Finance operations that run on spreadsheets
Map invoice intake, approvals, reconciliation and month-end close to see where hours, rework and late payments actually come from.
Learn: how many hours a month go to re-keying and chasing approvals?
Services / Automation
Process audit & automation roadmap.Know what to automate before you build.
We study how your work actually runs, measure where time, errors and delays come from, and score every automation and AI opportunity on value, feasibility and risk. You get a staged, costed roadmap your team can act on, with or without us.
Evidence from real work. Honest scoring. A roadmap you own.
The business case
Most automation fails because it automated the wrong thing.
Teams buy tools or start AI pilots before anyone has measured the process. The result is automation around a workaround, or a demo that never meets real exceptions. An audit first means the money goes where the evidence says it pays back.
When this service fits
Teams buried in manual work, a long list of AI ideas with no clear priority, leadership asking for an automation plan, or past automation projects that did not deliver.
When something else fits better
If you already know the exact workflow to automate and have the data and access ready, go straight to a build. We will tell you if an audit would only repeat what you know.
Where this applies
Different departments. The same method.
These are example situations, not pre-packaged promises or claims about past results.
Sales and CRM handoffs that leak
Trace a lead from enquiry to signed contract to find duplicate entry, missed follow-ups and data that never reaches the CRM.
Learn: how long does a qualified lead wait for first contact?
Customer support with repeat work
Classify tickets by type, effort and resolution path to separate what can be automated, what needs better tooling and what needs people.
Learn: what share of tickets follow a known resolution?
HR and onboarding paperwork
Follow hiring, onboarding and offboarding across email, forms and systems to find the steps that delay start dates or leave access open.
Learn: how many days between signed offer and a working laptop?
Operations across disconnected systems
Find the people acting as integration between ERP, inventory, logistics and reporting tools, and what it costs when they are away.
Learn: which processes stop when one person is on leave?
Leadership asking where to start with AI
Turn a long list of AI ideas into a ranked, costed set of opportunities backed by process evidence rather than vendor demos.
Learn: which three opportunities pay back first, and why?
How the audit works
Review every step. Move only the strong ones forward.
Each process step is reviewed against the same criteria. The steps that score well on value, feasibility and risk move onto a staged roadmap; the rest are simplified, kept with people or parked with a reason.
What we always include
Real volumes where data exists, the exceptions that break the happy path, a named owner for every process and the human checkpoint each automation needs.
What we keep out
Vendor-driven recommendations, savings figures without assumptions, AI for steps a simple rule would handle, and slide decks nobody can execute.
Discovery
Evidence from the work, not the org chart.
Process documents describe how work should happen. We look at how it does happen, with the people who do it.
Interviews and shadowing
Short structured sessions with the people doing the work, plus observation of real cases end to end, including the exceptions.
System and data review
Read-only review of the tools, exports, inboxes and spreadsheets each process touches, with your approval on what we access.
Volume and effort sampling
Counts, cycle times and handling effort taken from real records where they exist, and labelled as estimates where they do not.
Exception and failure mapping
The cases that break the happy path, how often they occur and who resolves them. These usually decide what automation is safe.
Assessment
Every step scored on the same terms.
Each candidate is scored on value, feasibility and risk so the comparison is fair and the reasoning is visible.
Value
Hours recovered, errors avoided, faster cycle time or revenue protected, expressed as ranges with stated assumptions.
Feasibility
Data availability, system access, rule clarity and how much variation the step has. Unstructured inputs are not ruled out, but costed honestly.
Risk and control
Financial, legal, privacy and customer impact if the automation is wrong, and the human checkpoint each step needs.
Change effort
Who has to work differently, the training involved and the dependencies on other teams or vendors.
Choosing the right tool
Not everything should be AI. Some things should not be automated.
The roadmap recommends the simplest approach that reliably works for each step.
Fix the process first
Remove duplicate approvals, unclear ownership or unnecessary steps before automating them. Automating waste makes it faster waste.
Rules and integrations
Deterministic workflows, system integrations and existing platform features where inputs are structured and rules are stable.
AI where judgement is narrow
Document extraction, classification, drafting and agents with scoped permissions where inputs vary but outcomes can be checked.
People where it matters
Steps with high consequence, low volume or real judgement stay with people, supported by better tools and information.
The roadmap
Sequenced so early wins fund the next stage.
The output is a delivery plan your team can act on, with or without us.
Quick wins
Low-risk changes that can ship within weeks, often configuration or small integrations, to build confidence and free capacity.
Core automation
The higher-value workflows with their dependencies, data preparation, controls and measurement plans spelled out.
Platform foundations
Shared pieces such as integrations, identity, logging and review queues that several automations depend on.
Measurement and review
A baseline for each process and the measures that decide whether an automation is kept, adjusted or switched off.
Tools & methods
Practical methods. Tools matched to your estate.
Discovery relies on people and records more than software. These are the methods and platforms we typically assess, not a list you must adopt.
Process discovery
Structured interviewsShadowingBPMN 2.0MiroLucidchart
Current-state maps built with the team, in notation your analysts can maintain.
Process and data evidence
SQLPythonpandasSystem exportsEvent logs
Volumes and cycle times sampled from real records; process mining tools such as Celonis only where event data supports it.
Workflow automation
n8nMakeZapierPower AutomateTemporal
Matched to complexity, hosting needs and what your team already runs.
AI capabilities
OpenAIAnthropic ClaudeGoogle GeminiDocument AISelf-hosted models
Assessed per step for accuracy, cost and data handling requirements.
Business systems
ERPCRMHRISHelpdeskAccounting platforms
Integration feasibility checked against real APIs, permissions and rate limits.
Planning and reporting
Opportunity scoring modelBusiness case workbookRoadmapNotion or Confluence
Delivered in editable formats your team owns.
Delivery approach
Two to four weeks from scope to roadmap.
Engineers and analysts work alongside your team. Findings are shared as they emerge, so there are no surprises in the final readout.
- 01
Scope and priorities
Agree the processes in scope, the business goals, the people to involve and the systems we may review.
Output: Scope note, interview plan and access list
- 02
Discovery
Interviews, shadowing and record sampling across the selected processes, including exceptions and workarounds.
Output: Current-state process maps with volumes and pain points
- 03
Opportunity assessment
Score every candidate on value, feasibility, risk and change effort, and choose the right approach for each step.
Output: Scored opportunity register with assumptions
- 04
Roadmap and business case
Sequence the opportunities into stages with dependencies, effort ranges, controls and measures.
Output: Automation roadmap and business case
- 05
Validation and handover
Review with stakeholders, test the top assumptions with a small proof where useful, and hand over everything.
Output: Final readout, editable files and a recommended first sprint
What your team brings
A sponsor who can make priority calls, time with the people who do the work, approved access to the relevant records and a finance contact to challenge the business case.
Decisions without surprises
Scope, access and duration are agreed in writing before discovery. Every score in the register shows its reasoning, so you can disagree with it and change it.
What you receive
Editable files. Reasoning you can audit.
Everything is delivered in formats your team can update as the business changes.
Current-state process maps
Each process as it really runs, with owners, systems, volumes, handoffs and exceptions.
Opportunity register
Every candidate scored and explained, including the ones we recommend not automating.
Automation roadmap
Staged plan with dependencies, effort ranges, controls and the measure that proves each stage.
Business case
Value and cost ranges with explicit assumptions your finance team can challenge and update.
The audit does not include building the automations. The final readout includes a recommended first sprint if you want to continue.
Related services
From roadmap to running automation.
The roadmap usually leads into one or more of these.
Questions before starting
The practical details.
How long does an audit take?
Typically two to four weeks for three to five processes, depending on how many teams and systems are involved. We agree the scope and duration before starting.
Do you need access to our systems?
Read-only access or exports for the processes in scope usually help, but nothing is accessed without your approval. Where access is not possible we work from walkthroughs and samples and label findings accordingly.
Will you guarantee savings?
No. We give value ranges with the assumptions behind them, and we recommend measuring a baseline so real results can be compared after delivery.
Does the roadmap commit us to building with you?
No. The roadmap is yours and is written so your own team or another partner can deliver it.
Do you recommend specific vendors?
We recommend approaches and name tools as options. We have no reseller arrangements, and we prefer platforms you already own where they fit.
What if automation is the wrong answer?
Then the roadmap says so. Some steps are better fixed by simplifying the process, changing ownership or keeping people in charge with better information.
How do you handle sensitive data during discovery?
We minimise what we collect, prefer anonymised samples, follow your access rules and delete working copies at the end of the engagement unless agreed otherwise.
Can the audit lead straight into delivery?
Yes. The final readout includes a recommended first sprint, which we can start immediately if you choose to continue with us.
Start with the evidence
Tell us where the hours go.
We will show you what is worth automating.
Share the processes that frustrate your team. We will propose a scope, the people to involve and a timeline before any commitment.