Skip to content
Back to Analytics & Experimentation

Micro-Funnel

A funnel within a single page or interaction —the 3-step micro-journey (e.g., open modal, fill field, submit) that determines whether a macro-funnel step converts.

Why

Macro-funnels (signup to deploy) are too coarse to diagnose a drop-off. If step 3 'create project' has a 35% drop, the fix is somewhere in the create-project interaction —a 5-field form with a modal, a submit button, and validation. The micro-funnel breaks that interaction into its own 3-5 step funnel: saw the modal, focused field 1, filled field 1, focused field 2, ..., clicked submit. Each micro-step has its own drop-off, and the micro-drop-off is what points to the fix. A 40% drop between 'focused email field' and 'filled email field' means the email validation is too strict; a 30% drop between 'filled all fields' and 'clicked submit' means the submit button is not salient or the user doubts success. Micro-funnels turn a single macro-drop-off number into a diagnostic map of the interaction, and the map is what makes the fix findable.

When

When a macro-funnel step has a high drop-off (> 20%) and the step is a single-page interaction (form, modal, wizard). Instrument the micro-funnel before diagnosing.

How

1. Identify the macro-step with the high drop-off (e.g., "create project" form)
2. Map the interaction as a sequence of micro-events: modal_opened, field_focused(name), field_filled(name), field_blurred(name), submit_clicked, submit_succeeded
3. Instrument each micro-event with a unique event name and properties (field name, time since modal open, validation errors)
4. Define the micro-funnel: modal_opened -> field_focused(first) -> ... -> submit_clicked -> submit_succeeded, with a time window (e.g., 5 min)
5. Compute the micro-funnel for the analysis period, segmenting by field if one field has a high drop
6. For each micro-drop-off > 10%, diagnose: is it validation (check error events), is it confusion (check session replays), is it abandonment (check time-to-abandon)
7. Fix the highest-impact micro-drop, re-measure, and iterate —micro-funnels enable fast iteration because the feedback loop is hours, not weeks
8. Roll up the micro-funnel into the macro-funnel: the macro step 3 conversion = product of micro-step conversions, so a 10% lift in one micro-step lifts the macro by a predictable amount

Example

# Micro-funnel: "Create Project" modal (macro step 3)
Macro drop-off: 35% (6,448 reached step 3, 4,193 completed)

Micro-events instrumented:
  modal_opened
  field_focused(name)      name in {title, template, region, visibility}
  field_filled(name)
  submit_clicked
  submit_succeeded

Micro-funnel (n=6,448 modal opens):
  modal_opened:            6,448  (100%)
  field_focused(title):    6,120  (94.9%)  drop: 5.1%  (some users close modal)
  field_filled(title):     5,876  (91.1%)  drop: 4.0%  (title validation?)
  field_focused(template): 5,640  (87.5%)  drop: 4.0%
  field_filled(template):  5,382  (83.5%)  drop: 4.6%
  ... (region, visibility similar)
  submit_clicked:          4,710  (73.1%)  drop: 10.4% from last field
  submit_succeeded:        4,193  (65.0%)  drop: 11.0% (validation errors)

Diagnosis:
  Biggest micro-drop: submit_clicked -> submit_succeeded (11%)
  -> 517 users clicked submit but got an error
  -> 380 of 517 (73%) hit "title already exists" (duplicate title check)
  -> Fix: add inline duplicate check on field_blur(title), before submit

Expected: recover 60% of 380 = 228 users = 3.5pp macro step lift
  (228 / 6,448 = 3.5pp, which rolls up to 0.9pp overall funnel lift)

Common Problems

  • ⚠Not instrumenting micro-events because 'the form is simple' —a 5-field form has 15+ micro-steps, and the 35% macro drop is distributed across them; without micro-instrumentation, you cannot point to which field or interaction causes the drop.
  • ⚠Instrumenting micro-events but not computing the micro-funnel —the events are logged but never analyzed as a sequence, so the team has the data but not the diagnostic, and the macro drop remains a mystery.
  • ⚠Fixing a micro-drop without re-measuring —the team 'improves the error message' but never checks whether the submit_clicked -> submit_succeeded micro-drop actually improved, so the fix is assumed to work without verification.
  • ⚠Treating micro-funnel and macro-funnel as separate analyses —the micro-funnel rolls up into the macro (macro step conversion = product of micro conversions), and connecting them is what lets you predict the macro impact of a micro fix.

Best Practices

  • ✓Instrument every field focus, fill, and blur as a micro-event —the instrumentation is cheap (one event per interaction) and the diagnostic value is high (you can point to the exact field that causes the drop).
  • ✓Compute the micro-funnel as a sequence, not just event counts —the sequence is what shows '40% drop between focused email and filled email', which is the diagnostic; event counts alone show 'email was focused 5000 times' which is not.
  • ✓Connect the micro-funnel to the macro-funnel: macro step conversion = product of micro conversions —this lets you size the macro impact of a micro fix ('fixing this 11% micro-drop lifts the macro by 0.9pp'), which is how you prioritize.
  • ✓Re-measure the micro-funnel after every fix, within hours —micro-funnels have a fast feedback loop (hours, not weeks), and this is their main advantage over macro-funnels; use it to iterate quickly.