Promotion Strategy: When an Event Is Worth Running

Candidate boundary

Decide now: Which single event, market, buyer job, and sellable product pool are under judgment?

Write into the action card: Write one line for event / market / language or currency / product pool / buyer job / current owner.

Read the result: If you cannot name one market or product pool, this is still an idea. Keep it in Watch; do not start campaign production.

Evidence boundary: A public calendar, hype, or platform date supplies a candidate only; it does not prove a store-specific reason to buy.

Hard stop line

Decide now: Is any one of price versus checkout, stock versus delivery, rights versus copy, or support versus returns already impossible to keep?

Write into the action card: Write the stop condition, the field that must be read back, and who can continue, narrow, or stop the work.

Read the result: One active stop line overrides the 25-point total: move to Prepare, narrow scope, or Skip.

Evidence boundary: This lesson does not read a live checkout, stock, carrier, rights file, or support state.

Value lane

Decide now: Without a sitewide price cut, why should the shopper act now: a bundle, member benefit, content, community relationship, or service promise?

Write into the action card: Choose one value lane and write the buyer job it serves, the page promise, and one pass signal.

Read the result: If the only answer is “a deeper discount,” return to margin guardrails and differentiation; do not make discounting the default route.

Evidence boundary: Content or community activity does not replace checks for price, stock, fulfillment, or consent scope.

Capacity and opportunity cost

Decide now: Whose creative, page, feed, stock, support, or media time will this consume, and what work will it displace?

Write into the action card: Name the capacity owner, the conflicting event, work removed from the queue, and the evergreen asset to protect.

Read the result: If you cannot state where the resources come from, keep Watch or Skip; busyness is not evidence of campaign value.

Evidence boundary: The example team’s capacity is not a real schedule or budget authorization for your team.

Evidence read-back

Decide now: Which current Shopify, GA4, UTM, order, inventory, or support fact supports or overturns this route?

Write into the action card: Record the field or URL, observation date, sample scope, pass/fail line, and next review date.

Read the result: When store facts conflict with the interaction rehearsal, revise the record and route—not the facts—to fit the score.

Evidence boundary: One screenshot, click spike, or engagement signal alone cannot prove profit, incrementality, or long-term brand effect.

Route and action card

Decide now: Is the route Do, Prepare, Watch, or Skip; what is the smallest next action, and when can evidence reopen it?

Write into the action card: Complete an action card with route, stop line, owner, next action, review date, and downstream lesson.

Read the result: Copyable notes carry the current choice and evidence gaps; write them into the real plan before moving into the next lesson.

Evidence boundary: An action card in this lesson is not campaign approval, a send instruction, discount configuration, or cross-team scheduling.

If the team later feeds event signals into demand forecasting, treat the forecast as an input to the preparation window, not as campaign approval. The method in EventCast encodes future promotion or holiday event summaries separately from historical demand sequences, then fuses them before a numerical predictor produces the forecast. This lesson borrows the separation of duties, not the paper's forecast errors: at T-45 / T-21, record the event assumption, historical baseline, forecast horizon, and product-pool / market scope, then recheck against actual inventory, delivery, and order read-back at T-7. The study uses daily demand records from one anonymous e-commerce platform, with offline evaluation from January through August 2025, a 14-day lookback, and T+1 through T+4 forecast horizons, based on arXiv v2; it does not show that a Shopify store will gain sales, have enough stock, or obtain causal sales lift from an event.

Copyable lesson notes: event fit scorecard
Candidate event: Prime Day / marketplace event
Candidate alternatives: Back to School, Annual retail event / BFCM, Cultural sports / national moment, Christmas / holiday gifting, Brand-owned member day
Campaign route: Prepare: fill evidence gaps
Primary route: Prime Day / marketplace event; enter the primary route only if it passes stop lines and can produce T-21 evidence in the current cycle.
Backup route: Back to School; keep it in Watch with explicit market signals, inventory changes, or creative validation.
Score: 14/25
Calendar lane: Next: prepare next month
Weak dimensions: Margin room
Shopify evidence gaps: Discounts / Offer, Markets / local promise, Pages / Theme sections, UTM / GA4, Inventory / delivery cutoff
First move: Check marketplace substitute pricing first, then design member perks, bundles, or a brand-search defense page.
Decision logic: the more popular event is not always the right event. Purchase intent, margin, creative reason, operations readiness, and risk control must all clear the line.
Admin evidence: Products / variants, inventory coverage, replenishment lead time, delivery cutoff, Discounts, Markets, landing page, UTM, GA4, ads manager, and support questions.
Source boundary: Shopify, Amazon, and NRF references provide candidate events and backend capability boundaries; Do / Prepare / Watch / Skip is decided by store evidence.
Owners: growth owns whether to run; product owns SKU and stock; finance owns margin; operations owns shipping and support; media owns traffic entry.
Review window: if score is below the execution line, fill evidence first; re-score 7 days before launch; check first-hour orders for evidence conflicts.
Quick Fit Gate: Candidate boundary
Decision: Which single event, market, buyer job, and sellable product pool are under judgment?
Write into the action card: Write one line for event / market / language or currency / product pool / buyer job / current owner.
How to read the interaction result: If you cannot name one market or product pool, this is still an idea. Keep it in Watch; do not start campaign production.
Evidence boundary: A public calendar, hype, or platform date supplies a candidate only; it does not prove a store-specific reason to buy.
Quick Fit Gate: Hard stop line
Decision: Is any one of price versus checkout, stock versus delivery, rights versus copy, or support versus returns already impossible to keep?
Write into the action card: Write the stop condition, the field that must be read back, and who can continue, narrow, or stop the work.
How to read the interaction result: One active stop line overrides the 25-point total: move to Prepare, narrow scope, or Skip.
Evidence boundary: This lesson does not read a live checkout, stock, carrier, rights file, or support state.
Quick Fit Gate: Value lane
Decision: Without a sitewide price cut, why should the shopper act now: a bundle, member benefit, content, community relationship, or service promise?
Write into the action card: Choose one value lane and write the buyer job it serves, the page promise, and one pass signal.
How to read the interaction result: If the only answer is “a deeper discount,” return to margin guardrails and differentiation; do not make discounting the default route.
Evidence boundary: Content or community activity does not replace checks for price, stock, fulfillment, or consent scope.
Quick Fit Gate: Capacity and opportunity cost
Decision: Whose creative, page, feed, stock, support, or media time will this consume, and what work will it displace?
Write into the action card: Name the capacity owner, the conflicting event, work removed from the queue, and the evergreen asset to protect.
How to read the interaction result: If you cannot state where the resources come from, keep Watch or Skip; busyness is not evidence of campaign value.
Evidence boundary: The example team’s capacity is not a real schedule or budget authorization for your team.
Quick Fit Gate: Evidence read-back
Decision: Which current Shopify, GA4, UTM, order, inventory, or support fact supports or overturns this route?
Write into the action card: Record the field or URL, observation date, sample scope, pass/fail line, and next review date.
How to read the interaction result: When store facts conflict with the interaction rehearsal, revise the record and route—not the facts—to fit the score.
Evidence boundary: One screenshot, click spike, or engagement signal alone cannot prove profit, incrementality, or long-term brand effect.
Quick Fit Gate: Route and action card
Decision: Is the route Do, Prepare, Watch, or Skip; what is the smallest next action, and when can evidence reopen it?
Write into the action card: Complete an action card with route, stop line, owner, next action, review date, and downstream lesson.
How to read the interaction result: Copyable notes carry the current choice and evidence gaps; write them into the real plan before moving into the next lesson.
Evidence boundary: An action card in this lesson is not campaign approval, a send instruction, discount configuration, or cross-team scheduling.
Priority rule - Primary route: Keep one or two events that pass their stop lines and can produce T-21 evidence in the current cycle.
Priority rule - Backup route: Keep the next candidate in Watch with explicit market signals, inventory changes, or creative validation to observe.
Priority rule - Watch signals: Name the Watch metrics, sample boundary, and review date; weak signals stay out of main budget.
Priority rule - Freed capacity: Write the design, development, support, or budget capacity freed by Skip back into the primary route or evergreen assets.
Priority rule - Sudden trend insertion: A sudden trend must name the resource moved from a locked event, the narrowed product and market, the shortest page and tracking evidence, and who can stop it immediately; otherwise keep it in Watch.
Catalog check - Small catalog (2–3 SKUs): Check every SKU, variant, stock level, and delivery promise, and leave a written campaign check record instead of sampling away a single gap.
Catalog check - Larger catalog sampling: Sample by hero product, market, margin risk, feed warning, and return history, while recording the sample scope and conclusion.
Campaign-window work: monitor budget, inventory, add-to-cart, checkout, support questions, refunds, stockouts, and page errors; do not scale from ROAS movement alone—locate product-pool, page, price, or channel issues first.
Conditional next route: Annual Event Calendar remains the single next-lesson route for annual planning; marketplace pressure goes to the Prime Day route case, offer issues to offer guardrails, feed issues to feed readiness, landing-page issues to landing-page work, owned-channel issues to owned-channel work, and localization issues to Markets localization.
Stop condition: pause main traffic when inventory coverage is below the campaign window, replenishment lead time cannot catch up, shipping cutoff cannot be kept, post-discount contribution profit is below the floor, or price, feed, IP, or support promises cannot be evidenced.
Watch / Skip value: Watch keeps observation metrics and the next review window; Skip moves budget and team time back to evergreen assets, page conversion, inventory preparation, or a higher-fit event.

The copied result is a plan draft to verify, not an automatic approval. Write it into the campaign plan alongside live product, price, stock, delivery, consent, and owner evidence before moving into the next lesson’s calendar, offer, or channel work.