Whichever one you pick, the thing that matters is that exactly one purchase action fires per order and you report on that action by conversion time. The failure mode is never the choice of method — it is running two methods at once and counting every order twice.

On this page

• What are the ways to track Shopify purchases in Google Ads?
• How do I check I am not double-counting Shopify orders?
• Which attribution model should I report Demand Gen on?
• What query gotchas hit Shopify conversion reporting?

What are the ways to track Shopify purchases in Google Ads?

The Google & YouTube channel app. Easiest, and it wires up the feed at the same time. It also installs its own conversion tracking, which is the source of most double-counting when a manual tag is already present.

Google tag through Shopify's customer events / a tag manager. More control, survives theme changes, requires someone to maintain it.

Server-side. Most durable against browser restrictions, most work, and worth it at volume.

How do I check I am not double-counting Shopify orders?

After setup, run one test order, then segment:

`` SELECT segments.conversion_action_name, metrics.all_conversions, metrics.all_conversions_value FROM campaign WHERE campaign.id = AND segments.date BETWEEN ... ``

One order should produce one purchase conversion. If it produces two, you have two tags. If it produces three, you also have a test action live.

Which attribution model should I report Demand Gen on?

A raw conversions column on a typical store rolls in page views, add-to-cart, begin-checkout, view-item, search and local actions like directions and calls. It is not a purchase count.

Measured case: the raw column read about 42 for a week where the real purchase count was 5.

Report the purchase conversion action only, by conversion time. That is the number that steers the account.

What query gotchas hit Shopify conversion reporting?

Cost is not segmentable by conversion action. You cannot pull spend and per-action conversions in one query. Run two — one for cost by campaign and day, one for the purchase action by campaign and day — and join them. Filtering on a conversion action also requires including that segment in the select.

Product identifiers. If you are also running Shopping, the identifier fields Shopify exposes are not always what Merchant Center consumes, and disapprovals that keep returning after every product update usually trace to the field mapping rather than the values themselves.

Related questions

How do I track which calls came from Google Ads if the client won't use a call tracking number?

You can still get most of the way there with dynamic number insertion on the website and Google's own call reporting, but you should tell the client plainly what they are giving up — and the gap is bigger than they think. Read the full answer.

Google Ads, GA4 and Shopify all show different conversion numbers. Which one should I trust?

Your order system is the truth for how many orders happened. Google Ads is the truth for what the ad platform is optimising toward. Read the full answer.

Can I use target ROAS to get Google to find new customers instead of repeat buyers?

No, and it usually does the opposite. If your bidder optimises a total purchase value, a target ROAS floor structurally rewards repeat-buyer traffic and penalises new-customer traffic — it is not neutral, it is directionally harmful. Read the full answer.

Written by Usama Khan, who runs Google Ads for e-commerce brands and home-services companies at AdventurAds. Figures are from accounts under active management and are anonymised. Last reviewed 2026-08-26.