Candidate set
Eligible products constrained by availability and catalog facts.
Ecommerce/Playbook
⊕ Playbook 02 / Merchandising
Turn product truth into deliberate discovery decisions.
Operate search, recommendations, collections, and bundles as reviewable programs instead of disconnected widgets.
Book a working session↗
Six phases from catalog context to storefront placement.
- 01OUTCOMEOPERATE
Define the shopper job
State whether the surface should help discover, compare, complete, or expand a purchase.
- 02CATALOGOPERATE
Build the eligible set
Use availability, category, market, product state, and explicit exclusions.
- 03SIGNALSOPERATE
Choose decision signals
Combine product similarity, behavior, curation, and commercial constraints deliberately.
- 04PROGRAMOPERATE
Author the ranking or connection
Make inputs, candidate logic, and output shape explicit.
- 05QUALITYOPERATE
Inspect examples and edge cases
Check cold starts, sparse catalogs, repetition, and unsuitable pairings.
- 06DELIVERYOPERATE
Bind the approved output
Deliver the decision to the intended storefront surface.
Decisions that can be inspected and reused.
01
Candidate rules
Eligibility and exclusion logic for each merchandising surface.
02
Product connections
Reviewable relationships attached to the products they connect.
03
Placement brief
Surface, shopper job, fallback behavior, and acceptance criteria.
The model proposes. The operator owns the shelf.
Review representative placements and explicit exclusions before delivery.
Prepare the work
Agents assemble the change and its evidence.
Review the exact output
The operator sees what changes and where it goes.
Approve the consequence
Publishing remains a deliberate human decision.
Merchandising checklist.
The shopper job is explicit
The placement has one clear purpose.
Eligibility respects catalog state
Unavailable or unsuitable products are excluded.
Cold-start behavior is defined
Sparse signals do not produce empty or arbitrary results.
Examples cover edge cases
Operators can inspect surprising decisions.
The destination is scoped
Outputs reach only the intended storefront surface.
Merchandising FAQ.
Is every recommendation personalized?
No. Some surfaces should be curated, contextual, similar, complementary, or rule-constrained.
Why separate product truth from ranking?
Catalog facts should remain durable while merchandising decisions change by surface and goal.
How should a team evaluate the result?
Review relevance, coverage, repetition, edge cases, and business outcome using approved measurement plans.
Can operators override the output?
Yes. Curation and exclusions should remain explicit parts of the decision system.
⊕ A working session, not a sales theater
Bring one merchandising surface.
We will map its shopper job, candidate set, decision logic, review, and delivery.
Book a working session↗