Online Stores: Where the Ad Money Actually Goes

1. 89% of your products may return nothing — and the answer is not to switch them off

In a sixty-day audit of one store: 567 products spent budget, and 506 of them — 89% — returned nothing at all. Those products absorbed 5,541 shekels, which is 67% of the entire spend.

The number is alarming, and the quick conclusion drawn from it is wrong. Switching off 506 products at once dismantles the campaign structure, sends it back into learning, and deletes products that never got a fair chance. The number is not an execution list — it is a sign that the segmentation is wrong, not that the products are bad.

2. Segment by click efficiency, not by price or brand

When we re-segmented the products in that same account — not by price and not by brand, but by cost per click — the real picture appeared:

The waste was not in the cheap products. It was concentrated in the expensive-click ones. Had we segmented by price the way everybody does, we would have switched off the most profitable slice of the account.

This is not a technical detail: it is the difference between a sales campaign that earns and one that merely looks busy.

3. The cheap product is not the problem — it is the entry to the basket

In the same account: a product at ₪67 pulled a basket worth ₪570 — roughly three times its own price. And the average basket in the cheap-click segment was ₪529.

The customer arrives for something inexpensive and leaves with a large basket. Anyone measuring product profitability by its price alone cuts off the entrance to the basket while believing they are cutting waste.

4. ROAS 1.45 is not profit — it is a loss

Two brands in that same account spent ₪889 and returned ₪1,285. ROAS 1.45; the number is above one, so it looks like profit.

It is not. At a normal retail margin — between 25% and 35% — ₪1,285 of sales leaves roughly ₪320 to ₪450 in your pocket. And you paid ₪889. That is a real loss, and it will keep appearing in reports coloured green.

The practical rule: calculate your own ROAS threshold from your margin, not from a generic figure. At a 30% margin, anything below roughly 3.3 loses money.

5. The conversion counter lies

In that account the ads dashboard reported 178.5 conversions. The real orders were 81.

The cause: two conversion actions recorded for the same event, both set as primary — so every order was counted twice. The effect is not cosmetic: cost per order read ₪47 when it was actually ₪104. The account appeared to be buying an order at half its true price, and every budget decision built on that was built on an illusion.

Before any performance reading: open the conversion settings and count how many events are set as primary. Then reconcile the figure against the order count in the store's own dashboard — not the ads dashboard.

6. The smallest budget may be the highest return

In the same audit, the highest return in the whole account came from a category that had spent only ₪221, at ROAS 19.17.

Budget was not the constraint. Allocation was. Before asking for a budget increase, ask: what is working right now on the smallest amount, and why is it not getting more?

7. The campaign that has not sold in months

In another store, three campaigns were together spending about ₪1,031 a month with zero sales — one of them at a 0.25% click-through rate. All three were switched off, daily spend fell from ₪123 to ₪90, and ROAS 1.92 remained on what actually sells.

Campaigns do not die on their own. They keep reading "active" and spending quietly, and nobody opens them because the monthly total looks reasonable.

8. A link with no image is a click that never happened

Auditing two different stores, we found zero OG tags. No `og:image`, no `og:title`.

That means the store's link, when sent on WhatsApp or posted to Facebook, arrives with no image and no title — bare text. In a market where most customers arrive from WhatsApp, that kills the click before it happens.

It is the cheapest fix with the highest return in any store we have seen: two tags in the page head.

9. The pixel alone is no longer enough

iOS settings cut browser-side pixel tracking, so a share of your purchases never reaches the platform at all — and the platform optimises towards what reached it, not towards what happened.

The minimum for a serious store today: the pixel plus a server-side conversions interface, and the full set of store events — product view, add to cart, checkout started, and purchase with value and currency. Without the value, there is no ROAS to speak of.

10. The catalogue is the campaign

In the fashion stores we worked with, what worked was not the ad but the catalogue structure: Google Merchant, the Meta catalogue and the Instagram feed connected from a single source, and product titles written the way people search rather than the way the importer names them.

In other stores the catalogue itself was the product: a ceramics showroom rebuilt its Instagram account as a visual catalogue by room and style, and the post became a direct route to a showroom visit.

11. Payment and shipping kill the basket more than price does

Auditing one store, we found the checkout page literally displaying "no available payment methods" — an entire store unable to take money. In another, a product priced at zero that could be ordered, and a product with no name.

By contrast, the stores that genuinely sell carry six payment methods or more — card, PayPal, Apple Pay, Google Pay, Bit and cash on delivery — with a shipping company wired into the order.

Before spending a shekel on advertising, buy from your own store, on a phone, all the way to the thank-you page.

12. The launch model: drops, not permanent stock

In the fashion and jewellery brands we launched, what built demand was not constant advertising but the rhythm of launches: a teaser, a reveal, then a collection — three or four times a year, with styling content showing a complete look rather than a single piece.

Deliberate scarcity creates a date the audience waits for. Permanent stock presented the same way every day creates nothing.

Recurring mistakes

• Reading performance from the ads dashboard instead of the store's
• Switching off everything with zero return, all at once
• Segmenting campaigns by brand or price instead of click efficiency
• Treating any ROAS above one as profit
• Running campaigns before measurement is installed with value and currency
• Product photos from the supplier's site instead of your own photography

Ninety days

Month one — cleanup: buy from your own store · fix the OG tags · install full measurement with value · count the primary conversions and delete the duplicate · calculate your ROAS threshold from your margin.

Month two — segmentation: pull cost per click for every product · re-segment by that efficiency · isolate the expensive-click products into a campaign you watch separately.

Month three — scaling: scale what works on the smallest budget · switch off what has not sold in two months · reconcile every number against the store dashboard before any decision.