Skip to content
CodelessOps

Case study · Exscientia · Clinical-stage biotech · 2022 to 2025

Month-end close, from 12 working days to under 8 hours

Before

12 working days

After

Under 8 hours

The run, step by step

Run record · Exscientia · month-end closeComplete · 5 of 5 · 1 sign-off
  1. 01Period closes in the ledgerNetSuitelogged
  2. 02Actuals land overnightscheduled load into Anaplanin the model by morning
  3. 03Headcount and warehouse data landHR and warehouse, same windowin the model by morning
  4. 04Consolidation runsinside the model, no workbook layerevery level at once
  5. 05Review and adjustmentsPersonfinance team, in the modelwith a personsigned offwaiting on the finance teamadjustments and reporting in the same place

Steps as built, 2022 to 2025. No AI in this chain.

01 · The situation

An AIM and NASDAQ-listed, clinical-stage biotech. I built their planning platform in Anaplan from scratch, and the close rebuild came with it. Most of the 12 days was numbers in transit: exports requested, tidied and loaded, headcount on its own loop, consolidation in workbooks waiting for everything else.

02 · What changed

  • The data stopped travelling by hand. NetSuite actuals, HR headcount and warehouse data landed in the model as scheduled overnight loads, so the numbers were there every morning of the close.
  • Consolidation moved inside the model. The workbook layer between the ledger and the reporting was retired.
  • The business went in as hierarchies: expense lines into phases, projects, partners and portfolios, so every number answered at every level at once.
  • The team never left the model: review, adjustments and reporting all in one place, with proper roles and access.
  • Training and handover, so the finance team ran the close themselves.

03 · The result

  • The mechanical close, from ledger close to consolidated actuals ready for review, went from 12 working days to under 8 hours.
  • Reforecasting went from a week to a day in the same programme.
  • Handed to the finance team to run.

Limits
Close here means the mechanical close. The judgement work, reviewing the result and explaining it, is what the recovered days were for. There was no AI anywhere in this chain. The result came from a full planning-platform build between 2022 and 2025. A first build applies the same approach to one scoped bottleneck, so it moves one part of the close, not all of it.

The full write-up: what actually moved the close

04 · Your version of this

If your close loses days to data moving by hand, send me one situation and I'll reply within two working days.