Repricing Done Right: Seven Rules That Save Your Margin

Automated repricing has a poor reputation, and part of that is deserved. Anyone who buys a tool and switches on the default rule "always one cent below the cheapest offer" has optimised nothing — they have automated and accelerated their own margin decline. Set up properly, repricing does the opposite: it protects margin at exactly the points where manual maintenance loses it every day.

What repricing is actually meant to solve

The starting point is mundane: in a shop with 40,000 items nobody can maintain prices. What happens in reality is a mixture of three states — and each of them costs money:

• Frozen prices. Most of the assortment was last touched months ago. Purchase prices have moved, so have competitors. Some items have become unsellably expensive since; others are sold far below their value.
• Panic reactions. Somebody notices a best seller has slipped to position seven on Idealo and cuts the price. Usually too far, because it is unclear how much would have been enough.
• Blind following. The competitor cuts, you follow — even when they have two units in stock and are back at the old level in three days.

Repricing does not solve these problems through faster reaction, but through decisions taken in advance. You determine once how to act in a given situation. The system executes that consistently — no daily form, no gut feeling, nothing forgotten.

The seven rules

1. The price floor is not a suggestion

Every rule needs a hard floor, and it is not built from the purchase price but from the full contribution margin: purchase, shipping, packaging, payment fee, portal commission and the expected return rate. Only that figure is the real floor. In practice it sits a regular 8 to 15 percentage points above what businesses believe their floor to be — the return rate is almost always underestimated.

2. Not every item gets the same rule

An assortment is not a single block. A sensible segmentation runs along competitive intensity and role within the assortment:

The most common mistake at the start: throwing every item into a single rule. That aligns your exclusive goods to a competition that does not exist.

3. Compare only with comparable offers

A competitor price is only a reference price if the offer behind it is comparable. A retailer with no stock, a fourteen-day lead time, no telephone support and €9.90 shipping is no benchmark for a shop that delivers next day. Filter out non-comparable offers — otherwise you chase a price nobody actually gets.

4. Limit the speed of reaction

Price changes need limits per unit of time: at most X per cent per day, at most Y changes per week. That prevents two things. First, price spirals in which two automated retailers drive each other down until both sell below cost. Second, loss of trust: customers watching an item and seeing a different price every day buy later, or not at all.

5. Price increases are the real lever

Most systems are only armed in the downward direction. Yet the larger effect lies the other way: when the cheapest competitor has sold out or delisted the item, the market is momentarily willing to pay more. If you only ever cut, you systematically miss those windows. A working rule steers position, not direction — and holding position when competition weakens means increasing.

6. Think of channels together

Shop, Idealo and Google Shopping follow different logics but draw on the same stock and the same margin. Without joint control your own channels undercut each other. The price difference between channels should be a deliberate decision — justified by commission and audience — not the by-product of separate rule sets.

7. Every change must stay explainable

When management asks why an item was 12 per cent cheaper yesterday, the answer has to be on the table in thirty seconds: which rule, which trigger, which limit. Systems that cannot do this get switched off after the first inexplicable incident — rightly so. Traceability is not a comfort feature; it is the precondition for a repricer surviving in day-to-day operation.

The typical structure of a rule set

A working rule set has three layers, checked in this order:

1. Target rule — what should be achieved? For example: position 1 to 3 on Idealo for A items.
2. Adjustment rule — how is that achieved? For example: cheapest comparable competitor minus €0.01.
3. Protective rule — what must never happen? For example: never below 18 per cent contribution margin, at most 7 per cent change per day, no adjustment with fewer than three comparable offers.

The protective rule always wins. When target and protective rule collide, the item is not adjusted but flagged — that is then a matter for assortment or purchasing, not for pricing.

What you need before you start

The most common cause of failed repricing projects is not poor software but a missing data foundation. Three things have to be right:

• Clean cost data per item. Without a reliable landed cost including ancillary charges, every floor is guesswork.
• Reliable matching to the competitor offer. EAN matching alone is not enough for variants and bundles.
• Clear ownership. Somebody has to own the rules and look at the exceptions once a week. A system without an owner runs wild within months.

With those foundations in place, the rest is craft. The effect shows up in two figures at once: manual maintenance effort drops noticeably, and margin rises — not because prices were raised across the board, but because the cases of selling too cheaply for no reason disappear.