Experiment rapidly.Maximize your impact.

Evidential is a free, open-source experiment engine built by and for nonprofits. Run experiments on your programs, learn what works, and make them better.

app.evidential.dev
Evidential console — experiments listEvidential console — bandit analysis with leaderboard and historical means
Trusted by partners
KabakooUdhyam Learning FoundationMAIANoora Health

Nonprofits learn faster with Evidential

Real experiments our partners have run, and what they found.

74
Experiments run
15
Organisations
“Since adopting Evidential, we've been able to surface clear, actionable insights much faster, and I'd recommend it to any team that needs to test product or engagement ideas quickly without sacrificing rigor.”
Godwin HoudjiQuantitative Researcher, Kabakoo
“Evidential transformed how we approach our digital engagement. […] It takes the complexity out of rigorous experimental design and gives you the exact data you need to drive real-world impact.”
Shashanka KunduDirector of Product, Udhyam Learning Foundation
“We've found Evidential to be a very useful tool in building up the culture and practice of A/B testing. […] I love logging in and checking progress regularly, and that builds up the test & iterate muscle.”
Bailey KlingerFounder & Director, MAIA
IDinsight team reviewing records with health workers at a clinic
IDinsight field team member guiding a respondent through a form
Enumerator training session with a live map of survey locations
A program officer using a phone alongside printed survey materials
Kabakoo team members smiling as a colleague passes them a microphone at a workshop
Group photo of participants at the 2026 Nairobi Sprint hosted by The Agency Fund
A workshop participant speaking into a microphone while the room listens

Built for how non-profits work

Plenty of tools can run an A/B test. Almost none are built to measure real program outcomes across the channels you deliver on.

  • Designed for the social sector. The roadmap is steered by mission-driven orgs, not advertisers.
  • Impact-first metrics. Measures behaviour change and program outcomes, not vanity reach.
  • Made for continuous experimentation. Set it up once, then experiment indefinitely.
  • Your data stays yours. Kept on your own infrastructure.

How Evidential compares

Category (examples)Built forTypically measuresAssumes you have
Product & web experimentation (Optimizely, Statsig)Product & growth teamsClicks, sign-ups, conversionsIn-house engineers
Messaging & chatbot A/B tools (Intercom, WATI)Marketing teamsOpens, clicks, engagementA marketing/ops team
Product analytics (Mixpanel, Amplitude)Product teamsUsage & funnelsA product + eng team
Traditional RCTsOne-off impact studiesCausal impactResearchers, months, funding
Nonprofits improving live programsEngagement, behaviour change & outcomesAn appetite to experiment

What partners have tested

+125%
more user responses

Noora Health used interactive buttons to get more mothers to respond

Added buttons and prompts to WhatsApp health messages for about 2,400 new mothers.

Udhyam Learning Foundation
+69%
lesson-report submissions

Udhyam used WhatsApp reminders to get teachers to submit lesson reports

Sent reminders to about 700 teachers in Maharashtra and compared them with a control group.

Kabakoo
WhatsApp
beat push notifications

Kabakoo tested which channel gets learners to show up

WhatsApp reminders lifted exposure and participation. Sending more reminders didn't deepen engagement.

From idea to insight in five steps

01

Connect your data

Link the channel you deliver on and the data you already collect.

02

Design the experiment

Set your hypothesis, choose what to compare and how to measure it.

03

Assign at random

Target the right-sized audience (for all your variations) with just a few clicks.

04

See results live

Outcomes update as people respond, so you see what works early.

05

Learn & act

Clean data and lessons in one place, ready to act on.

What Evidential can do

Pick the method that fits your question. Evidential handles the set-up and the statistics.

Experiment cards labelled A/B and bandit experiment types
Analysis panel showing a beta prior and a binary reward type
Leaderboard of estimated average outcomes for each arm
Historical means chart with uncertainty ranges over time
Evidential sidebar with the Integrations menu and experiments list

Where are you starting from?

We'll look at your program and data together and set Evidential up with you. How much support you need depends on where you're starting from.

What it costs:Free

You track outcomes in a database or event system and have a question in mind. We connect Evidential to your data and get your first experiment live.

What it costs:Free, with extra setup support

The data exists but isn't connected yet, or you're not sure what to measure. We'll help with the database and integrations, and design your first experiment with you.

What it costs:Paid retainerExperimentation Advisory Service

An IDinsight economist works with your team on a fractional basis. They help you design experiments, run them, and turn the results into decisions you can act on.

Not sure which one fits? That's what the first conversation is for.

Start a conversation with us
A program officer using a phone alongside printed survey materials

Curious whether Evidential fits your program?

Our team will help you find a first experiment worth running. Tell us a little about your work and we'll take it from there.

Free and open source · Hands-on support · Your data stays yours

Built byIDinsight | The Agency Fund

Takes under a minute. We'll reply within a few working days.

Already using Evidential?

Email the team about your experiments or the Experimentation Advisory Service.

support@evidential.dev