THE CHECKS

What Madar checks

Ten checks run on every audit. Each one finds something no other tool thinks to look for. Every figure on this page came out of a real store.

The big tools are not lying to you. They simply never ask. They read GA4's conversion count and take it at its word — the way every tool before them did.

Madar asks one question first: was that actually a sale?

Measurement integrity

Conversion integrity

Whether the number labelled conversions means sales — or means people dropping things in a basket and walking out.

Why the other tools walk straight past it

In GA4 you decide what counts as a conversion. The ecommerce templates decide for you: add_to_cart, ticked by default, without a word of warning. From there it is easy — every tool reads `conversions` and believes it.

What it found

One store was reporting 279 conversions. It had made 21 sales. Thirteen times over. Its real basket average was 146 SAR — every dashboard it owned was showing 11.

ecom_conv

An event is not a person

Funnel events firing more than once per person — quietly inflating every rate you build on top of them.

Why the other tools walk straight past it

GA4 puts Event count where you will see it and keeps Total users one click away. We fell for it ourselves, on a real customer, and reported 59 lost buyers who did not exist. This check is the scar.

What it found

add_payment_info was firing 2.77 times per person. That one number turned 11 lost customers into 59 — and it fooled us before it could fool anybody else.

event_integrity

Web tracking configuration

Data sources that were never connected — and sources that were connected, then went quiet.

Why the other tools walk straight past it

A source that stopped reporting looks identical to a source with nothing to report.

web_track

Internal traffic

Your own dashboard and your own ad tools counted as customers — purchases, revenue and all.

Why the other tools walk straight past it

GA4 has a filter for this. Almost nobody switches it on, because nothing ever tells them to. So the owner checking his own store ends up inside his own conversion rate.

What it found

In one store, 13.1% of all reported revenue came from the owner's own Shopify dashboard and Meta's ad tools. Not one riyal of it came from a customer.

internal_traffic

Measurement conflict

Two analytics streams measuring the same site — counting every session, every purchase, every riyal twice. Plus the dormant streams a migration left behind, still live.

Why the other tools walk straight past it

GA4 never warns you, and no analytics tool can see it: they all read from the PROPERTY, one level above the streams. So the question never gets asked. And a dormant stream from an old platform is a lit fuse — the day something switches it back on, every number in the business doubles overnight and nothing tells you why.

What it found

One store was running a live Shopify stream alongside a dormant MonsterInsights stream — left over from the WooCommerce site it abandoned three years ago. Still installed. Still live.

measurement_conflict

Attribution

Payment-gateway self-referral

Your payment gateway claiming the sales your ads paid for — while your own domain shows up as an outside referral.

Why the other tools walk straight past it

Every Gulf gateway is a redirect: Tap, Moyasar, HyperPay, PayTabs, Tamara, Tabby. The customer leaves to pay and comes back a brand-new visitor, courtesy of the gateway. Attribution looks perfectly healthy while crediting the wrong source on every single sale. And fixing it today does not fix last month.

self_referral

Channel quality

growth · plan

Which paid channel brings buyers and which brings window-shoppers — measured in revenue per session, not cost per click.

Why the other tools walk straight past it

Each platform reports its own clicks and its own conversions, and taken alone each one looks reasonable. Nobody lines revenue per session up side by side — the single view that shows which channel is paying you and which is billing you.

What it found

One store: Snapchat was returning 24x more revenue per session than Facebook and converting 8.2x better. Facebook was taking 93% of the paid traffic.

channel_quality

Google Ads conversion setup

Conversion actions that are dead, frozen, or misconfigured — while Smart Bidding optimises hard against the broken signal.

Why the other tools walk straight past it

Google Ads writes it out plainly — Misconfigured — on a screen nobody opens.

gads_conv

Performance

Page weight on mobile

A page too heavy for the phone in your customer's hand — and whether the weight is in the images (rarely) or the scripts (usually).

Why the other tools walk straight past it

The owner tests on a laptop and the page is fine. On a phone it takes seventeen seconds. Nothing tells him, because he has never once stood where his customer stands. Then a consultant tells him to compress his images — confidently, and wrongly.

What it found

One store: 1.6 seconds on desktop, 16.9 on mobile. The images were already compressed to the bone — 661 KB. The weight was 277 script files, 2.9 MB of them. The owner had decided his product was the problem. His product was fine. His page never opened.

page_weight

Tag sprawl

Several Google tag containers on a single page — some of them fetched twice — and the real price your tracking charges your customer in patience.

Why the other tools walk straight past it

Each container was added by some app or some setting, years apart, and nobody ever went back to take the old one out. Every one of them was installed to measure the store or to grow it. Together, they are why it crawls.

What it found

One store was loading four Google containers. Two of them twice — the same bytes, fetched again. 933 KB across six files. More than every image on the page put together.

tag_sprawl

What we do not check

Madar reads what your numbers say — GA4, Search Console, performance data. It does not step inside your page, read your code, or record your visitors.

So when you ask why that customer walked away from checkout, the honest answer is: we do not know. We can show you that she left, when she left, and how many did the same. The reason is not in the numbers.

We report the effect, and we say so plainly — instead of guessing at a cause and dressing the guess up as a diagnosis.

And the mistake we made

On 12 July we told a real store that fifty-nine customers had entered their card details and walked away. It was false. We had divided one event counter by another and invented forty-eight people who were never there.

We corrected it in the open. Then we built a check out of the wreckage — an event is not a person — so that nobody would repeat it. Not us. And not the store owners repeating it right now in their own dashboards, because GA4 puts Event count where you cannot miss it and keeps Total users one click away.

Every tool swears it does not get things wrong. Madar shows you the thing it got wrong, then hands you the guard it built out of it.

Run an audit →

No card. Connect GA4 and see what Madar sees in your store.

Every figure on this page was measured in a real store and can be checked. Not one of them is an estimate.