All case studies
06 / Unit economicsSilver jewellery · Egypt · cash on delivery

A 618-product catalogue that didn't know its own break-even

A silver jewellery store scaling without a break-even number turned out to need 1.77 ROAS to cover costs and 2.11 once cancellations were counted, against a discount rate 16.2% deeper than intended.

1.77
True break-even ROAS
2.11
With cancellations
16.2%
Realised discount
The situation

A large catalogue — 618 live products — being advertised without an agreed definition of a good result. That sounds like a bookkeeping detail and it is the whole game: without a break-even, every ROAS is just a number to have opinions about, and an account can be scaled confidently for months in a direction that quietly loses money on delivered orders.

What I found
  • The true break-even was 1.77 ROAS, and 2.11 once the cancellation rate was included. Campaigns clearing 1.9 had been treated as winners and were losing money on every delivered order.
  • Realised discounting ran 16.2% deeper than the discount the store thought it was giving, through stacked codes and legacy promotions nobody had switched off.
  • Benchmarked against the closest comparable Egyptian silver retailer, the catalogue was priced around 27% higher — which is a viable position, but not an accidental one, and nothing in the ads was doing the work to justify it.
What I changed

Set the break-even, then re-read every campaign against it

Cost of goods, shipping, the payment fee and the cancellation rate resolved into two numbers: 1.77 to break even on paper, 2.11 to break even on what actually gets delivered and kept. Re-scoring the account against 2.11 rather than a vague sense of good moved several campaigns from the winners column to the losers column overnight, which is uncomfortable and correct.

Closed the discount leak

Stacked codes and forgotten promotions were audited and shut off, which recovered margin without touching the ad account at all. A store discounting 16.2% more than it means to is running a hidden price cut through every campaign it launches, and no amount of targeting work will find that.

Made a 618-product catalogue navigable

Tags, Arabic product naming, collection structure and on-page SEO were rebuilt, and a buy-two-get-one collection gave the account a way to raise order value that fit how the category actually sells. On a catalogue this size the constraint stops being traffic and becomes whether a shopper can find the second thing to buy.

What happened

The store came out with a break-even it can measure every campaign against, a discount rate that matches the one it intended to give, and a catalogue structured so a visitor can move through it. The scaling decisions did not get more aggressive; they got answerable.

The part usually left out

Most of the value here was accounting and store structure, not media buying, and the ad account looked worse the week after the audit than the week before — because campaigns that had been passing a soft target started failing a real one. That is the correct sequence, but it does mean the first honest report is a downgrade.

About this case

What people ask about this one.

A large catalogue with no known break-even — where do you start?
By computing break-even per product before any scaling, so every campaign is judged against a real number rather than a soft target. Without it you scale campaigns that pass a fake target and lose money believing you are winning.
Why does the first honest report after an audit look worse?
Because campaigns that had been passing a soft target start failing a real one. That is the correct sequence: the true number first, even if it dips, then scaling.
WhatsApp