Learn · Lesson 12 of 12
When to kill a product
The decision to stop is arithmetic plus a rule written before you were attached to the answer. Learn the four honest kill conditions and the one bad reason people hold on.
What this lesson teaches
- Write kill conditions before launch, in a form that can be checked
- Distinguish a product failure from a channel failure from a page failure
- Recognise sunk cost in your own reasoning and route around it
Killing a product is the most valuable skill in this curriculum and the least practised, because every incentive in the moment argues for one more test. The defence is to decide the conditions in advance, when the product is still an abstraction and you can be honest about it.
Four conditions worth writing down
- Economic: contribution per order, recomputed with the real landed cost including actual returns and actual shipping, falls below the level the channel requires. This is a kill regardless of any other signal, because it cannot be fixed downstream.
- Acquisition: after spending your chosen multiple of breakeven CPA across at least two distinct audiences and two distinct creatives, cost per order remains above contribution. State the multiple and the false-kill rate you accepted, as in lesson nine.
- Fulfilment: the return, damage or complaint rate makes the economics false. A 12% return rate on a £18.82 contribution removes roughly £2.26 per order before the returned units' shipping is counted twice.
- Opportunity: the product occupies capital or attention that a better-evidenced candidate needs. This is a legitimate kill even when the product is mildly profitable, and it is the one people never write down.
Locate the failure before concluding it is the product
| Evidence | Where the failure is | What to do |
|---|---|---|
| CTR above bar, CPC fine, few orders | The page or the offer | Fix the page. The audience wants the click. |
| CTR below bar across several audiences | The offer, or the market | Kill, or reposition to a different buyer entirely. |
| Orders arrive, contribution negative | The economics | Kill, or re-source. No page fixes this. |
| Orders arrive, returns eat the margin | The product or its description | Fix the description first; it is cheaper and often it is the cause. |
Sunk cost, concretely
Money already spent on stock, photography, samples and testing is gone under every future course of action, so it cannot distinguish between them and must not appear in the decision. The only question is whether the next pound spent on this product returns more than the next pound spent elsewhere. The tell that sunk cost is operating is a sentence of the form 'we have already put so much into this' — which is an argument about the past offered in place of an argument about the future.
A kill is a finding
Record what was tested, what the pass bar was, what was observed, and which condition triggered. A killed product with a written record narrows the search for the next one: it eliminates a price band, an audience, a supplier's terms, a category's shipping physics. A killed product with no record eliminates nothing, and the same candidate will look appealing again in four months.
The end of the curriculum is the beginning of the loop
Landed cost, contribution, a pass bar written in advance, a stopping rule with a known error rate, and a record of what was learned. That loop is the entire job. Everything else is execution speed.
Figures in this lesson are illustrative inputs chosen so the arithmetic can be checked. They are not measurements. What FlowFinds Solutions actually measures is published, with its artifacts, in research.