Back-to-back Shopify sales on the same products
Running 20% off this week and 30% off next week on the same products: why SafeSale will not stack the second sale on the first, and the sequence that works.
By Beacon Wave Studio · · 7 min read
A common promotion shape is a ladder. Twenty percent off the autumn range this week, thirty next week, fifty for the last three days before the new collection arrives. Or the reverse, a daily deal that rotates through the same bestsellers at different depths. Either way you end up wanting a second sale on products that are already in a first one, and in SafeSale you will meet a refusal when you try to confirm it. This post explains the rule behind the refusal, why we chose it, and the sequence that gets you a clean step-down without it.
The rule: one sale per variant at a time
When you confirm a sale, SafeSale takes a lock on every variant the sale will change. The lock exists for as long as the sale is scheduled, applying, active or ending, and it is removed when the sale ends or is cancelled. While it exists, no other sale can claim that variant.
You see the rule in two places. The preview of any draft lists conflicts near the top: the name of the other sale and how many of your variants it already holds. And if you press Confirm anyway, the app stops with a message along the lines of: some of these variants are already in another active or scheduled sale, SafeSale never layers two sales on one variant, end or edit that sale first. Nothing is changed in Shopify at that point. The second sale stays a draft.
Note the word scheduled. The lock is taken at confirmation, not at start time. A sale confirmed today for next Friday already holds its variants, so a sale for this Friday on the same products is refused even though their dates do not touch. That is the part people trip over, and it is intentional, for the reason below.
Why layering is refused rather than allowed
SafeSale has two mechanisms. A price-rewrite sale writes the discounted price into each variant and keeps a snapshot of what was there before. An automatic-discount sale leaves prices alone and creates a Shopify discount. The lock applies to both, but the reason is clearest with the rewrite.
A rewrite snapshot records the price the sale found. If a second sale were allowed to apply on top of a first, it would find the sale price and record that as the original. Now there are two snapshots for one variant and they disagree. End the first sale and it restores the pre-sale price under a live second sale. End the second and it restores the first sale price as though that were normal. Whichever order you end them in, something is wrong, and a stocktake three months later finds a product that has been quietly 20% cheaper since October.
With automatic discounts the catalog is safe, but the shopper is not. Shopify decides how discounts combine, and two product discounts on the same items either stack to a depth you did not intend or compete in a way you cannot explain to a customer who saw the first banner. Rather than have the rule depend on the mechanism, SafeSale uses the same rule for both: one sale owns a variant, full stop. The post on cancel, end and retry covers what releasing that ownership involves.
The sequence that works
A step-down ladder is three sales run one after another, not one sale edited three times. Each has its own snapshot and its own log, and each starts from the real original prices because the previous one restored them. The practical steps:
- Give the first sale a fixed end. The end job restores prices or removes the discount at that time and releases the locks. A sale with no end date, as the open-ended sale post explains, holds its variants until you remember it.
- Build the second sale as a draft now. Targets, percentage, rounding, dates, the name. Drafts take no locks, count against no plan limit, and can sit there for weeks. Just do not press Confirm yet.
- After the first sale ends, preview and confirm. Open the draft, look at the preview computed from the restored prices, and confirm. If you want the second sale to begin at a precise later time, keep its Starts field set; confirming a sale with a future start simply schedules it.
If you previewed the draft while the first sale was still live and come back to confirm it afterwards, expect the message that prices changed since you previewed. The earlier preview was built on reduced prices, the restored ones are different, and the fingerprint the app checks no longer matches. Reopen the preview and confirm from the fresh one. It is a few seconds of friction in exchange for never applying a plan you did not see against the prices it will actually touch.
Minding the gap
Between the first end job and your Confirm the products are at full price. If the first sale ends at midnight and you confirm at nine in the morning, that is nine hours of no sale, and if your emails said otherwise, nine hours of support tickets. Two ways to shrink it:
- Set the first sale to end a little before the advertised switch, be at the keyboard, and confirm the second the moment the first shows Ended. The end job and the apply job each take a minute or two on a few hundred variants, so a ten minute gap is realistic.
- End the first sale manually from its page, which releases the locks as soon as the restore finishes, then confirm the second. Same idea, with you choosing the moment rather than the clock.
What you cannot do is make the switch atomic, and we think that is fine. A short window at full price is a known quantity. Two sales with intertwined snapshots are not.
Only the overlap is blocked
Locks are per variant, not per sale. If week two adds a second collection that week one never touched, the variants in that collection are free, and a sale that targets only them can be confirmed and scheduled today. A tidy pattern for a growing ladder is therefore two sales per step: one you can schedule immediately for the new products, and one you confirm after the previous step ends for the carried-over ones. The mixed targets post explains how collections, tags and vendors resolve to variants, which is what the lock is counted on.
The same per-variant logic is why a conflict does not necessarily mean the whole second sale is wrong. The preview tells you how many variants the other sale holds. If it is three out of four hundred, the fix may be to exclude those three from the new sale rather than to wait.
Stepping up instead of down
Everything above works in either direction, but a word on the reverse ladder. Starting at 50% and climbing back to 20% is technically a sequence of sales like any other. Commercially it tends to anger the people who bought on day one of step two, and in several markets a later, smaller discount shown against the same compare-at price invites questions about which price was ever the real one. If you run a reverse ladder, keep the compare-at honest, as the was-now pricing post describes, and prefer a sequence that always deepens.
A worked ladder
Autumn range, 320 variants, three steps. Sale A: 20% off, starts Monday 00:00, ends Sunday 23:50, price rewrite with .99 rounding. Sale B: 30% off, same targets, drafted on Monday, left unconfirmed. Sale C: 50% off, same targets plus the accessories collection, drafted at the same time.
Sunday night, A ends at 23:50 and restores the originals. You preview B against those originals, confirm it at 23:56 with no start time, and it applies by a few minutes past midnight. The following Sunday the same happens with C. Because the accessories were never in A or B, you could have confirmed an accessories-only 50% sale a week earlier and left C to cover the carried-over range. Three sales, three snapshots that each point at the true original prices, and a change log that reads the way the promotion was planned. That is what the refusal is buying you.
Frequently asked questions
Can I schedule next week's sale on the same products while this week's sale is still running?
Not in SafeSale, if the two sales share variants. A variant is locked from the moment its sale is confirmed, scheduled or live, until that sale ends or is cancelled. The preview of the second sale lists the conflict by sale name and variant count, and Confirm refuses until the first sale is over. Variants that are only in the second sale are unaffected.
Why not just let the second sale deepen the first?
Because a price-rewrite sale records the price it finds as the original to restore later. A second rewrite applied on top of the first would record the sale price as the original, and ending either sale would leave the catalog wrong. Refusing to layer is what makes the restore exact.
I previewed the second sale during the first one and now Confirm says prices changed. Why?
The preview you approved was computed from the first sale's reduced prices. When that sale ended, the originals came back, so the fingerprint no longer matches. Open the preview again, check the new numbers, then confirm. This is deliberate: the app never applies a plan you have not seen against the current prices.
Will there be a gap at full price between the two sales?
A short one, yes: between the end job restoring prices and you confirming the next sale. If a gap matters, end the first sale manually a few minutes before the published switch time and confirm the second immediately after. Automatic-discount sales have the same lock rule, so the sequence is the same.