Close cycle cut from 18 to 6 working days at a PE-backed SaaS

Why the close had drifted

The business had grown from £6m to £22m ARR in two years. Headcount in finance had not grown with it, and the chart of accounts still reflected the founder-era bookkeeping setup. Most month-end activity was sequential rather than parallel — a single person owned the bank reconciliations, a single person owned revenue recognition, and any holiday or sickness pushed the entire timetable to the right.

Layered onto this, the lender had recently tightened covenant reporting from quarterly to monthly. Two consecutive months of late submissions had triggered a strongly-worded letter from the bank. The board had asked for confidence that the next refinancing window would not be derailed by avoidable reporting failures.

When we walked in, the controller had been in seat for six weeks. She was strong technically but had inherited a culture where 'finished by month-end' meant 'a draft exists somewhere in OneDrive'. There was no shared definition of done.

The first two weeks: visibility before change

We resisted the temptation to redesign anything in the first fortnight. Instead, we mapped every line on the trial balance to three things: the upstream dependency, the named owner, and the evidence required to sign it off. The map was deliberately produced in a single A3 page and pinned to the wall of the finance area.

That artefact alone changed the conversation. It surfaced six line items with no clear owner, four reconciliations that depended on data the operations team was not aware they needed to produce on a schedule, and two manual journals that had been carried forward unchanged for fourteen months without anyone reviewing the underlying assumption.

Crucially, we did not change a single process during this phase. We just made the existing process visible. The team could see for the first time why the close kept slipping — and they began to volunteer fixes before we proposed them.

Redesigning around dependencies, not departments

From week three onwards, the close calendar was rebuilt around the dependency chain rather than the org chart. Tasks that could be parallelised were lifted out of the critical path. Tasks that genuinely had to be sequential were given hard cut-off times with named escalation routes if upstream data was late.

We standardised journal evidence using a one-page template: source, calculation, reviewer, materiality threshold. Anything below the threshold no longer required the senior reviewer's time. This single change freed roughly two days of controller capacity per close — capacity that was redirected into variance commentary for the board pack.

Reviews moved to a daily 15-minute stand-up during close week. The agenda was rigid: what closed yesterday, what is closing today, what is blocked. No status theatre, no detailed discussion of accounting treatment. Treatment questions were taken offline and resolved within a defined SLA.

Rebuilding the management pack

The legacy pack was forty-two slides and arrived a fortnight after month-end. Nobody read past slide eight. We agreed with the board what they actually needed to make decisions: revenue performance against cohort, gross margin movement with a small number of named drivers, cash position with a thirteen-week forward view, and a single page on operational KPIs that the executive team owned.

The redesigned pack was sixteen pages including appendices. It was structured so that the first four pages would answer the questions a non-executive director would ask in the first ten minutes of the meeting. Variance commentary connected each material movement to a named operational cause — not 'higher than expected costs' but 'three engineering hires brought forward by six weeks to support the enterprise pipeline'.

By month three, the pack was landing two working days after close. By month four, the chair commented in a board meeting that he had stopped writing his usual list of pre-meeting questions because the pack was already answering them.

Controls hardening for the refinancing diligence

In parallel with the close work, we ran a controls remediation track aimed squarely at the refinancing diligence. The lender's diligence provider had a known appetite for documented evidence of segregation of duties, approval thresholds and journal review trails.

We produced a controls matrix mapping every key process to its owner, its evidence location and its review cadence. Where the existing process did not meet the threshold, we either redesigned it or documented a compensating control with a named end date for full remediation. Nothing was left ambiguous.

The controls matrix became a live document maintained by the controller. The handover at the end of the engagement included a quarterly review checkpoint to ensure it remained accurate as the business changed.

What the numbers say four months in

Close cycle reduced from 18 working days to 6, a 66 per cent improvement. The reduction held across three consecutive closes, which is the threshold we use internally before claiming a process change has stuck.

Covenant reporting was submitted on time for every month of the engagement. The lender removed the heightened reporting requirement at the next review. Year-end audit completed with zero adjustments, the first time in three years.

More important than any single metric was the change in tone. Finance team meetings stopped being apologetic. The controller was promoted to head of finance the month after the engagement closed, and the permanent CFO returned to a function that was demonstrably running itself.

What we would do differently

We underinvested in the operational data feeds in the first month. The dependency map surfaced four reconciliations that needed upstream data from operations, but we only formalised the data SLAs in month two. With hindsight, that conversation should have happened in week one, because operational teams need lead time to change their own cadence.

We also waited too long to involve the auditor in the controls remediation work. Bringing the engagement partner into a thirty-minute working session at the end of month one would have shaved a week off the year-end fieldwork.