Inputs vs. Outputs vs. Outcomes in Impact Monitoring and Evaluation

Iulian Circo
May 20, 2019 · 6 min read

Understanding the difference defines how we think about impact

Say you want to throw a party. You’ll probably need some food. A 6-meal white cloth affair or maybe a laid back barbeque? But what if the weather is bad? You’ll need a plan B. And you’ll definitely need some beverages. A good mix. Something classy, perhaps? But also a few beers. Should you cook yourself? You’ll need the ingredients. And the knowledge. Maybe just call a caterer?

You end up organizing everything, and the party is a success. Or, at least you think it was a success. Is there a way to know for sure?

If it was a success, why was it a success? Was it the host’s charm? Or perhaps that chicken casserole? The music helped, too. And the wine. And isn’t so-and-so such an amazing story-teller? How good that he came.

Outcomes

Turns out a good party is a pretty complex system. There are many elements that you can control and some that you can’t. A good party is an outcome. It is the compound result of our hard work, many small decisions, the weather, some luck and, maybe, just a bit of magic. Outcomes are where it’s at. They are existential — the reason behind everything we do. If you are in the business of organizing parties, if you fail at organizing parties, you are out of business.

Outputs

But outcomes are complex things, that take time and many variables to achieve. We need some sort of system that help us prioritize what we are doing on a daily basis. To separate what absolutely needs to be done (we need guests!) from the nice to haves (balloons!). What are the must-haves of a good party? Venue? Guests that actually show up? Awesome food, presented wonderfully? A spectacular cake? Bubbly at the right time? These would be outputs. Outputs correlate strongly with our outcome.

And while they don’t necessarily guarantee the outcome, failing to achieve the outputs strongly reduces the chances that the outcome will be achieved. Can you have an awesome party if your guests don’t show up? Or if there is no food & nothing to drink? Maybe, but that’s not the sort of party we are in the business of organizing. Outputs are easy to verify, standardize & tick off a checklist. They are great operational proxies for the success of the outcome. If we are data-driven party organizers, it is the outputs that we are obsessing over.

Inputs

But how do we obtain these outputs? There is more than one way to get guests to show up at a party, right? You can be a very popular person. Or you can lure the guests with a game. Or you can outright pay for them to be there. The cake? You can make it with flour, butter and the rest, or you can order it at the baker’s. All of these are possible inputs. It’s the stuff you put into the system that will convert to outputs and eventually to outcomes:

Impact: Inputs vs. outputs vs. outcomes

As a complex system, delivering impact is not so different. The purpose is to achieve a higher level result — say, a reduction in child mortality. Or a cleaner ocean. More resilient communities. This is what it’s all about. These are our impact outcomes.

RELATED: You Should Stop Traditional Data Collection Right Now and Smarten up your Monitoring & Evaluation

But getting there takes time and effort and even when we get there, it is not easy to know that we did. It’d be awesome to have some easier ways to track our progress as we go. And as it happens, we do: Impact Outputs.

If you want to reduce child mortality, vaccinating a child is a good thing to do. Cleaning the environment works that way too: taking plastic out of the environment correlates with a cleaner environment. Sure, the child may get sick of another disease, or get hit by a car. Someone may pump sewage in the river where we just took the plastic out, but at the level of a community, at the higher level, the more children are vaccinated, the lower the child mortality. The more garbage we take out of the river, the cleaner the river.

But how do we achieve these outputs? Maybe we need to train the nurses? Procure vaccination? Hire people to pick up plastic? Or maybe invest in lobbying to influence policy & ensure people who pollute must pay a fine? Regulate sewage disposal?

All of these are Impact inputs. They are the resources that we commit towards achieving the desired outputs. There is more than one way to achieve any given output and whichever way we chose requires specific inputs. But inputs by themselves mean nothing.

Here is the same framework for an impact implementation example, in the area of child survival:

Outputs, baby

If I am in the business of organizing parties my definition of success is organizing parties that people love. I need a simple system to track the elements that I can control leading to the awesome party.

The success of my business depends on my team creating the highest quality outputs. I need to make sure my team finds the best guests, the most spectacular cakes and so on. The smart thing to do would be to define their success around those outputs.

RELATED: Output Markets

It’d be a mistake to evaluate the quality of my team based on how many KGs of flour they used, how many invitations they sent out or how much money they have spent (inputs). If I’d do that I’d be out of business pretty soon.

The fallacy of measuring inputs

Yet, in the impact delivery business, this is exactly what we do. Donors fund implementers and they judge the quality of the implementation based mostly on inputs:

“This is a great charity because their overheads are so low”;

“This is a bad charity because so and so director has a high salary”

Donor-imposed evaluation frameworks (“logframes”) contain tens of inputs that recipients of funds have to report on regularly:

  • no. of trainings conducted;
  • no. of people trained,
  • no. of commodities purchased.

They have whole teams and tools dedicated to this — an entire industry. Forms. Reports. Conferences. While they create the illusion of data-driven implementation, they are mostly noise, and are very easy to game. Here are two things that we could do differently & turn the whole thing around:

  1. We shouldn’t even bother with inputs. At all. Leave them at the discretion of the implementing team, who will measure them in relation to outputs (as a way to optimize their own outputs); If someone figures out a way to roll out expensive technology as a force multiplier and deliver high quality outputs, power to them, high overheads be damned;
  2. Put a fixed, public price on outputs. Then offer the same price to anyone who is willing to deliver. This would mean means that, rather than contracting ONE implementer and hoping they will deliver an outcome, we would pay ANYONE who demonstrates outputs achieved; No contracting, no grant management, no politics. An open market for outputs & true output based financing.

RELATED:

***********

I build businesses that create value for the world.

Connect with me on Linkedin | Twitter

More here

Proof of Impact

Making every data point count