning team cut the data accordingly.
What the data showed. The vulnerable-customer cohort — people in financial difficulty, recently bereaved, or flagged as needing extra support — was waiting materially longer than everyone else and abandoning at nearly twice the headline rate. Their contacts landed disproportionately in an understaffed late-afternoon window and routed to a general pool with no specialist skilling. The aggregate had been averaging their poor experience away: nothing in the headline was wrong, it was simply blind to exactly the customers the regulator looks at first.
What the planning team did. The fixes were ordinary planning moves, which is rather the point. A small specialist group was skilled for the flagged contacts; the late-afternoon coverage was protected instead of being the default place to give ground; and the vulnerable-cohort metrics — wait time, abandonment, routing success — went into the weekly pack as a separate line with its own triggers, so the gap could never silently reopen.
The outcome. The cohort’s abandonment fell back towards the headline rate within a quarter, and the operation could evidence the improvement — which, under the Duty, is half the job. The blended number stayed green throughout, before and after. That was the lesson.
The discipline: the aggregate isn’t just less useful than the segmented view — it can be the thing actively hiding the risk. Deeper: Planning for vulnerable customers · Consumer Duty for planners.
Two saves: the one nobody heard about, and the one they did
The situation. Two incidents, a couple of years apart, operationally almost identical. In the first, a planner spotted that a product team’s mailing — around half a million letters — was due to land in the thinnest-covered week of the quarter. In the second, a billing-system change certain to drive a contact spike was spotted early in just the same way.
What the data showed. Both times, the capacity model made the collision obvious weeks out: demand and cover crossing in a way that modelled to a service collapse if nothing changed.
What the planning team did. First time: the mailing was split across three weeks, training was moved, and the fortnight passed without incident — and nobody outside the team was ever told. The save’s only monument was the disaster’s absence. Second time, the same operational move came with four sentences of communication: the save went into a dated log the week it happened, appeared as two factual lines in the monthly pack, and surfaced in the quarterly review with the counterfactual attached — without the early move, the fortnight modelled out around 70% against the 82% delivered — claimed as a contribution, and reported alongside that quarter’s forecast miss.
The outcome. After the silent save, the function was described in the budget round — without malice — as “the team that does the reports”, and lost a head. After the visible one, the team was in the room when the next year’s decisions were made. The difference in cost between the two saves was nothing; the difference in consequence was everything.
The discipline: invisible good work gets cut; honestly evidenced work gets invested in. Keep the saves log. Deeper: Showing the planning team’s success · Planning function credibility.
Want the disciplines behind the stories? Start with the planning cycle, sense-check your numbers against the benchmarks, or put your own figures into the calculators.