A strategy is the design of the offer: which products may appear, how much off each one, how many products a catalog holds, and who gets the scarce deals. The campaign only decides who receives it and when. Most of what makes a Randio offer different from a sale is decided here, in three choices.
1. The range matters more than the discount
A fixed 20% is a sale with extra steps: every customer sees the same number, and the number is what they remember. A range does something a sale cannot. With a range of 10–25% every product in every catalog draws its own percentage, so two customers see different prices on the same shirt, and one customer sees different depths across their own page.
The floor is the number to set first. It is the worst case for your margin, and nothing in the range ever goes below it. The ceiling is the best deal any customer can see. Between them, the Higher discounts setting decides how often a draw lands near the top: Even, Uncommon, Rare or Very Rare. The range is never violated; only the odds inside it move.
So a wide range with Rare behaves very differently from the same range with Even. The page looks just as generous, since the ceiling is visible in some catalogs, but the average discount stays close to the floor. That gap between what the offer looks like and what it costs is the whole trick, and it is yours to set.
2. Mix high and low margin products on purpose
A catalog that is all staples is dull. A catalog that is all hero products is expensive. The strategy lets you build both into one page without blending their margins, because every pool has its own discount.
Put staples, the products everyone should see, in a Guaranteed pool at a shallow fixed percentage, say 10%. Put the aspirational products, the ones people want but hesitate on, in a Chance pool with a deeper range. Each catalog then holds a sure thing and a maybe, and the maybe is the deal that makes the page feel like a win. The staple margin never funds the hero discount, because they never share a rule.
If stock matters, the Distribution setting weighs it: Even draws by chance alone, Balanced leans toward products with more inventory, Inventory strongly prefers them. A store with a warehouse to clear uses Inventory and lets the engine do the clearing.
3. Limits keep it from turning into a sale
Three limits, all on by default or one setting away:
- One discounted unit per product per catalog. The shop-wide default lives in Settings → Preferences. A customer can buy five of the shirt; four of them are at full price. This is what stops a personal price from becoming a wholesale price.
- Limited pools cap the deep deals. A pool marked Limited goes into catalogs until the number of customers you set has received it, then stops. In Outreach the engine hands it to random recipients at the start; on the storefront it paces the allocation across visits. Your 40% off goes to 100 people, not to everyone who shows up.
- Previously bought products stay out. Settings → Inventory can keep a product out of a customer's next catalog for a window after they bought it through Randio.
Size
A catalog holds between 2 and 12 products. Guaranteed and Limited pools take their slots first; Chance pools fill the rest up to the maximum. The editor shows how many distinct products your pools cover and the minimum it recommends for catalogs to stay mostly unique. When you cannot reach that number, lower the maximum rather than widening the pools with products you would not discount.
A first strategy
Starting points, not rules. One Guaranteed pool: your two or three best staples at a fixed 10%. One Chance pool: a collection you are happy to discount, at 10–25% with Higher discounts on Uncommon. One Limited pool: a single hero product at a deeper fixed percentage, capped at a number of customers you would be glad to lose that margin on. Six to eight products per catalog. Run it, read the Discount dependency chart in Analytics, then move the handles.
How to pick the depth itself is in How deep should your discounts be?, and the three pool types in detail are in Distribution types.