September 1, 2019

Lasting impact at scale, the work of Kevin Starr & Mulago


Designing for "lasting change that goes to scale" has been the work of Kevin Starr and the Mulago Foundation over the last 25 years.

This post provides a step-by-step summary of Kevin Starr & the Mulago Foundation’s approach to designing for impact at scale. It is designed for social entrepreneur's, but much of this approach works for standard businesses too.

All content, ideas, and images here are from different talks by Kevin on YouTube. If I messed anything up, let me know. After writing this post, I found much of the same content on the blog of the Mulago website. That is all linked at the end of this piece.

I highly recommend going back and watching Kevin’s original talks on this. They are all linked at the bottom of this post.

This article came about after summarizing the book Lean Impact by Ann Mei Chang, published in 2018. In my opinion, many of the ideas discussed in that book, are actually directly attributable to Kevin Starr. Who has been talking about the ideas for over a decade prior to that books publication.

In particular, he has been discussing how the analog of profit is impact. And the analog of 'return of investment' is the cost of that impact.

“Businesses are ruthless about designing for profit, you have to be the same about designing for impact”

Watch this article as a video


Step 1. Clarify the Mission

Step 2. Big Idea

Step 3. Outcome (Impact)

Step 4. Behaviour Map

Step 5. Intervention

Behaviour change: Emotion, Calculation & Bias, Ability

Step 6. Impact Model

Step 7. Scale / Doer at scale / payer at scale / scalability audit

About Mulago / Further Reading / Videos

Before starting, is this right for you?

A process designed to understand how to develop projects that are designed to scale. Many worthwhile projects are not designed for scale. For many years after reading Small is Beautiful in college, Kevin was interested in creating "small jewels" of projects. However, as he becomes more interested in impact, his approach to scale changed entirely, and he began to see that scale is beautiful.

Not all project should be for scale. But it is important to know ahead of time the type of work you are getting into.

Step 1. Clarify the Mission

The first and most important step is to clarify the mission.

Kevin points out that the average mission statement is rather useless:

Screen Shot 2019-08-18 at 2.06.02 PM.png
View fullsize

Therefore, he proposes an eight-word, or less, mission statement.

It must include:

  • a verb
  • a target population (/setting / species / etc)
  • an outcome (something you can measure)

The key thing is that the mission statement has nothing to do with the idea. It should be what one wants to happen in the world. Not how this will happen, or why it should happen.

The statement should be specific enough to focus the organization, yet ambitious enough to build a vision at scale. It is ok for different organizations to have the same mission statement, this is normal.

A few examples

  • Save coral reefs in Micronesia
  • Save kids lives in Liberia
  • Get African farmers out of extreme poverty
  • Save Blue Wales in Shri Lanker shipping corridor
  • High-quality light to very poor people
  • Get clean water to poor people
  • Improve the health of the poor
  • Make rural moms with kids healthier in Asia
  • Get poor people more money
  • Improve living standards in poor communities
  • Prevent childhood malaria in remote African villages

Tips for clarifying he mission

  • The verb must be assertive and clear. The word “get” can often help with this, e.g. “Get clean water to people” is better than “provide access to clean water”
  • The mission must be the state you leave people in at the end. 'Empowering people' is not an end state. What is the end state? Do people have more money? If so, that is the mission
  • Consider making the mission positive (e.g. create healthy reefs) rather than negative (e.g. end reef destruction).
  • In clarifying the mission, it can be helpful to reflect why you decided to do this thing in the first place? Why did you think this was an important problem to solve? What moved you to change careers?
  • The verb is the least important part of the sentence, the target population and outcome covers most of the idea "healthy reefs in Indonesia.”
  • Don't use any of the words from the Mulago foundation's 'banned word list' that they keep in the office. These are words that are too vague, overused, or don't get you all the way to impact at scale: such as empowerment, sustainability, impactful, access.

Write the mission at the top of the page.

"The mission is your success or failure statement. If you accomplish this you succeed if you don’t you fail....Be ruthless towards trying to do one thing well."

The same 'big idea' may have different potential missions. Find the mission with the biggest impact. For instance, of the potential advantages of cook stoves, improved health has the highest benefit.

Screen Shot 2019-09-01 at 3.47.08 PM.png
View fullsize

Step 2. Big Idea

This is the how. How you will execute the mission.

The big idea needs to be summarized down into an "elegant short idea". If it can't be contained that way, the idea is likely too complicated to scale.

Write the big idea right underneath the mission.

Step 3. Outcome (Impact)

If you could measure only one specific outcome to determine the success of your mission, what would it be?

How much you change this outcome determines your impact.

Your Impact = What happened with you - minus - what would have happened without you.

If the same outcome would have happened without you, then you have not had impact.

"Impact is a change in the state of the world brought about by an intervention. It's the final result of behaviors (outcomes) that are generated by activities (outputs) that are driven by resources (inputs)."

If your organization was increasing access to mobile phones in Africa, you may report success. But in fact you may find your organization’s impact has been no different than areas without your organization.

The impact jackpot: is when you deploy the big idea into a population where the intervention yields the highest impact possible (aka. the delta between the impact with the intervention vs the outcome without the intervention is the largest.

Often interventions targeted at the poor, have the potential for the largest impact. Generally, if the poor can access it, it means the rich will be able to afford it too. Or similarly, if you measure childhood mortality, there is likely a chance that adults and youth are healthier too. Focusing on the impact jackpot likely is a good surrogate for other groups, and lets you refine your approach.

Write the outcome way down at the bottom of the page.

Tips for designing the outcome / impact

  • It must be objectively measurable
  • It must capture fulfillment of your mission
  • You must know how to get real numbers for the outcome
  • If you are having trouble determining the outcome, perhaps either your mission is too broad, or you are not an expert yet in the area

Example of outcomes

M: Save blue Wales in Sri Lanka
O: More blue Wales in Sri Lanka

M: Get farmers out of extreme poverty
O: Farmers have more income

M: Get clean water to poor people.
O: People drinking only clean water. (better than 'access to water')

or perhaps the mission should be: Prevent water-born disease

M: Prevent HIV infection in Brazil
O: Decrease in HIV infection rates

M: Educate slum kids in West Africa
O: Functional literacy and numeracy

M: Get money to poor people.
O: Number of jobs created

M: Save lives of marginalized women
O: Maternal mortality

M: Make rural moms and kids healthier in Asia
O: Reduce maternal and child mortality rates

M: Improve health for the rural poor
O: Better disease outcomes

Step 4. Behaviour Map

Impact (the desired outcome) comes from someone doing something different.

Be very clear: who (exactly) has to do what (exactly) differently for the desired outcome and impact to happen? This is captured in a behaviour map.

Write this as a single sentence. For instance, "Mother's in rural households have to use the cookstove exclusively".

The precision of the behaviour is important. For instance, with cookstoves, it has been discovered you only see a benefit if you use it every time and if it removes over 90% of the pollutants from the air. That is why the above behaviour includes the word "exclusively".

You already have written at the bottom of the page the desired outcome. Now, outline above this the behaviour map of who has to do what differently to go from the current state to that outcome.

Some examples:

View fullsize
View fullsize
Screen Shot 2019-09-01 at 8.26.56 PM.png
View fullsize

Tips for the behaviour map

  • Map only those steps and people that are outside of your control. Meaning - they don't work directly for you - aka you can't fire them.
  • The behaviour map traces things people have to do. Not things that people 'feel' or 'think'. The behaviours happen with people's hands, not between their ears.
  • Be specific with the actor. E.g. "the mother" vs "the family".
  • Be specific in defining the actor's role.
  • The behaviour map has to logically flow together to get from A to Z, it cannot leave out any critical steps, or make leap of faith assumptions of how the steps get to impact/outcome.
  • The goal is to require as few behaviours as possible to get to the outcome.
  • Likely you want no more than 6 steps in total. Otherwise, it gets to complicated and unlikely that you can change behaviour in more than that number of steps.

Step 5. Intervention

The intervention is the specific way the behaviour will be changed from the current state to the desired state. It is the thing you do.

For each behaviour on the map, a corresponding intervention must be mapped. If the behaviours are listed down the left side of the page, map horizontally across each behaviour its intervention.

For each intervention, consider

  • conditions: can it happen?  make it easy
  • incentives: will it happen?  tap into pre-existing demand
  • will it last?

Tips for interventions

  • Map an intervention for each behaviour.
  • It is ok for the same intervention to change several behaviours.
  • Some behaviours may be changed by other organizations. This is ok...long as you can ensure this actually happens.
  • The interventions result in actions (things people do with their hands), not changing the way people think between their ears. (Changing people's feelings may be required along the way, but it is not enough).
  • Continuously improve the interventions, to make them more effective, scalable, and cheaper.

"Good entrepreneurs get excited by the prospect of having to continuously improve the intervention"

Screen Shot 2019-09-01 at 8.21.57 PM.png
View fullsize

Tips for behaviour change

The success of the interventions depends upon how people currently feel about the intervention. How you can make them feel better about it. And how you can make it easier for them.

Apply these tips for behaviour change. The likelihood that someone will complete an action is based on three "dials".

Dial 1: Emotion (heart)

How do I feel about the action? Positive? Negative?  

Generally, aim to develop interventions that will evoke a positive emotion. A negative emotion is hard to overcome. Sometimes, a danger emotion may help alert people to something important.

Drivers of emotion

Try to optimize all four, and avoid creating a negative emotion on one

1. Status: does this increase or decrease my / family / community's status?

2. Fairness: would anyone be perceived as loosing? this is a strong deterrent. This is a large problem with giveaways - its seen as unfair to people in the community. Need to bring up the entire community together.

3. Belonging: does this increase or decrease belonging to a group or community

4. Safety: people are risk-averse, does this increase emotional, physical, cultural safety?

Dial 2: Calculation and Bias (head)

People often make decisions based on biases. Understand what people's biased reasoning maybe how they perceive your intervention.

The intervention needs to have a positive cost-to-benefit ratio when analyzed 'objectively' and through biases that people may have.

Wikipedia lists dozens of cognitive biases. A few examples

Confirmation bias: people may continue to hold an idea if it confirms their pre-conceived ideas

Availability bias: people are unduly influenced by recent information and events

Loss aversion: people avoid loss more than seek gain

Sunk cost fallacy: people will remain with something even if they shouldn't continue with it

Dial 3: Ability (hands)

If Dial 1 and Dial 2 work, then people will want to change their behaviour. But are they able to? How can you make it easier for them to do the right thing than not?

Increase 'Internal ability'

  • break tasks into smaller chunks to make it easier to start
  • make something a habit
  • 'tap the heard': if everyone does something, its easier to get someone else motivated

Increase 'External ability'

  • Make something physically easier to do
  • Make doing the wrong thing, physically harder to accomplish
  • Design the product based on familiar things

For an example of how these Dials of behaviour change can be incorporated, see the Examples document.

Step 6. Impact Model

The model is the systematic process by which you apply the intervention to achieve the outcome.

The model is the machine that replicates the activities outlined in the interventions on the behaviour map.

A model is a: systematic, replicable, set of activities that can be done repeatedly, and iterated on.

It helps to know who your doer and payer at scale will be (discussed in Step 7). This influences the type of model you will build.

The model informs you what type of organization needs to be built. Who to hire.

Strive for narrative attribution

In narrative attribution, the story of is so straightforward how the intervention and model lead to the outcome that there is no other way to explain the intervention. This works best.

For instance, putting this fence up to keep animals from grazing has a profound impact at six months time.

Other times, you end up with matched controls: One Acre Fund for instance, measures baseline farmer productivity before they go into a region, and compares that to their results in other areas, and how the farmers perform after the intervention.

Last, is randomized controlled trials: they work well when there are many variables and low-frequency events.

Preserve a working model

The only way to keep the model intact is to tell funders, "this is what we do". Do not follow the money. If you follow the money, you will change your model, and this will result in solutions no longer designed for scale and impact.

"The best organizations lead, and the money follows."

Step 7. Scale

Scale happens after you have an initial idea, undergone research and development, proven it has impact, and proven it has a replicable model.

This sounds like scale happens at the end. However, one should think about scale from the start. Design from the beginning interventions that are designed to scale. Ideally, it is helpful to have an understanding as early on as possible of who the doer at scale and who the payer at scale will be. You want to design early on for this.

To start thinking about scale, Kevin asks organizations to consider what "1 million looks like". This can be 1 million units of whatever your outcome is measuring - such as children educated or hectares preserved.

Why 1 million? It is just a good number to start with. After thinking about 1 million, consider 10 million, and 1 billion.  

How founders react and respond to thinking about operating at scale, is a good indicator to Kevin on their potential to achieve it. Is it something that excites them?

Step 7a. Doers at scale

Who are the doers at scale? Who is going to do / deliver / replicate the model at scale?

There are only four types of doers at large scale.

1 - You: it is unlikely that your organization alone is able to do all the work at scale. In general, this is rare and only found in highly leveraged interventions, typically in the digital space. The number of these types of interventions is must more rare than people think.

2 - Lots of businesses: a model that enables an autonomous business to adopt and implement it while creating a profit. It has to be a good business; marginal businesses don't scale. This works best when increasing the profit of the business leads to more impact.

For instance, if a company grows Eucalyptus trees as a cash crop on poor people's farms, the more trees they sell the more poor farmers that will receive funds for growing these trees.

3 - Lots of NGOs: if the doer at scale are NGOs, it's important to remember they are notoriously bad at delivering other people's models. The model will, therefore, need to be very simple.

4 - The Government: has the mandate to implement interventions at scale that have a huge benefit across many people. Again, the intervention must be super simple for it to be successful.

Ideally, you must pick only one category of doer. It is hard to design for multiple doers at scale. By knowing the doer as early as possible, you can test if they are actually able to do it.

Step 7b. Payer at scale

Who is the payer at scale? Who is going to pay for the model at scale? Where does the revenue come from?

Again, there are only four options.

1 - Customers buying it: again, if this model is possible to build, it is great as it aligns the doer and payer

2 - Government taxes: its important to keep in mind though, that many poor countries don't have enough taxes to afford interventions and most basic services in poor countries are provided by big aid

3 - Big aid: such as USAID, DFID, World Bank. These organizations generally avoid paying for anything until it is ready to scale

4 - Private philanthropy: the advantage is this funding is more flexible and risk-tolerant. However, it often can't take an idea beyond a certain level of growth.

The doer and payer at scale can be different. For instance, for a community health worker model, the doer at scale maybe the government, and the payer at scale may be big aid.

Step 7c. Scalability Audit

The ability to scale requires three things

1 - Is it broad enough? are there enough settings the model can be replicated

2 - Is it simple enough? Often it won't work at scale if the model isn't simple enough. This is frequently seen when scaling a model and the secondary groups implementing it fails to achieve the same impact and outcome numbers as the original group.

3 - Is it cheap enough? (or profitably enough if a business) You must know the payer's price point. How much is that payer currently, and willing to, spend on this type of intervention? Look to see what the government is currently spending on comparable services, and see if you can beat that price point by several-fold.

About the Mulago Foundation

The Mulago Foundation was formed to continue the work of Dr Rainer Arnhold, a pediatrician in San Fransisco who dedicated his life to working in almost all humanitarian crisis through the 1960s, 70s, and early 80s.

Mulago looks for “the best solutions, to the biggest problems, in the poorest places”.

The foundation only provides unrestricted funding to organizations, with a pathway to scale.  Why unrestricted funding? “If you don’t believe an organization can use funding better than you know, you shouldn’t give them funding.”

They offer two fellow programs. Both programs look for scalable solutions. Rainer Fellows focus on scalable solutions to poverty. Henry Follows: focus on scalable conservation solutions.


The Inyenyeri Clean Cookstove example:

- Example of this DIF exercise, by Fitzroy Academy

- Example of the Behavior Change exercise, by Fitzroy Academy

Further Reading

The Mulago Foundation

** The Design Iteration Format, 2019 Edition from Mulago Foundation **

How Mulago Tracks Progress
Annually evaluate: delivery activities towards the mission; organizational capacity towards the mission; and impact achieved.

How Mulago Measures Impact

And many more articles at

The Seven Commandments of Funding
2015, Stanford Social Innovation Review

Screen Shot 2019-09-01 at 10.49.53 PM.png
View fullsize

Lectures & Talks

** Design for impact at scale, Kevin Starr of Mulago **
50 min (across 8 videos), 2017, by Fitzroy Academy
Excellent series to explain the process

How to change behaviour, with Kevin Starr, Mulago 1-6
41 min (across 6 videos), 2018, by Fitzroy Academy
Goes into depth on behaviour change

** Mulago Foundation Workshop: Unreasonable Institute **
2 hours, 2017, Unreasonable Institute
Watch Kevin work through this exercise with a group and help them to clarify their process.

Design For Impact | SWF 2014
1h10m, 2014, Skoll World Forum

How To Create Lasting Change At Scale With Dr. Kevin Starr Of The Mulago Foundation
37 min, 2017, Aidpreneur

Kevin Starr: Lasting Impact
18 min, Pop Tech 2010

Thinking about happiness
Kevin Starr: Fighting Poverty, Designing for Happiness
1h20m, Skoll 2015 world Forum

Kevin Starr, Mulago Foundation
5 min, 2015, Goodwell Investments BV

3 questions for Mulago Foundation's Kevin Starr
4 min, 2017, Devex

Designing for Impact with Kevin Starr | Startup Accelerator | SU Labs
4 min Oct 28, 2015

Interview: Kevin Starr, Mulago Foundation
6 min, 2012, SOCAP

Subscribe for updates

Get an email update from time to time.