Are Your Google Ads Conversions Real? Check the Tracking Against Your Orders

Take a hypothetical perfume store in Saudi Arabia, with round numbers. Google Ads says its campaigns brought in 120 conversions this week. The store's dashboard says the whole business took 90 orders across every channel, 72 of them paid or delivered. An ad account that claims more sales than the entire store isn't measuring differently. It's counting sales that never happened.

A Google Ads conversions number is real when it sits close to what your store or CRM recorded for the same days, and never above the store's total from every channel. The weekly reconciliation sheet below settles it in eight checks. Interface details are correct as of October 2026.

I read the sales sheet before the ad report. The report shows what Google's tag saw; the sheet shows what reached the bank. No conversions number should steer bidding until it has survived a week against that sheet.

Why does Google Ads show more conversions than your store has orders?

Usually because the account counts one sale twice, counts things that aren't sales or adds modeled conversions. The first two are setup errors. The third is by design: the Conversions column includes modeled conversions for real sales Google couldn't observe, such as in browsers that limit cookies. On their own, they shouldn't push the total past your store's orders.

Hypothetical example, rounded numbers. The perfume store's 120 conversions are 64 from the Google Ads purchase tag, 40 from a GA4 purchase event someone imported and set to primary, and 16 taps on the WhatsApp button. Eight of the 64 are customers reloading the thank-you page. Purchases the tag recorded from ads: 56. On SAR 12,000 of weekly spend, the account showed SAR 100 per conversion; the truth is nearer SAR 214 per purchase.

What does Google Ads count as a conversion, and when?

Anything that matches a primary conversion action and follows an ad interaction. The default column files it under the day of the click, not the purchase. A conversion action is the account's definition of what counts.

The Conversions column counts primary actions only, unlike All conv., so compare against it. Which actions deserve to be primary is the first check in the Google Ads audit checklist.

Timing is the trap: a Thursday click that becomes a Sunday purchase shows up on Thursday, so last week keeps growing through a default 30-day click window. Reconcile with Conversions (by conv. time), from the columns menu, which files each sale under the day it happened.

Is one order being counted twice?

Usually through a missing transaction ID, two primary actions for the same sale or a tag installed twice.

No transaction ID. A transaction ID is a unique code sent with each purchase. Google ignores a repeated ID, but only within one conversion action. An ID starting GG_ means your page sent none. Open Chrome DevTools' Network tab, tick Preserve log and place a test order. Filter for googleadservices.com/pagead/conversion and find oid. If it's empty or identical on every order, you've found the duplicate.

Two primary actions for one sale. Google sets GA4-based conversions to secondary by default, to prevent counting the same event twice for bidding. Make the GA4 purchase primary alongside the Google Ads tag, and every sale both tools credit to Google Ads appears twice. That's the 40 in the example, and no transaction ID fixes it across two actions. Why GA4 and Google Ads disagree even without duplication is covered in why every platform claims the same sale.

Tracking installed twice. A store-platform integration sends the purchase, and the agency's Tag Manager tag sends it again. I found no published documentation of exactly what the Saudi store platforms Salla and Zid send. Whoever installed what, the test is the same: place one order and count the conversion requests leaving the thank-you page.

That test isn't specific to Google. At Argan Package in Saudi Arabia, I worked on the Snapchat, TikTok and X integrations, pixels and conversion tracking included, and every pixel faces the same questions: does it fire once per order, with which ID, on which page?

Are WhatsApp and call clicks being counted as conversions?

Yes, if someone set them up that way. For a phone number on a mobile site, Google says only the click is tracked, not the call, and a WhatsApp button works the same way: the conversion records a tap, not a message, and certainly not a sale.

In Saudi Arabia and Jordan, much selling starts in a WhatsApp chat, so the button is the cheapest result bidding can chase. A chat doesn't carry the Google click ID (GCLID) the way a page visit does, so a sale closed in WhatsApp reaches Google Ads only if you connect it to the ad click, for example by storing the GCLID with the chat in your CRM and uploading the sale as an offline conversion. Until then, they appear only in the store sheet. My rule, without exception: WhatsApp and call clicks are secondary. Watch them; don't bid on them.

When does Google Ads show fewer conversions than real orders?

When the tag never reaches the purchase or can't tie it to the click. The first is worse, because the lost order appears in no ad report and no model recovers it.

Start with the Status column under Goals → Summary (in some accounts, Goals → Conversions → Summary): Misconfigured means recording has stopped; Needs attention, that something needs fixing; Awaiting conversions, that none arrived in the last seven days.

Then test the payment gateway. If the tag waits for the thank-you page and the customer pays on the gateway's page and never returns, the sale goes unrecorded (my reading of the mechanism, not a Google rule). Place one order per payment method: card, mada (Saudi Arabia's national debit network), installments through Tabby or Tamara, cash on delivery. In the hypothetical store, the installment order never reaches Tag Assistant.

Before you blame the gateway, rule out the tool: Google says Tag Assistant's debug window stays open while you navigate within the same domain, and that multiple redirects can stop it from loading, so an off-site checkout may hide a tag that did fire. Repeat the order with DevTools on Preserve log and look for the conversion request on the thank-you page.

Consent mode rarely explains the gap for a store selling only in Saudi Arabia or Jordan: Google requires consent in the European Economic Area, the UK and Switzerland, and models consent-mode conversions only from 700 ad clicks over seven days per country and domain grouping. The exception is a cookie banner shown to every visitor with consent denied by default. In basic consent mode, tags stay blocked until the visitor accepts, so a purchase from anyone who ignores the banner goes unrecorded. Check what yours does for a visitor in Riyadh. Local data-protection law is a question for a lawyer.

How do you reconcile one week of Google Ads conversions against your orders?

Put three numbers side by side for one week, then run eight checks that explain the gap. Pick a full week that ended at least three days ago. If you sell cash on delivery, leave the paid-or-delivered column empty until that week's orders are delivered or refused.

Step 1: The week table

Enter Conversions (by conv. time) day by day, and take the other columns from your store dashboard or CRM. Daily figures follow the account's time zone; if it isn't Riyadh or Amman time, compare weekly totals only.

Day	Conversions (by conv. time)	All store orders, every channel	Paid or delivered orders
Sunday
Monday
Tuesday
Wednesday
Thursday
Friday
Saturday
Total

1. If the second column is above the third, the account is overcounting: run checks 2, 4 and 5. If it's far below and Google brings most of your paid traffic, look for a missing tag: checks 1 and 6.
2. Between the third and fourth columns sit cancellations and refused cash-on-delivery orders, which bidding never hears about unless you tell it. Recorded value versus collected cash needs its own sheet.
3. Don't expect a perfect match; expect a gap you can explain.

Step 2: Eight checks that explain the gap

#	Check	Where to look (Goals → Summary unless shown)	Fault	Fix
1	Tag fires on purchase	Status column; Tag Assistant, with a test order started from the home page	Misconfigured, Needs attention, or test order missing	Google tag on every page, thank-you pages and new domains included
2	Transaction ID	Tag Assistant Hits tab (transaction_id) or DevTools (oid)	Empty, identical, "undefined" or GG_	Send the order number: no customer details, 64 characters max
3	Count	Action → Edit settings → Count	Lead form on Every (the website default)	One for leads, Every for purchases; changes apply to future reporting only
4	Duplicate primaries	Primary actions and their sources	Ads tag and GA4 or store-app purchase both primary	One primary per sale
5	WhatsApp and call clicks	Click-based actions	A click counted as primary	Make it secondary; bid on purchases, or on CRM-qualified form leads uploaded back to Google Ads
6	Gateway return	Test order per payment method, with DevTools on Preserve log	A method whose order never appears	Ask the provider or developer, in writing, how the purchase is sent without a return
7	Enhanced conversions	Goals → Settings; request payload; Diagnostics after about 72 hours	em shows only tv.1~em, no hash	Connect the checkout email field to the tag
8	Click vs. purchase date	Conversions (by conv. time)	The week changes each time you look	Reconcile on conversion time

Same hypothetical example. The 120 conversions by conversion time against 90 orders proves inflation before you open a setting. Checks 2, 4 and 5 remove 64 of them (8 + 40 + 16), leaving 56, below the store's 90 orders from every channel, as it should be.

Which conversion should Google Ads bidding learn from?

From something the store or sales team can find in its own records: one purchase with a transaction ID, or, for a service business, a qualified lead. Not a click.

Fixing the count moves the hypothetical store's cost per conversion from SAR 100 to about SAR 214. If a campaign runs on Target CPA or Target ROAS set from the old number, reset the target to the corrected baseline the day you fix the tracking, or bidding will cut spend chasing a cost that never existed. Google says Smart Bidding strategies need a 7- to 14-day learning phase and that changing targets can restart it, so make the fix at least two weeks before White Friday (the name Black Friday goes by in Saudi Arabia and the Gulf).

For businesses that sell by phone, my rule is to optimize ads to a "qualified booked call" confirmed in the CRM and sent back to Google Ads as an offline conversion, not to raw form fills. Google's documented route is enhanced conversions for leads: the tag captures a hashed email or phone at form submission, and you later upload the qualified lead with the same data (Goals → Summary → Create conversion action → Conversions offline). If your CRM was wired in through the API some time ago, ask whoever built it: since 15 June 2026 Google has moved these uploads to the Data Manager API and blocks them in the Google Ads API, keeping legacy access only for integrations already sending uploads in the months before that date.

What should you ask your agency this week?

Four questions, answered in writing:

1. Which primary actions are live, and where does each come from?
2. Does every purchase send the order number as its transaction ID?
3. Do WhatsApp or call clicks feed bidding?
4. How is a purchase tracked when the customer pays off-site and doesn't return?

Without clear written answers, hold the White Friday increase until you have them. A bigger budget on an inflated number only buys more of the wrong conversions.

For a check from outside the agency, a Google Ads account audit starts with how conversions are defined and tracked, and I deliver findings within about 48 hours of read access. For team training, see corporate training programs.