EstimandLab

Every decision has an estimand.

EstimandLab is a Vancouver consultancy for economics, evaluation and AI. We measure what programs, policies and investments change, and model what they would change before you commit.

Frame your question
Illustrative program effect An outcome rises slowly over twelve periods without the program. After the program starts in period six, the observed outcome rises faster, ending about 11 points above the estimated counterfactual. Program starts Before After Estimated effect With program Without
Illustrative example. The question is never just "did the outcome rise?" but "how far above what would have happened anyway, and how sure are we?"

Three practices, one discipline

State precisely what needs to be estimated, choose a method that can estimate it, and report the answer with its uncertainty. That holds whether the question is about a health program, a grain market or an AI workflow.

A researcher pipetting samples in a laboratory safety cabinet

Evidence and evaluation

Did it work, for whom, and was it worth it? Causal evaluation, health economics and study design that hold up under review.

A grain elevator on the Saskatchewan prairie at dusk

Economics and strategy

Where to invest, what a policy would do, how a market is moving. Strategy and economic analysis grounded in data.

Cabled servers in a data centre rack

Data, AI and automation

AI that does real analytical work, built and tested the way we test any estimate: against what it is supposed to deliver.

Turn your decision into a question

Every good analysis starts with an estimand: the effect of what, on what, for whom, compared with what, over what period. Choose an example or write your own, then send it to us.

What is the effect of a monthly basic income payment on household spending on healthy food among low-income households, compared with no payment, over two years?

Send this question to us

Opens your email with the question filled in. We reply within one business day with how we would answer it.

Insights

All insights

AI and agentic workflows · October 26, 2026

Agentic AI for research and reporting: where it helps, and where it quietly fails

AI agents can now search, read, calculate and draft. The risk is not obvious nonsense; it is a polished report that leaves out the one comparison that mattered.

Program evaluation

What is an estimand, and why does your program evaluation need one?

Agriculture

How should a producer organization judge the return on its research?

Policy simulation

Testing a policy before it exists: what agent-based models add

How an engagement works

  1. A 30-minute call

    About the decision you face, the data you hold and your timeline. No charge.

  2. Scope and fixed fee

    Within two business days: the question, the method, the deliverables and the price, in writing.

  3. The work

    With check-ins at agreed points, so there are no surprises at the end.

  4. A clear handover

    A plain-language report, a technical appendix, and everything needed to reproduce the results.

Who we work with

Governments and agencies, research funders and teams, producer organizations, companies and non-profits across Canada.

The British Columbia Parliament Buildings in Victoria

Public sector and policy

Ministries, agencies, municipalities and funders who need to know whether programs work and what proposed policies would do.

Agriculture and food

Commodity commissions, grower-funded research bodies and agri-food associations, from research investment to market reports.

Health, business and AI

Research teams, life-sciences firms, insurers and companies: health economics, strategy, analytics and AI workflows.

More on the sectors we serve

Have a decision that needs evidence?

Tell us what you are deciding, what data you have and when you need an answer. We reply within one business day.

Contact us