AI Savings Calculator
How the AI Savings Calculator works
Understanding how we estimate your potential savings and capacity uplift
The calculator has two modes
Both are grounded in real evidence on productivity, task automation, and adoption patterns in higher education.
A quick, directional estimate for your whole department.
An optional detailed model that accounts for different academic and professional service roles.
1. The Standard Calculator
A quick, directional estimate for your whole department.
What it shows
Annual savings (£), productivity increase (%), capacity created (FTEs), and five-year projection (£).
How it works
You enter the number of staff, their average annual salary, and an adoption level: just starting out, building momentum, or fully embedded. The calculator adds typical on-costs such as NI and pensions.
Core assumptions
1 FTE = 37.5 hours per week; on-cost factor is 1.25 × salary; AI-eligible tasks are 55% of total workload; productivity uplift is 8-15% faster completion of eligible tasks depending on adoption level.
2. The Advanced Calculator
A more precise model that accounts for different academic and professional service roles.
What it adds
The advanced calculator allows you to enter headcounts for different academic and professional service roles for a more precise result.
AI-eligible task share
This is the proportion of a person's job made up of activities that AI can assist with directly, such as drafting or summarising text, writing communications or reports, preparing teaching materials, reviewing or analysing data and feedback, searching, referencing, collating information, and formatting or structuring content.
What it excludes
It does not include work that requires in-person teaching, decision-making, or practical delivery that AI cannot (yet) perform reliably.
Role categories and data foundations
When you enter counts for these roles, the calculator replaces the single headcount figure and weights savings by role type, producing a blended productivity and capacity estimate. Roles marked "Default share" use the 55% figure until role-specific evidence is added.
| Role type | Example departments or functions | Typical AI-eligible task share | Rationale |
|---|---|---|---|
| Educators | Teaching and learning | 55% | Lesson prep, resource creation, and feedback benefit strongly from AI support. |
| Researchers | Research staff, postdocs | 50% | Literature review, summarising, and drafting papers show measurable gains. |
| Student Services / HR / Registry | Student-facing and administrative services | 55% | Default share. |
| Educator-Researchers | Combined teaching and research roles | 53% | Balanced mix of academic and administrative tasks. |
| Library & Learning Resources | Librarians, learning developers | 50% | Metadata, summarisation, and research assistance tasks are moderately AI-amenable. |
| Finance | Finance and procurement | 55% | Default share. |
| IT Services | IT and digital services | 55% | Default share. |
| Marketing and Comms | Marketing, communications and recruitment | 55% | Default share. |
| Estates / Facilities | Estates, facilities and operations | 55% | Default share. |
| Other Admin | Any other professional services role | 55% | Default share. |
3. Understanding Target Adoption Level
The Target Adoption Level shows how widely and confidently staff are expected to use AI tools in their everyday work. It captures both reach (how many people use AI) and depth (how effectively they use it). High adoption doesn't just mean more licences - it means AI is built into daily workflows.
| Adoption level | Approximate adoption rate | Typical efficiency uplift | Example overall productivity gain |
|---|---|---|---|
| Just starting out | 35% | 8% faster completion | 0.55 × 0.35 × 0.08 = 1.5% overall |
| Building momentum | 60% | 12% faster completion | 0.55 × 0.60 × 0.12 = 4.0% overall |
| Fully embedded | 80% | 15% faster completion | 0.55 × 0.80 × 0.15 = 6.6% overall |
4. KPIs and how the calculator works them out
The calculator produces four key results.
Annual savings = Number of staff × Average loaded salary × Productivity improvement.
Productivity increase = Eligible work × Adoption rate × Efficiency gain.
Capacity created = Number of staff × Productivity increase.
Five-year savings = Annual savings × 5.
Worked example
Using the Building momentum level (12% efficiency on eligible tasks, 60% adoption rate), 55% eligible task share, 60 staff, £45,000 average salary, and an on-cost factor of 1.25:
Productivity improvement = 0.55 × 0.60 × 0.12 = 3.96% (rounded to 4%)
Annual cost savings = 60 × (£45,000 × 1.25) × 0.0396 = £133,650
Capacity created = 60 × 0.0396 = 2.4 FTEs (about 89 hours a week)
Five-year projection = £133,650 × 5 = £668,250
5. Data and Evidence Sources
| Area | Evidence base |
|---|---|
| Productivity uplift (8-15%) | Bick et al. (2025); OECD / Filippucci et al. (2025); Brynjolfsson, Li and Raymond (2025). |
| Adoption levels (35-80%) | BCG (2025); McKinsey (2025). |
| Task eligibility (~50%) | Henseke et al. (2025); McKinsey (2025). |
| Higher education adoption | Microsoft (2025); HEPI / Freeman (2025); UNESCO (2023). |
| Cost avoidance and back-office efficiency | Dell’Acqua et al. (2023, Harvard / Wharton / BCG). |
| Error reduction and quality control | Dell’Acqua et al. (2023); PwC (2024). |
| Salary and working patterns | ONS ASHE (2024). |
References
- Bick, A., Blandin, A. and Deming, D.J. (2025) The Rapid Adoption of Generative AI. Federal Reserve Bank of St. Louis Working Paper 2024-027C (revised February 2025). Available at: stlouisfed.org
- Bick, A., Blandin, A. and Deming, D.J. (2025) The State of Generative AI Adoption in 2025. Federal Reserve Bank of St. Louis, On the Economy, 13 November 2025. Available at: stlouisfed.org
- Boston Consulting Group (2025) AI at Work 2025: Momentum Builds, But Gaps Remain. Third edition, June 2025. Available at: bcg.com
- Brynjolfsson, E., Li, D. and Raymond, L.R. (2025) 'Generative AI at Work', The Quarterly Journal of Economics, 140(2), pp. 889–942. Available at: academic.oup.com
- Dell'Acqua, F., McFowland III, E., Mollick, E.R., Lifshitz-Assaf, H., Kellogg, K., Rajendran, S., Krayer, L., Candelon, F. and Lakhani, K.R. (2023) Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality. Harvard Business School Working Paper No. 24-013. Available at: ssrn.com
- Filippucci, F., Gal, P., Laengle, K. and Schief, M. (2025) Opportunities and Risks of Artificial Intelligence for Productivity. OECD AI Papers. Available at: oecd.org
- Freeman, J. (2025) Student Generative AI Survey 2025. HEPI Policy Note 61. Higher Education Policy Institute, in partnership with Kortext. Available at: hepi.ac.uk
- Henseke, G., Davies, R., Felstead, A. and Zhou, Y. (2025) How Exposed Are UK Jobs to Generative AI? Developing and Applying a Novel Task-Based Index. arXiv preprint 2507.22748. Available at: arxiv.org
- McKinsey & Company (2025) The State of AI in 2025: Agents, Innovation, and Transformation. November 2025. Available at: mckinsey.com
- Microsoft (2025) 2025 AI in Education Report. Available at: microsoft.com
- Office for National Statistics (2024) Employee Earnings in the UK: 2024. Annual Survey of Hours and Earnings statistical bulletin, 29 October 2024. Available at: ons.gov.uk
- PwC (2024) 2024 Global AI Jobs Barometer. PricewaterhouseCoopers. Available at: pwc.com
- UNESCO (2023) Guidance for Generative AI in Education and Research. Authored by Miao, F. and Holmes, W. Paris: UNESCO. Available at: unesco.org
Ready to estimate your potential savings?
Use the calculator, then book a free AI Capability Check to see where the time is really going.