Skip to main content

Evidence-based program management: deciding on what the work shows

Evidence-based program management is a simple idea: every status, forecast and decision about a program should rest on evidence from the work, and that evidence should be visible to whoever relies on it. This guide sets out the principles, the practices and a way to start.

By the Tarivra team · · 7 min read

What it means

In many organizations, program management runs on reports: a status chosen by the program manager, a forecast from the original plan and decisions recorded in meeting notes. Evidence-based program management keeps those things but asks one question of each: what in the work supports this?

The idea borrows from evidence-based practice in medicine and management: decisions are better when they rest on the best available evidence, and when the evidence is open to the people affected by the decision.

Five principles

  • Status follows the work. A reported status is a claim, and claims are checked against the work underneath.
  • Forecasts come from real pace. Dates are forecast from how fast work is actually finishing, as a range, not a single promised day.
  • Evidence is cited. Every statement in a status report or brief points to the records it rests on.
  • Decisions are tracked to done. Each decision has an owner, a due date and a record of whether it was carried out.
  • The same rules for every program. Programs in different tools are measured the same way, so they can be compared fairly.

What it looks like in practice

  • Weekly reviews start from the gap between reported and supported status, not from a round of updates.
  • A program manager reports Amber with the numbers that show why, and nobody treats that as failure.
  • When a date moves, the old date stays on record, and the portfolio shows which programs keep their dates.
  • Leadership decisions are written into the team's own tool as work, so follow-through is visible without asking.
  • Auditors and steering committees can trace any status or decision back to its evidence.

How to start

  • Pick five to ten programs that matter and that leaders already discuss every week.
  • Agree a small set of checks: finish rate, work left against time left, scope added, blocked and overdue work.
  • Run the checks for four weeks beside the normal reports, without judging anyone.
  • Discuss the programs where the reported and supported status differ, and ask what help is needed.
  • Record every decision from those discussions with an owner and a due date, and review them the following week.

Done by hand, this takes a few hours a week. The value usually shows within a month: fewer surprises, shorter reviews and decisions that get carried out.

Pitfalls to avoid

  • Using the evidence to blame. If numbers are used against people, the numbers get gamed.
  • Measuring too much. A few checks everyone understands beat a dashboard nobody reads.
  • Ignoring data quality. Work with no owner, no due date or no updates weakens every measure; fix the basics first.
  • Forgetting the decisions. Evidence that does not lead to a recorded decision changes nothing.

Where Tarivra fits

Tarivra was built to make evidence-based program management the default. It reads the work in Jira, Azure DevOps, ServiceNow, GitHub and GitLab, checks every reported status, forecasts from real pace, cites the evidence in a weekly brief and tracks each decision to done in the team's own tool. We call the category Verified Program Intelligence.

Questions

What is evidence-based program management?

Managing programs so that every status, forecast and decision rests on evidence from the work itself, and that evidence is visible to the people who rely on it.

Does it replace frameworks like PMI or SAFe?

No. It works alongside any framework. It changes what status and decisions are based on, not how the work is planned or run.

Can we do it without a tool?

Yes, for a handful of programs. Across a large portfolio and several tools, doing the checks by hand every week becomes slow, which is where a tool helps.