Seasonal Knowledge: The Once-a-Year Processes Nobody Remembers How to Run

Every company has a handful of processes that only run once a year. The financial year-end close. The external audit. The budget cycle. Insurance renewals. Compliance filings. The seasonal ramp-up. They are among the most important things the business does, and they happen so rarely that nobody remembers exactly how.

So the same scene repeats every cycle. Someone digs through last year’s emails, finds a spreadsheet three versions out of date, and calls the one person who ran it last time - who has since changed roles, or retired, or left. The process gets re-learned under deadline pressure, last year’s mistakes get repeated, and a week of senior time evaporates rebuilding knowledge the company already had and then lost.

This is the problem of seasonal process knowledge. It is uniquely fragile because it combines the three things that erase institutional memory: it happens rarely, it lives in people’s heads rather than in systems, and the people move on. This guide explains why it disappears, what re-learning it actually costs, and how a company memory that stays connected to your real systems can hold the once-a-year playbook so it survives turnover and gets sharper each cycle.

The Once-a-Year Problem

Most knowledge management thinking is built around daily work: the ticket queue, the sales pipeline, the production line. But the processes that hurt most when they go wrong are often the ones that run least often. They carry hard deadlines, regulatory weight, and financial consequences, and the knowledge to run them is concentrated in a few heads.

• Rarity breaks the feedback loop - a task you do every day corrects itself through repetition. A task you do once a year gives you twelve months to forget the steps, the exceptions, and the reasons behind them.
• The knowledge is tacit - the person who runs the annual audit knows which numbers the auditor always questions and which folder holds the reconciliation nobody documented. That judgment was never written down. Nonaka called this tacit knowledge: personal, context-specific, and hard to formalise.
• The owner moves on - roles change, people retire, colleagues leave. When the one person who ran last year’s cycle is gone, the playbook goes with them, because it only ever existed in their memory.
• The stakes are high - a missed filing deadline, a botched renewal, or a delayed close is not a minor inconvenience. It draws penalties, audit findings, coverage gaps, or a nervous board.
• Nobody owns the gap - because the process runs once a year, no one is responsible for keeping its knowledge current the other fifty-one weeks. It is everyone’s problem for one week and nobody’s problem for the rest.

The result is a strange blind spot. Companies invest heavily in tools for the work they do every day and leave their highest-stakes annual processes running on the memory of one or two people. Here is what that actually costs.

Why Seasonal Knowledge Is Uniquely Fragile

Every kind of knowledge in a company faces some risk of loss. Seasonal process knowledge faces three at once, and they compound. Understanding why is the key to fixing it.

Three forces that attack it together

1. Low frequency starves the memory - human recall of a procedure fades sharply without practice. Between two runs of an annual process sit roughly 250 working days of other work. By the time the cycle returns, even the person who ran it is reconstructing rather than remembering.
2. Tacit judgment resists documentation - the steps might be written somewhere, but the reasoning is not. Which supplier always sends the invoice late. Which line the auditor probes. Why the workaround from two years ago exists. Gartner estimates that 70 to 90 percent of enterprise information is unstructured and mostly uncaptured, and the judgment behind a rare process is exactly this kind of content.
3. Turnover removes the carrier - low frequency and tacit knowledge would be survivable if the same person ran every cycle for decades. They do not. In Germany, 12.9 million working people will reach retirement age within fifteen years, and around 30 percent of the current workforce is set to retire by 2040. When the carrier leaves, the uncaptured playbook leaves with them.

The multiplier effect

Each force alone is manageable. Combined, they create a knowledge type that is almost designed to disappear. A daily task survives turnover because the team shares it. A well-documented rare task survives because the document holds. Seasonal process knowledge often has neither the team reinforcement nor the reliable document, so it depends entirely on one person continuing to be there and continuing to remember.

What Re-Learning Rare Processes Actually Costs

“We figured it out eventually” hides a real bill. The cost of re-learning a rare process is paid in senior hours, missed deadlines, repeated errors, and outside fees. It is rarely on a line item, which is exactly why it never gets fixed.

Where the money goes

• Senior time reconstructing the steps - knowledge workers already lose about 1.8 hours a day, nearly a full working day a week, searching for and gathering information¹. A rare process concentrates that waste into an intense reconstruction sprint by your most expensive people.
• Duplicated effort across the org - poor knowledge sharing is estimated to cost large enterprises billions annually, with one widely cited figure putting the loss for a typical large firm in the tens of millions per year⁴.
• Repeated mistakes - without the record of why last year’s shortcut backfired, teams walk into the same trap. In manufacturing, periods of elevated turnover among experienced workers correlate with measurable increases in defect rates even after staffing recovers¹⁸.
• Deadline slippage - only 18 percent of finance teams close their books in three days or less, and half take longer than a week, largely because the process depends on manual, memory-driven steps⁷.
• Consultant and auditor fees - when internal knowledge is missing, companies pay outsiders to rediscover their own process, then pay again next year because the knowledge was never retained.
• Key-person risk premium - the whole cycle waits on one person’s availability. Illness, holiday, or departure at the wrong moment turns an expensive process into a crisis.

A concrete example

Consider a 300-person manufacturer running its annual budget cycle. The controller who built the model left in spring. In autumn, three people spend two weeks reverse-engineering her spreadsheet, guessing at the assumptions behind formulas nobody documented. They miss a cost driver she always adjusted manually, the first draft is wrong, and the board review slips by ten days. None of this appears as a cost. All of it was avoidable.

The Processes That Live in One Person’s Head

Seasonal process knowledge is not abstract. It attaches to specific, recognisable processes that almost every company runs. Here are the most common, and why each is exposed.

The usual suspects

• Year-end financial close - the annual close carries accruals, adjustments, and reconciliations that the monthly close does not. 94 percent of teams still run close activities in Excel, and half say spreadsheets are a key reason it drags⁷. The person who knows which manual entries to make is running on memory.
• The external audit - preparing for auditors is a rare, high-pressure exercise in knowing exactly which documents they will ask for and where they live. Miss a reconciliation and the finding follows you all year.
• Budget and planning cycle - the annual model is often one controller’s creation, full of undocumented assumptions and manual overrides that nobody else can trace.
• Insurance renewals - renewing policies once a year means re-learning coverage details, claims history, and negotiation points every cycle. A gap or lapse is expensive and only visible after something goes wrong.
• Regulatory and compliance filings - annual filings, permits, and certifications come with specific formats, portals, and deadlines. Miss one and the penalty is real. The knowledge often sits with a single compliance owner.
• Tax filings - annual corporate tax preparation blends internal data gathering with rules that change year to year. The internal side, knowing where every number comes from, is the fragile part.
• ESG and sustainability reporting - a newer annual burden, so the process is immature and almost entirely dependent on whoever assembled it last time.
• Benefits open enrolment - an HR process that runs once a year with its own timeline, vendor coordination, and employee communication that gets rebuilt each cycle.
• Seasonal production or hiring ramp-up - businesses with a peak season re-learn how to scale up staffing, inventory, and logistics every year, often losing the lessons from the last peak.

What they have in common

1. They span multiple systems - a rare process usually pulls from ERP, spreadsheets, email, SharePoint, and a portal or two. The knowledge is partly “where everything is”.
2. They mix rules and judgment - most steps follow rules, but a few require experience: which exception matters, who to escalate to, what changed since last year.
3. They have hard external deadlines - auditors, regulators, insurers, and boards do not move their dates because your expert left.
4. They depend on one or two people - the concentration of knowledge is the defining risk, not the complexity of any single step.

Why Documentation Never Fixes It

The obvious answer to a knowledge gap is “write it down”. It is also the answer that has failed every company that tried it for rare processes. Understanding why static documentation breaks down points to what actually works.

The four ways SOPs fail for seasonal processes

• They go stale between cycles - a document written this year is untouched for twelve months. In that time the ERP gets an update, a portal changes, a tax rule shifts, and a colleague moves on. By the next run the SOP is quietly wrong and nobody trusts it.
• They capture steps, not judgment - a checklist can say “reconcile intercompany balances”. It rarely captures which entity always mismatches and why, or the shortcut that looks right but causes an audit finding. The judgment is the valuable part, and it is the part that does not fit in a numbered list.
• Nobody maintains them - there is no owner for a document describing a once-a-year process. Maintaining it is never this week’s priority, so it is never anyone’s priority. APQC finds most organisations have no formal programme to retain this kind of knowledge¹⁵.
• They are disconnected from the work - the SOP sits in a folder while the actual work happens in ERP, email, and spreadsheets. It never learns from what really occurred, so last year’s hard-won lesson never makes it back into the document.

The lesson is not to stop documenting. It is that a document alone cannot carry seasonal process knowledge across a year of change and turnover. What can is a memory that stays connected to the systems where the work happens.

Capturing the Once-a-Year Playbook in a Company Brain

A Company Brain is a shared company memory that captures how your people, processes, and data actually work, and keeps itself current by staying connected to your live systems. For seasonal processes, it is the difference between a playbook that ages in a drawer and one that is ready every cycle.

What it does differently

• Captures the full playbook, not a summary - the steps, the systems touched, the documents used, the people involved, and the decisions made, recorded as the process runs rather than reconstructed from memory afterwards.
• Holds the reasoning - it records why a step exists, which exceptions came up, and what changed since last time, so the judgment survives alongside the mechanics.
• Stays connected to live sources - email, Teams, SharePoint, CRM, and ERP are treated as living inputs, so the memory reflects the current state of your systems rather than a stale snapshot.
• Survives turnover by design - the knowledge sits with the company, not the individual. When the person who ran last year’s cycle is gone, the playbook is still there.
• Improves every cycle - each run adds to the memory. What went wrong last time becomes a flagged risk this time. The process gets more reliable rather than starting from zero.
• Answers in plain language - anyone can ask how the process ran last year, what the deadlines are, and who signed off, and get a grounded answer instead of hunting through folders.

From memory to institutional asset

The shift is from “the process lives in Anna’s head” to “the process lives in the company, and Anna helped teach it”. Anna is still the expert. But if Anna is on holiday, changes roles, or retires, the next cycle does not collapse. The playbook, the reasoning, and last year’s lessons are all still there, ready to guide whoever runs it next.

AI Employees That Actually Run the Cycle

Capturing the playbook is half the answer. The other half is having something that can act on it next cycle. An AI employee is a digital worker connected to your real systems that can run the routine parts of a seasonal process and guide a person through the judgment-heavy parts.

What an AI employee handles

• Pulls the data - reconciliations from the ERP, balances from finance systems, claims history from the insurer portal, assembled without a person copying between tools.
• Follows last year’s checklist - it works from the captured playbook, running each step in order and flagging where this cycle differs from the last.
• Chases the missing pieces - it emails the supplier who always sends the invoice late, reminds the department that owes a figure, and tracks who has not signed off.
• Assembles the deliverable - the reporting pack, the audit binder, the renewal submission, the filing, drafted from the source data and ready for review.
• Keeps a human in the loop - approvals, exceptions, and judgment calls route to a person. The AI employee does the legwork; the expert makes the decisions.
• Learns from the correction - when a person adjusts something, that adjustment feeds back into the Company Brain, so next cycle the process starts smarter.

A year-end close, run differently

1. Before the close - the AI employee reviews last year’s playbook, checks which systems and rules have changed, and prepares the checklist with the known exceptions already flagged.
2. During the close - it pulls reconciliations, matches intercompany balances, drafts the accruals it handled last year, and lists what needs a human decision.
3. At the deadline - it assembles the reporting pack, chases the two figures still outstanding, and routes the final review to the controller.
4. After the close - it records what was different this year, why, and how it was resolved, so next year’s cycle inherits the lesson instead of relearning it.

The same pattern applies to audits, renewals, filings, and planning cycles. Connect the AI employee to where the work happens, ground it in the captured playbook, and the once-a-year process stops being a memory test.

How to Capture a Seasonal Process (Step by Step)

The single most important principle: capture the process while it runs, not from memory afterwards. Memory-based documentation is where SOPs go to die. Here is a practical approach that works around the annual cadence.

The capture playbook

1. Pick the process with the highest failure cost - not the easiest one. Usually the year-end close, the audit, or a critical filing. High stakes plus concentrated knowledge equals highest return on capture.
2. Identify the carrier before they leave - find the one or two people who actually run it. The urgency is highest when that person is close to retirement or has signalled a move.
3. Capture the next live run - instead of interviewing from memory, record the real cycle: the steps taken, systems touched, decisions made, and exceptions handled, as they happen.
4. Connect the AI employee to the real systems - email, Teams, SharePoint, CRM, and ERP, so the playbook is grounded in the actual sources of truth rather than a separate document.
5. Record the reasoning, not just the steps - for each judgment call, capture why. Which exception, which contact, what changed. This is the part a checklist always loses.
6. Let the AI employee run the routine parts under supervision - during the same cycle, have it handle the rule-based steps with the expert watching, so both the capture and the automation are validated at once.
7. Review and correct together - the expert checks the AI employee’s work, corrects it, and every correction improves the stored playbook.
8. Test with a fresh person next cycle - the real proof is whether someone who has never run the process can complete it with the Company Brain and AI employee guiding them.

How Superkind Fits

Superkind builds a Company Brain and AI employees for mid-sized and larger companies. The approach is process-first, not technology-first: the starting point is always your real workflows and systems, which is exactly what a seasonal process needs.

• Company Brain as company memory - it captures how your people, processes, and data work and keeps itself current, so the once-a-year playbook survives turnover instead of walking out the door.
• Capture during the live cycle - we document the real run, not a memory-based idealisation, so the exceptions and workarounds are caught where they actually appear.
• Connected to your real systems - email, Teams, SharePoint, CRM, and ERP are live sources, so the playbook reflects the current state of your tools, not a stale snapshot.
• AI employees that run the routine parts - reconciliations, chasing documents, assembling the pack or filing, with a human in the loop for approvals and judgment.
• Improves via feedback each cycle - every correction your team makes teaches the system, so the process gets more reliable each time it runs.
• Sits on top of your stack - no rip-and-replace, nothing new for the team to learn. The AI employees work with the tools you already have.
• Outcomes, not licences - engagement is tied to a specific process and a measurable result, not a seat count or a multi-year lock-in.
• Security by design - data stays within your infrastructure, protected by encrypted connections, access controls, and audit logs, built to meet GDPR and industry requirements.

Which Rare Processes to Capture First

You cannot capture everything at once, and you should not try. Prioritise by two factors: the cost of failure and the concentration of knowledge. The processes that score high on both are where you start.

Frequently Asked Questions

Henri Jung

Co-founder of Superkind, where he helps SMEs and enterprises deploy custom AI agents that actually fit how their teams work. Henri is passionate about closing the gap between what AI can do and the value it creates in real companies. He believes the Mittelstand has everything it needs to lead in AI - it just needs the right approach.