Build a logic model (results chain) from your program document
You upload the document and the components already scattered through the text are put in order as a logic model: objective, action, output, outcome. What is missing is proposed from what it says, for you to validate. Each chain ends in an indicator with its formula, suggesting the data sources you need to calculate it.
- What your document says is distinguished from what was proposed
- Each component with the page it comes from
- Indicators with formulas and data sources
Nobody's authorization is needed: it's your file and your browser.
One chain per objective
A program usually pursues several objectives that are independent yet necessary and interdependent. A housing plan that also creates a registry of the housing deficit isn't using the registry to build houses: it uses it to know where they are needed. And if it also promotes local employment, that is something else again.
Each of those objectives comes back as its own chain of the logic model, with its actions and its indicators. No overall objective is invented to bring them together, because that would tidy up the document at the cost of asserting a subordination the program doesn't have.
Each component comes back marked one of three ways: your document says it, with the page; proposed from your document, with the passage it was deduced from; or you answered it. Where the chain goes from what is written to what is proposed there is a line, and you can see it.
Missing elements are shown
Declared, and nobody is pursuing it
Your document commits to something and puts no activity in place to deliver it. It appears as its own chain, with the objective quoted and everything else empty.
No target is proposed
“Increase by 20%” from what. If your document doesn't give the starting point, the chain stops at the indicator and says what would have to be measured first.
Your document contradicts itself
The contradictions are shown so you can decide.
What is defined elsewhere
If the document says that amounts and deadlines are set in a later agreement, that is not an absence and isn't treated as one.
With a formula, and pointing out which data are needed
An indicator that is only a name can't be calculated. Each one comes back with a simple formula relating variables, with each variable declared separately: whether the data already exists and who produces it, whether the program itself will produce it, or whether it has to be built and with what.
That distinction is the part that is most often missing. A formula usually has one term that is already measured and another that isn't, and knowing which is which is the difference between an indicator you can start reporting next month and one that needs a survey first.
Indicators hang from the expected outcome. Counting how many neighborhoods were covered says the action was carried out, not that anything improved, and confusing the two is what makes a monitoring dashboard useless for making decisions.
Whoever has to give shape to what is already written
Government programs and plans
A program approved by resolution that describes activities, and now you have to report progress against something.
Funded projects
The donor asks for a results framework and what you have is the proposal the project was submitted with.
Diagnostics and evaluations
Reconstructing the intervention logic that opens every evaluation, done in an afternoon and with the citation for each thing.
Before committing to indicators
See what can really be measured with the data there is before signing a dashboard that can't be filled in later.
Three steps, and you decide what stays
- You upload the document or paste its text. A .docx or a PDF with text. If it's a scan, it's best to paste the text: you are told when the file couldn't be read, instead of working on half of it.
- What is already written gets ordered. Each component with its citation. What isn't a rung —principles, targeting criteria, legal grounds— stays in view with the reason it isn't one.
- It is completed down to the indicators. Each chain separately, so they don't get mixed. What is proposed is marked and can be corrected: what you correct becomes yours.
You sign the chain. Nothing that is proposed has authority of its own: it is there so you can discuss it with your team and defend it before whoever asks for it.
Yes, and it's the most common case.
It proposes and marks what is proposed. What it doesn't do is contradict your document.
It asks, but it still proposes. A chain that stops at a question leaves you with nothing. The question says what would change in the chain depending on your answer.
Try it with the program you have at hand
Upload the document and see which logic model comes out, and where your document falls short.