Skip to content

Concept Screening Funnel

A staged narrowing process — instantiates Divergence-Convergence Cycle Orchestration

Narrows many candidates to few through a sequence of successively stricter, successively more expensive screens — cheapest filters first — so evaluation budget is spent only on options that have already survived.

When you have eighty candidates and can only afford to deeply evaluate three, the question is not which criteria but in what order you apply them. Concept Screening Funnel is the answer: a staircase of screens arranged cheapest-first, where each stage kills the options that fail a coarse test before any expensive test is run. Early screens are quick knockouts on hard constraints; later screens cost real money and prototyping effort but face far fewer survivors. Its defining idea is sequencing evaluation by cost — it is not a scoring model and not a single gate, but the deliberate ramp of investment that lets a team pay little to reject the many and reserve deep appraisal for the few.

Example

A snack-food company runs an innovation sprint that generates about 80 flavor-and-format concepts. Sending all 80 to consumer taste panels and shelf-life testing would be ruinous, so the concepts enter a funnel. Screen 1 — desk feasibility (an afternoon): can our lines actually produce it and does it fit the price ceiling? Roughly 80 → 30. Screen 2 — internal cutting (a day): the team tastes bench prototypes against a short must-have list; 30 → 10. Screen 3 — small consumer test (a week, real cost): 10 → 4. Screen 4 — shelf-life and unit-cost validation (weeks, expensive): the 4 survivors get the full workup, and 2 go forward.

The point is not that the criteria got harder — it's that the dollars per evaluated concept rose stage by stage while the number of concepts fell, so total spend stayed bounded. A concept killed in Screen 1 cost almost nothing; only genuinely promising ones ever reached the expensive tests.

How it works

What makes it a funnel rather than a checklist is the ordering discipline:

  • Rank screens by cost, not importance. The first screen is whatever cheaply eliminates the most options; the priciest evaluation goes last, when survivors are few.
  • Each screen is a pass/fail gate. Options either clear the stage's bar or leave; there is no carrying a weak option forward "to be safe."
  • Investment escalates with survival. Reaching a later stage earns an option more evaluation budget — commitment deepens only as evidence accumulates.
  • The stage budget is fixed. Each screen has a set effort and a target survivor count, which keeps the whole funnel bounded regardless of how many concepts entered.

Tuning parameters

  • Screen strictness curve — how aggressively each stage cuts. Steep early cuts save the most budget but raise the risk of killing a slow-burn winner on a cheap proxy; gentle cuts are safer and costlier.
  • Number of stages — more screens grade investment more finely but add overhead and handoffs; too few and you are back to one blunt gate.
  • Per-stage budget and survivor target — the dials that bound total cost; tightening them focuses spend but narrows earlier.
  • Kill reversibility — whether an eliminated concept is truly dead or parked for possible recall. Hard kills keep the funnel clean; a recall lane hedges against early false rejection at the cost of clutter.

When it helps, and when it misleads

Its strength is economic: it makes the cost of narrowing scale with the value of the decision rather than the count of options, so a team can start wide without committing to evaluate wide. It also makes the order of reasoning explicit, which is auditable later.

Its characteristic failure is the false negative — a genuinely strong concept eliminated early because the cheap proxy screen happened to misjudge it, a Type II error the funnel structurally invites by design.[1] Cheap early screens are exactly the ones most likely to misfire, and once an option is out it usually stops being thought about. The related misuse is stacking the early screens with criteria that quietly encode the incumbent's shape, so novel candidates die in Screen 1 for looking unfamiliar rather than being weak. The discipline that guards against both is to keep early screens to hard, defensible knockouts only, to log why each option was cut, and to keep a recall lane for options killed on thin evidence.

How it implements the components

The funnel owns the narrowing-and-investment machinery, not the generation or the fine scoring:

  • convergence_criterion_set — it holds the criteria as an ordered sequence of screens, each with its own bar, rather than one flat scorecard.
  • phase_transition_gate — every screen is a pass/fail gate that an option must clear to advance; the funnel is a chain of such gates.
  • reversible_commitment_ladder — surviving a stage earns an option more evaluation investment, so commitment depth rises step-wise with evidence.
  • cycle_cadence_and_budget — fixed per-stage effort and survivor targets bound the total cost of the whole narrowing pass.

It does not generate the candidates it filters (candidate_option_inventory, exploration_dimension_map — see Assumption Reversal and Recombination and Morphological Matrix, then Scoring), nor does it produce the weighted trade-off scoring of a final option_comparison_frame — that belongs to Multi-Criteria Decision Matrix.

References

[1] A false negative (Type II error) — rejecting an option that would in fact have succeeded — is the error a cost-ordered funnel systematically risks, because it decides most eliminations on the cheapest, least-informative evidence. Naming it is what motivates the recall lane and the "hard knockouts only" rule for early screens. ↩