Productivity without proof
Is AI delivering value?
One of the most striking findings from the research is the contrast between the benefits respondents believe their organisations are receiving from AI and their ability to demonstrate those benefits formally.
Across the survey, respondents reported a wide range of positive outcomes. Time savings were the most frequently cited benefit, selected by 45 respondents, 68%. Productivity improvements were identified by 29 respondents, 44%, while 14 respondents reported cost savings and a further 14 reported improvements in data quality. Other reported benefits included better decision-making, better services, improved forecasting, reduced response times and improvements to resident experience.
What benefits have been realised?
Time savings
Productivity improvements (staff)
Cost savings
Improved data quality
Better decision making
No clear benefits
Better services
Reduced response times
Better forecasting
Improved resident experience
Collectively, these findings suggest many organisations believe AI is already creating practical benefits. This is reinforced by responses to broader performance questions. Forty-three respondents, 65%, agreed or strongly agreed that AI had improved productivity, while 41 respondents, 62%, agreed or strongly agreed that AI was delivering measurable value.
Improved productivity Agree or strongly agree
Delivering measurable value Agree or strongly agree
Examples identified elsewhere in the research, including reporting automation, contact centre support, dashboard development, analytical tasks and workflow automation, indicate that organisations are already finding practical ways to use AI to improve efficiency and support decision-making.
However, the picture becomes more complex when organisations are asked whether they can prove the impact AI is having.
Demonstrating return on investment
Only 14 respondents, 21%, agreed or strongly agreed that their organisation could demonstrate return on investment from AI. Nineteen respondents, 29%, disagreed or strongly disagreed, while 33 selected a neutral response. The 62% measurable-value figure therefore reflects respondents’ assessment of value rather than evidence of systematic measurement. Relatively few reported being able to demonstrate that value confidently.
Strongly agree
Agree
Strongly disagree
Disagree
Neutral
This challenge becomes even clearer when considering measurement activity. Forty-six respondents, 70%, reported that they are not currently measuring AI’s impact at all. Only three respondents, 5%, said they measure impact consistently. A further 10 reported measuring impact in some cases, while seven said they were planning to introduce measurement processes in the future.
Lack of structured evaluation
In effect, many respondents appear to be relying on informal indicators of success rather than structured evaluation.
This is understandable. Many of the most common AI applications, such as content generation, research support, meeting summaries, coding assistance and productivity tools, can create benefits that are difficult to quantify. Staff may save time, complete tasks more quickly or improve the quality of outputs without those improvements being formally measured.
The survey responses illustrate this challenge clearly. Time savings were the most commonly reported benefit, with respondents referencing report writing, coding, dashboard development, research and summarisation. The results do not show how widely organisations have established baselines or formal evaluation frameworks, but the low level of measurement indicates that structured evaluation remains uncommon.
The findings suggest that increasing usage alone may not be sufficient. As AI becomes more integrated into business processes, organisations are likely to need stronger approaches to measurement and evaluation. Questions around value, outcomes and return on investment are likely to become increasingly important as adoption matures.
The research therefore suggests social housing is moving beyond the question of whether AI can generate benefits. Increasingly, the challenge is proving those benefits, understanding where value is being created and ensuring future investment is directed towards the use cases that deliver the greatest impact.
Housemark expert insight:
Why measurement matters
One of the clearest lessons emerging from the research is that adoption and value are not the same thing.
Many organisations can point to examples where AI is saving time, improving productivity or supporting better decision-making. These are genuine successes. However, as AI becomes more widely embedded, organisations will increasingly need evidence as well as experience.
Measurement does not need to be complicated. In many cases, it begins with establishing a baseline, identifying a desired outcome and assessing whether AI is helping to achieve it. This might involve measuring time saved, reductions in manual processing, improvements in service performance or changes in customer outcomes.
Housemark believes organisations are more likely to realise long-term value from AI when they understand which use cases are working, which are not and where future investment should be focused. Measurement is not about proving that AI works. It is about identifying where it creates the greatest value and ensuring resources are directed accordingly.
Andrew Jackson,
Director of Consultancy & Partnerships, Housemark