Public consultations are a cornerstone of democratic policymaking in the UK yet analysing the vast volumes of responses government receives has become a bottleneck in the process. With thousands of consultations each year, decision-makers often struggle to extract meaningful insight in a timely, cost-effective manner. This has recently sparked the debate about the use of AI in analysing open-text responses. Across the public sector new initiatives are emerging, including early-stage tools such as Consult and PlanAI, developed by the UK Government and the University of Liverpool respectively. These efforts signal a broader recognition that consultation analysis is costly and labour-intensive, and that carefully deployed automation can help it keep pace with the demands of modern policymaking.
Among the most common pain points in consultation analysis is the growing risk of missed deadlines and inflated budgets, especially when large volumes of open-text responses need to be processed quickly. These pressures are made worse by the logistical demands of large consultations, which often involve manual data handling and the coordination of decentralised teams working across complex workflows. As a result, keeping everyone aligned becomes a project management challenge, requiring constant oversight just to maintain momentum. Under these conditions, the consistency and quality of labelling can easily drop, leading to uneven or unreliable outputs.
Ultimately, this environment makes it far more likely that important insights remain buried simply because no team has the capacity to review every submission in depth. To tackle those issues, it is becoming increasingly clear that a paradigm shift is necessary. And the timing is right. Consultation work can indeed benefit from a tech solution protecting methodological rigour, ensuring full compliance with statutory requirements, and preserving human oversight, while still enabling faster and more consistent analysis.
At Alma Economics, we have spent years working directly with government departments on complex consultation and evidence-gathering projects. Time and again, we saw the same challenge: highly capable analysts spending significant amounts of time on manual data cleaning, thematic coding, and other repetitive tasks that inevitably slow down the pace of analysis, synthesis and report delivery.
This led us to take action and develop a platform specifically designed to address those issues at their root. After delivering multiple consultations using just spreadsheets, our team started developing tech solutions aiming at creating efficiencies. In this journey, and as our tech synergies were already mature, we received seed-funding from MHCLG and the Prop tech innovation fund to build a digital platform that manages consultations. We invested heavily in new features development ever since. We embedded our know-how around consultation management and analysis into the system, including the use of AI to support the coding of open-text responses.
This is the context in which we developed Cobflow, a secure planning and analysis platform specifically built for public consultation work. Cobflow centralises data management, streamlines data cleaning, and facilitates both qualitative and quantitative analysis workflows. Importantly, the deployment of AI in the platform is focused on qualitative coding and designed around a simple but fundamental principle: Humans remain in control. The analysis starts with extensive reading and response labelling carried out by users as they are solely responsible for creating themes. Cobflow then can leverage the team’s expertise by training AI models to replicate and scale their open-text response labelling.
For organisations requiring alignment with specific government standards, Cobflow integrates the same proven AI models used in tools like Consult AI. However, Cobflow’s capabilities extend far beyond this basic offering and classification. It provides more advanced and secure models that deliver greater accuracy and speed. More fundamentally, Cobflow is a complete operational platform. It incorporates essential project management, workflow automation, and reporting tools, all designed by analysts, for the specific challenges of consultation work. This integrated approach streamlines everything from data cleaning to cross-departmental collaboration and final publishing, driving significant time savings, richer insights, and a greener bottom line.
The promise of AI in public consultations
Cobflow’s AI is optional and tightly controlled by humans. Users start the analysis by labelling responses themselves; once they have defined the key themes, they can trigger the AI to finish the job using their exact rules. It won't go 'off-script' or create new categories. Every suggestion the AI makes can be checked and corrected by a human, keeping users in total command of the final report.
Equally vital is the platform’s approach to compliance, privacy and ethics. All personal data is encrypted end-to-end and stored in secure cloud-based databases in the UK. Cobflow’s AI models are deliberately restricted from accessing identifiable information and raw data including respondent characteristics or demographics. They can only access the open-text responses that users choose to share, and only after those responses have gone through Cobflow’s data-cleaning tool to remove personal information, profanity, duplicates, and any campaign-style submissions.
All the above constitute Cobflow’s core guardrails and safeguards. We built Cobflow on strict data-handling and AI-governance to meet statutory requirements and public sector standards. Data is anonymised before any processing, AI is used only for a narrow, human-defined task, and every output is checked by human users. It runs on secure, UK-hosted Google Cloud infrastructure aligned with NCSC principles, supported by security best practice, our Cyber Essentials Plus certification and our commitment to GDPR compliance. Together, these measures ensure full transparency, auditability, strong privacy protections and clear human oversight throughout.
Case study
A recent consultation illustrates the potential. In November 2024, the Department for Education (DfE) launched the Curriculum and Assessment Review Call for Evidence, receiving 7,021 respondents and a total of 159,047 responses across 45 qualitative questions. Reading this volume alone would have taken an 8-person team roughly 2 months, with full thematic analysis requiring significantly longer. Also, keeping a large team perfectly aligned on how to categorise responses is a massive logistical challenge, and human fatigue naturally leads to inconsistencies over time.
Faced with tight timelines, DfE commissioned Alma Economics to conduct the analysis. Using Cobflow, expert researchers carried out extensive initial reading and built a robust coding structure. The platform then used these human-defined themes to automate large portions of the labelling. This hybrid approach allowed the team to complete the qualitative coding around 85% faster than manual methods, without sacrificing quality and transparency. The result was a more accurate, policy-ready report delivered within a timeframe that would not have otherwise been achievable at a fraction of the cost.
Conclusion
Public bodies are under pressure to deliver more with less, and Cobflow directly tackles that challenge. It cuts the costs of consultation analysis, accelerates turnaround times, and strengthens the quality of insights, freeing budget for other priority policy areas and delivering better value for taxpayers.
By making high-quality analysis faster and cheaper, Cobflow gives teams the capacity to run more consultations, handle larger response volumes, and do so far more often. This creates opportunities for a more inclusive and representative policymaking process, allowing a wider range of perspectives to be considered without increasing the burden on teams or budgets.
The opportunity now is simply to use these tools strategically. With a humans-in-control design already built in, platforms like Cobflow let government modernise consultation analysis responsibly, expanding capability while keeping humans firmly in control.
Interested in learning more? Contact us for a free demo.