The practice is new. The methodological grounding is not.
Meridian is an early-stage evaluation practice, deliberately built as an independent vehicle, with AI-augmented workflows underneath rather than bolted on. What follows is the professional experience that grounds it: the methods, the contexts, and the operational disciplines that Thomas Crossman has developed across a decade of evaluation, advisory, and analytical work.
Three capability areas anchor the practice. Each is grounded in specific engagements, some independent, some as part of consultancy teams, some in senior roles inside programme structures. What matters here is what the practice brings forward from that experience, not the chronology of the assignments themselves.
Where the brief is to measure what actually changed.
One of Meridian's core capabilities is designing results architectures for programmes operating where standard M&E frameworks don't reach: post-authoritarian security institutions, contested political environments, programmes whose real work is institutional rather than programmatic.
The anchoring experience. Thomas served as independent MREL advisor to an FCDO counter-terrorism programme operating in Somalia, inheriting a results framework from an earlier phase that measured training attendance where it should have been measuring operational change. The task was to rebuild the evidence architecture from the ground up, so the programme could credibly demonstrate the institutional shifts it was actually delivering, not just the workshops it had run.
That work sat at the interface of FCDO accountability requirements and a programme team operating in one of the most constrained security environments in the world. The framework had to be rigorous enough to satisfy CSSF reporting standards and honest enough to capture what couldn't be measured cleanly. Quarterly and annual reporting, programme management alongside senior Embassy staff, and a continuous learning cycle that treated every review as a chance to sharpen the approach.
Where the methods get tested.
Meridian's second capability area is theory-based evaluation applied in genuinely difficult operating environments. The kind of contexts where the standard toolkit either doesn't work or doesn't produce evidence anyone can defend.
Thomas has led and contributed to mid-term reviews, endline evaluations, real-time evaluations, and multi-country programme assessments across humanitarian response, cash programming, food security, livelihoods, governance, child protection, gender, climate-migration, and security sector reform. Delivered through consultancy teams and in independent roles, against donor frameworks including FCDO, AFD, BHA, ECHO, UNTFHS, JICA, KOICA, and GAC, for clients including Mercy Corps, UNICEF, IOM, ACTED, UNMAS, and the British Council.
On the qualitative side: contribution analysis, process tracing, outcome harvesting, thematic coding with inter-coder reliability, Theory of Change reconstruction from programme documentation. On the quantitative side: reproducible Python pipelines, Kobo and ODK CAPI survey processing, indicator calculation against standard frameworks, disaggregated analysis, and quasi-experimental approaches where the data supports them. Applied in access-constrained, remote, and conflict-affected settings, with enumerator training, field supervision, and validation workshops.
Before the evaluations, the pipelines.
Meridian's third capability, and the one that most directly enables the AI-augmented workflows the practice is built around, is data engineering. Reproducible pipelines, statistical modelling, data quality systems, and the analytical infrastructure that lets AI tooling earn its place rather than be bolted on to traditional methods.
Thomas built and led a data analytics function from scratch in the private sector before moving into evaluation. Team building, pipeline architecture, data quality systems, statistical modelling. That technical foundation now underpins the AI-augmented evaluation workflows Meridian delivers. Additional consultancy work has included data systems improvement for organisations responding to gender-based violence, and resource allocation modelling in the education sector.