We compared what Meta and Google reported to what 11 Israeli online stores actually banked over nine months of 2026. Together, the two platforms took credit for 106% of everything the stores sold.
ROAS, return on ad spend, is the revenue attributed to your ads divided by what you paid for them: for every shekel of media, how many came back. Every store owner knows the moment. Meta's Ads Manager says 6.2. Google says 8.1. The bank account says something else. The question we hear most, from clients and from brands we meet for the first time, is which number to believe.
So we measured it. Our dashboards pull three things straight from the source for every client: store orders from Shopify, WooCommerce or Wix, spend and reported purchase value from Meta, and the same from Google Ads. For this article we took 11 Israeli ecommerce stores, all billed in shekels, all running Meta and Google side by side, from 1 January to 28 September 2026: 39,727 orders, ₪15.4 million in gross revenue and ₪2.9 million in media.
Two definitions, and we will keep to them for the rest of the piece:
Across the 11 stores, Meta and Google together attributed ₪16.3 million in purchases to their ads. The stores sold ₪15.4 million in total. Not from ads: in total, including organic search, returning customers, WhatsApp orders, email, word of mouth, everything.
That is 106%. According to the two platforms, nothing in these stores happened without them, and a little more on top.
The monthly view is even more instructive. In every single month, the reported blended ROAS sat above the real one. The platforms' claim was lowest in January (101% of revenue) and highest in July (115%), a month with big sales events, when email, WhatsApp and returning customers drive a large share of orders and the platforms still book them as theirs.
Pooled numbers hide the real problem. Look at the stores one by one, and the platforms' claim ranges from 48% of what the store sold to 129%.
In 5 of the 11 stores the platforms reported more revenue than existed. In 2 stores they claimed less than 80% of it. The median store sat at 94%.
This is why you cannot benchmark yourself against a reported ROAS you saw somewhere. A store with a 6.0 in Ads Manager can be doing better or worse than a store with a 4.0. The distortion is not a fixed tax you can subtract; it depends on your mix of returning customers, brand search, promo calendar and attribution settings.
| Measure | ROAS |
|---|---|
| Meta reported ROASpurchase value Meta attributes to its ads, divided by Meta spend | 4.84 |
| Google reported ROASsame, on Google Ads (10 of the stores run Google) | 7.27 |
| Both platforms, reportedall attributed value divided by all spend | 4.99 |
| Real ROASstore revenue divided by all media spend | 4.58 |
Per store, the median reported ROAS on Google was 7.27. On Meta, 4.84. Put the two together and the median reported blended ROAS was 4.99, against a median real ROAS of 4.58.
Google's number looks 50% better for a structural reason: a lot of Google's conversions come from brand search. Someone saw the product on Instagram, searched the brand name two days later, clicked the ad at the top, bought. Google reports the sale; often Meta reports the same sale inside its own 7-day click window. The store paid twice for the credit and received one order. Moving budget from Meta to Google because "Google's ROAS is higher" is the most common expensive mistake we see in new accounts.
Four mechanisms create the gap, and all four work exactly as designed:
The platform is not lying. It is counting what it saw. Your job is to count what you banked.
Real ROAS is total store revenue divided by total media spend across every platform, for the same period. It ignores attribution completely: it asks how much came in for every shekel that went out. In our data, the median Israeli store in 2026 ran at a real ROAS of 4.58.
Because Meta counts purchases it can connect to an ad within its attribution window, including view-through and modelled conversions, and it never subtracts cancellations or refunds. Google does the same with its own window. Both can count the same order. In 5 of the 11 stores we measured, the two platforms together reported more revenue than the store had.
There is no universal number: a good ROAS is one that clears your break-even with room to spare, and break-even is 1 divided by your contribution margin. At a 40% margin, a real ROAS of 2.5 only covers costs, so 4 and above is healthy; at a 25% margin you need 4 just to break even. For reference, the median Israeli store in our data ran at a real (blended) ROAS of 4.58 in 2026, while its reported blended ROAS was 4.99. Judge the real number against your margin, never a reported number against someone else's.
A real ROAS goal, derived from your margin: 1 divided by contribution margin is break-even, and the goal should sit above it. Each platform's reported ROAS is then used for one thing only: ranking campaigns inside that platform.
Google Ads has no price list, but it does have what stores actually pay. We opened the data of 10 Israeli online stores, 467 thousand clicks over nine months of 2026: what a click costs, what stores spend a month, and what an order costs.
Daniel Hirsh30 September 20267 min readEvery Preneuriat client gets a live dashboard: store revenue, all media spend, real ROAS, updated all day. The first call goes both ways.
Check the fit