Home / Skills / E-commerce / Promotion Campaign Planning
E-commerce

Promotion Campaign Planning

AIBarsMITSkill language: English
License verifiedNo scriptsLow risk
DemoReal run ยท claude-sonnet-5-5 ยท October 2026
I sell handmade candles online, mostly to the US and Germany, and I'd like to run a promotion for the holiday season. Last year I did 20% off everything for one week: orders tripled, but I ran out of my three best scents on day 2 and I had lots of customer service emails about late delivery. My contribution margin before ads is about 45% of price at full price. I have about 1,800 units in stock, a lead time of 3 weeks to make more, and ad budget of 2,000. Help me plan this year's promotion, and tell me how to judge whether it was worth it afterwards.
Skill: promotion-campaign-planning

Holiday promotion plan

I haven't invented any numbers beyond your own. Where a number is unknown, I've marked it as an input you need to supply.

What last year's data says

At 20% off, margin per unit drops from 45% of full price to 25%, which is only 56% of the normal profit per unit. The break-even sales lift is 1/0.56 โ‰ˆ 1.8x baseline units, before ads and extra support cost.

Discount Margin per unit (% of full price) Profit per unit vs. normal Lift needed to break even
10% 35% 78% 1.3x
15% 30% 67% 1.5x
20% 25% 56% 1.8x
25% 20% 44% 2.25x

Last year's tripled orders probably cleared break-even on paper. Three things probably ate into that:

  • Lost sales: your three best scents sold out on day 2, so you missed demand you could have served.
  • Support cost: the late-delivery emails cost time and goodwill.
  • Pull-forward: some of those orders were likely bought early rather than extra. Check whether December and January dipped.

1. One-page plan

  • Goal: profit from the promotion, with a secondary aim of new customers who reorder. I wouldn't aim for "orders tripled" again, because volume isn't what limited you. Stock and delivery were.
  • Margin floor (my recommendation, your call): no offer below 30% of full price in margin, which means a discount of at most about 15%. Test stacked codes against this floor.
  • Offer structure:
    1. Don't discount the three best scents. They sell anyway and they are your supply constraint. Offer them at full price, or only inside bundles with a cap.
    2. Discount the rest of the range by 10โ€“15%. Pair the offer with a bundle, such as "3 candles for X", where the slow scents carry the discount.
    3. Gift with purchase or free shipping above a threshold. Either one raises order value and protects your price better than a deeper percentage. Set the threshold above your typical order. I need your average order value and shipping cost to size it.
    4. Early access for your email list, 24โ€“48 hours before the public launch. This rewards repeat customers and spreads out the first-day rush.
  • Ads (2,000): I'd put most of it into retargeting and your own list, because those audiences are cheapest. Hold back about 20โ€“25% as a reserve for the mid-campaign check. Broad prospecting for new customers is the riskiest use. I'm not assuming a cost per order, so track it in the first 48 hours and adjust.
  • Window: 4โ€“5 days around Black Friday and Cyber Monday, not a full week. A shorter window concentrates demand but caps the support load. Confirm the dates for each market (see the calendar).

2. Stock and the 3-week lead time

This is where last year went wrong, and the lead time is why you need to act now.

  • Order production by about Oct 20. Stock then arrives around Nov 10, with a buffer before Black Friday (Nov 27, 2026). An order placed after Nov 3 can't arrive in time.
  • Size the order from a range, not one guess. Take your normal daily units (I need this number), multiply by roughly 1.5x, 2x and 3x over the promotion days, then split by scent. Your 1,800 units should be allocated by scent against that forecast. Don't assume they are spread evenly across scents.
  • Decide the sell-out rule in advance. For example: the best scents get a waitlist or a "back on [date]" label, and a substitute scent is suggested in a bundle. Don't leave a sold-out item on a promotional page.
  • Ask your maker for the cost of a bigger order and check the cash needed. Overstock of candles keeps, but cash is limited.

3. Calendar (confirm every date for the US and Germany)

When Stage Must be true before the next step
Oct 7โ€“12 Pull last year's data, set the margin floor, pick the offer Baseline daily units and per-scent sales known
By Oct 20 Place the production order Quantities set per scent
Oct 20โ€“Nov 13 Build the email list, plan the creative, set up site and codes Codes tested, including stacking
~Nov 10 Stock arrives; count it Actual stock by scent confirmed
Nov 16โ€“23 Teaser emails; write support answers; confirm carrier cutoffs Delivery promises are true for each country
~Nov 25 Early access for the list
Nov 27โ€“30 Public sale (Black Friday to Cyber Monday) Check at hours 6, 24 and 48
Dec 1 onward Close; ship; send "order by [date] for Christmas" messages Shipping cutoffs published
Dec 7 First review of sales
Mid to late Jan Final review once the return window closes Refunds counted

Germany also has strong Advent-season buying, with the first Advent on Nov 29. Ask yourself whether a second, smaller moment makes sense there. Also check whether your US and German customers ship from the same stock and how long delivery really takes to each country.

4. Readiness checklist (open items for you)

  • [ ] Every offered price clears the margin floor after fees, shipping and returns
  • [ ] Stacked codes tested for the worst case
  • [ ] Stock by scent counted and compared to the forecast range
  • [ ] Sell-out rule decided (waitlist, substitute, hide)
  • [ ] Packing capacity per day known, since last year's delays were here
  • [ ] Delivery dates promised on the site match carrier peak-season times, for the US and Germany separately
  • [ ] Pre-written support replies for delivery, returns and price adjustments
  • [ ] Checkout and payment methods tested under load
  • [ ] Terms written: exclusions, end time and time zone, return rules, what price returns to
  • [ ] Reference prices: for "was" prices, the EU and Germany have specific rules, including showing the lowest price from the previous 30 days. The US has FTC guidance on reference prices. Please verify the current rules for both, and don't use a "was" price that wasn't real.

5. Risks

  • Baseline unknown. If you can't say what a normal week sells, you can't say whether the promotion made money. Get that number first.
  • A smaller discount may lift sales less than 1.5x. Your 3x last year was at 20%. Treat any lift I mention as an assumption.
  • Supply: if the order slips past Nov 10, you repeat last year's problem. Keep a fallback of limiting best-scent bundles.
  • Delivery to Germany is a cross-border risk, including customs and VAT, if you ship from the US or the reverse. I don't know your setup.
  • Worst case: the sale sells out of the discounted scents early, ads keep spending on traffic you can't serve, and delays produce refunds. The mid-campaign check should pause ads on sold-out items.

6. How to judge whether it was worth it

Compare against the baseline, not against zero.

Incremental profit = (contribution profit during the promotion โˆ’ ad spend) โˆ’ baseline contribution for the same days โˆ’ pull-forward dip โˆ’ later refunds and extra support cost.

Collect:

  1. Units, orders and revenue per day, and per scent, for the window, compared with your baseline
  2. Average order value and units per order
  3. Contribution profit after discounts, gifts and free shipping, and after the 2,000 in ads
  4. New vs. returning customers, and cost per new customer
  5. Sales for 2โ€“3 weeks after the promotion vs. the baseline, to net out pull-forward
  6. Refund and return rate, rechecked after the return window closes
  7. Support: number of emails, topics, and hours spent
  8. Sell-out hours: how long each scent was unavailable
  9. Repeat purchase rate of new customers after 60โ€“90 days

Decision rule: it was worth it if incremental profit is positive after the dip and refunds, with no best scent out of stock during the window. Note what to repeat, change or stop.

What I need from you

  1. A normal week's units and orders (the baseline), and the split by scent
  2. Average order value, units per order, and shipping cost per order for each country
  3. Whether the margin figure includes shipping and payment fees
  4. Where you ship from, and how long delivery takes to the US and Germany
  5. How many email subscribers you have
  6. Your cash limit for a production order

With these I can turn the forecast ranges into per-scent production quantities and set the bundle prices and free-shipping threshold.

Real run in an isolated folder with only this skill installed. Only the Skill and Read tools were enabled, so no command was run and no file was written. The store and all numbers are invented and the model had no access to a real store. The break-even figures were checked by hand afterwards. Event dates and advertising rules in the answer must be confirmed for each market before relying on them.

What it does

Starts from one goal and a margin floor, then helps you choose the offer (percentage off, spend threshold, bundle, free shipping, gift, early access, limited time) and which products can afford it, keeping best sellers and supply-constrained items out of deep discounts. It builds a calendar backwards from the event date with owners and checks, and a readiness list covering margin and stacked codes, stock as a demand range, fulfillment, support, site and payments, consistent messaging and written terms. After the sale it compares results with a baseline and nets out the post-sale dip, later refunds and support costs, because a busy campaign can still lose money.

Output

A one-page plan, a calendar, a readiness checklist with open items, a risk list and a review template.

Good for

Planning a holiday or flash sale, deciding how deep to discount, and judging afterwards whether the promotion was worth it.

Notes & risks

Low risk: pure instructions with no scripts, no network access and no file writes. Event dates, advertising and reference-price rules, consumer-protection law and platform campaign rules differ by country and change; the skill flags them for you to confirm and does not state what is allowed. Expected sales lifts are ranges and assumptions, not forecasts. It cannot see your store data. In the trial its dates and break-even figures were checked by hand, but dates still need confirming for each market. Original skill by AIBars (MIT). Tried once on an invented candle store.