Measurement / SELF-GUIDED

Choose a measurement approach

Match the decision to attribution, an incrementality test or a broader modelling approach.

Start the first check Discuss a difficult question

Use a non-sensitive summary. Notes stay in this tab; Save keeps a temporary copy for reloads in this tab. Nothing is sent until you confirm a consultation request.

Your question

What do you want to understand?Which result matters?

CHECK 1 / 3

Name the decision

Describe the action under consideration and the result it should improve. Distinguish credit reporting, additional outcomes and portfolio allocation.

Check your interpretation. A model choice is useful only if its answer can change an action.

What will you do differently depending on the answer?Where are you with this check?

If you cannot resolve this check

If no action changes, narrow the question before collecting more data.

CHECK 2 / 3

Inventory the comparison you can make

List available attribution records, possible holdouts and historical spend/outcome variation. Mark gaps explicitly.

Check your interpretation. Attribution describes assigned credit. A causal comparison needs a credible counterfactual. MMM needs variation and explicit assumptions.

Which comparison is feasible, and which assumptions remain untested?Where are you with this check?

If you cannot resolve this check

If a control is infeasible, document the identification limits rather than treating another method as automatically causal.

CHECK 3 / 3

Specify the smallest useful measurement

Choose an approach, outcome, observation window and decision threshold. Record what an inconclusive result would mean.

Check your interpretation. Check power or precision, contamination and data maturity before committing to the design.

What result would support action, further measurement or no change?Where are you with this check?

If you cannot resolve this check

Get help when feasible data cannot distinguish the alternatives at a useful level of uncertainty.

What would help you move forward?

You can continue with partial observations. Recording a check does not verify its conclusion.

Work with your data

Carry your question and observations into Project Studio. Refine the scope, then use Builder and CSV Analyzer. Sign-in is required.

Keep the next check

Save your observations here or download them for your own work. No CSV rows are attached.

A QUESTION WORTH DISCUSSING

Bring the part you could not resolve.

Describe the difficult part in your notes above. You can ask for a review before finishing every check.

The next page lets you review a free first-consultation request. After sign-in, your notes fill the request; you can edit them before sending. Scope and timing are agreed after review.