Anna Totterdell
Projects Director
Most mid-market businesses overestimate their operational maturity. Not because they are dishonest - but because the view from the leadership team is different from the view on the ground.
The executive team sees the tools that have been purchased, the projects that have been funded, and the strategy that has been approved. The operations team sees the workarounds that keep things running, the data that does not match, and the manual steps that nobody has time to fix.
This checklist bridges that gap. Five dimensions, five levels each, scored honestly. The result tells you where your business actually sits - and where to focus next.
How to use this
For each dimension, read the five levels and select the one that most accurately describes your current state. Be honest. Pick the level that matches what happens day to day, not what the org chart says should happen.
Score 1 for Level 1, through to 5 for Level 5. Total your scores across all five dimensions.
Dimension 1 - Data
Level 1 - Fragmented. Data exists in multiple systems with no consistency. The same entity has different names, formats, or definitions in different platforms. Nobody trusts the data without manually checking it first.
Level 2 - Acknowledged. The team knows the data is inconsistent. Someone has started documenting where the problems are. Cleanup efforts are manual and periodic.
Level 3 - Standardised. Master data definitions exist for key entities, usually as the first output of serious data and systems integration. A single source of truth has been established for customers, products, or financial data. Not everything is clean, but the standards are defined.
Level 4 - Connected. Data flows automatically between systems with validation rules. Discrepancies are caught and flagged. Reports can be generated without manual assembly.
Level 5 - Trusted. Leadership makes decisions based on data without questioning its accuracy. Data quality is monitored continuously. New data sources are onboarded against established standards.
Dimension 2 - Process
Level 1 - Tribal. Processes exist in people's heads. If a key person is absent, the work does not get done or gets done differently. Documentation is nonexistent.
Level 2 - Documented. Key processes have been written down, though the documentation may not reflect current practice. There is a gap between the documented process and the actual process.
Level 3 - Mapped. Processes have been formally mapped end to end, including inputs, outputs, handoffs, and exception paths. The team knows where the bottlenecks and manual steps are.
Level 4 - Optimised. Bottlenecks have been addressed. Manual steps that could be automated have been. The process is designed around efficiency, not history.
Level 5 - Continuously improved. Processes are measured, reviewed, and refined on a regular cadence. Changes are tested, measured, and adopted based on evidence.
Dimension 3 - Integration
Level 1 - Isolated. Systems do not communicate. Data moves between them via manual export and import, email, or re-keying.
Level 2 - Ad hoc. Some systems are connected via basic integrations - file transfers, scheduled imports, or simple Zapier-style automations. These connections are fragile and lightly monitored.
Level 3 - Connected. Key systems are integrated via APIs with error handling and logging. The most important data flows are automated and reliable.
Level 4 - Orchestrated. Data flows are managed centrally with monitoring, retry logic, and alerting. The integration layer is treated as infrastructure, not an afterthought.
Level 5 - Event-driven. Systems communicate in real time via events. Changes in one system propagate immediately to dependent systems. The integration architecture is documented, scalable, and maintained.
Dimension 4 - Automation
Level 1 - Manual. Most operational tasks are performed by people. Automation is limited to basic spreadsheet formulas or email rules.
Level 2 - Task-level. Individual tasks have been automated - data entry, report generation, notification sending. These automations are standalone and not connected to each other.
Level 3 - Process-level. Full business automation workflows run end to end - from trigger to outcome. The team manages exceptions; the system handles the routine.
Level 4 - Intelligent. Automation includes decision logic, branching, and AI-assisted steps. The system handles not just the routine but the semi-routine, escalating only genuine exceptions.
Level 5 - Adaptive. Automation learns from outcomes and adjusts. Exception rates decrease over time. New patterns are identified and incorporated without manual reprogramming.
Dimension 5 - AI readiness
Level 1 - Absent. No AI is in use. The team may be using ChatGPT for ad hoc tasks but there is no operational AI.
Level 2 - Experimenting. One or two AI tools are in use - usually for content generation or data summarisation. No connection to core operations.
Level 3 - Applied. AI is used in at least one operational process - document classification, data extraction, or anomaly detection. The impact is measured.
Level 4 - Embedded. AI is part of multiple workflows, working on operational data, producing outputs that feed directly into decisions or downstream processes.
Level 5 - Strategic. AI enablement is planned and prioritised as part of the operational roadmap. New AI applications are evaluated against business impact, not novelty.
Scoring
5-10: Foundation stage. The priority is data and process. Do not invest in automation or AI until the basics are in place.
11-15: Building stage. The foundations are forming. Focus on integration and structured data to unlock the next level.
16-20: Operating stage. The business has solid fundamentals. Automation and applied AI will deliver measurable returns.
21-25: Advanced stage. The business is operationally mature. Investment in adaptive automation and strategic AI will compound existing gains.
What to do with the score
The score is not a judgement. It is a starting point. Most mid-market businesses score between 8 and 14 - which means the highest-value investment is almost always in data quality, process mapping, and system integration. Not AI. Not automation platforms. The fundamentals.
Fix the foundations. The results follow.
