Google Ads Conversion Tracking: A Correct Setup For 2026

Every audit I run starts in the same place, and it is never the campaigns. It is the Goals page. I sort the conversion actions, look at what has been marked primary, and inside two minutes I usually know why the account has stopped scaling. Four primary actions. Two of them counting the same purchase. A newsletter signup training the bidding algorithm as hard as a £900 order.

Correct Google Ads conversion tracking in 2026 is narrow, not comprehensive. One primary action per campaign goal, measured through a single source, with enhanced conversions switched on, consent mode passing all four signals, and counting and attribution settings that match how people actually buy from you. Everything else goes in as secondary for observation. Below: how to decide what counts, how to stop double counting between GA4 and the Google Ads tag, what changes with enhanced conversions in June 2026, and the settings that quietly distort your numbers.

Decide what counts as a conversion before you touch a tag

Most accounts track too much. That is the root problem, and no amount of tagging skill fixes it.

Smart Bidding optimises towards whatever you label primary. Primary actions report in the “Conversions” column and train your bid strategy when they sit in a goal used for bidding. Secondary actions are observation only: they report into “All conversions” and never steer a bid. So the decision is not “what can I measure”, it is “what do I want more of”.

The rule I apply on every account: one primary conversion action per campaign goal. Everything else drops to secondary.

Before: Purchase (primary), Add to basket (primary), Newsletter signup (primary), Contact page view (primary).

After: Purchase (primary). The other three secondary, still visible in All conversions, no longer voting on your bids.

Lead gen accounts feel this hardest. A form fill and a brochure download are not the same event and should never carry the same weight. If your sales team can tell you which enquiries turn into money, that is the signal worth optimising to, and it is the argument behind why ROAS on its own misleads you.

Pick one source of truth: GA4 import or the Google Ads tag

This is the most common broken setup I see, and it is invisible unless you look for it.

Put a Google Ads conversion tag on the thank-you page, mark the GA4 purchase event as a key event, import it into Google Ads, and both fire on the same order. Google Ads counts two. Your Conversions column doubles, your CPA halves on paper, and Smart Bidding chases a target that does not exist. Nobody notices because the numbers look good.

Pick one and delete or demote the other. My preference: GA4 import for ecommerce, where purchase value already flows through the data layer and you want reporting to agree across both platforms. The Google Ads tag direct for lead gen, where the event is simpler and enhanced conversions for leads gives you a route back to closed revenue.

How to find the problem in your own account: open Goals, sort by conversion volume, and look for two actions with similar names and suspiciously similar counts. That is your duplicate. Set the loser to secondary rather than deleting it, so you keep the history.

Switch enhanced conversions on, and know what changes in June 2026

Enhanced conversions supplement your existing tags by sending hashed first-party data, typically an email address, and matching it to signed-in Google accounts. It recovers conversions that browser restrictions would otherwise lose. Google hashes the data with SHA-256 before it leaves the page, and you need to accept the customer data terms in your account before it works.

Two things change this year, and both matter if you have a developer involved.

From April 2026, Google Ads accepts user-provided data from website tags, Data Manager and API connections at the same time, so you no longer have to commit to a single implementation route. From June 2026, enhanced conversions for web and leads merge into one on/off feature, and the method picker disappears from the interface. Existing advertisers who have already agreed the customer data terms migrate automatically.

The part to action now: offline conversion imports and enhanced conversions for leads uploads move to the Data Manager API and get blocked in the Google Ads API from June 2026. If you push closed-won deals from a CRM through a script or a middleware tool, check with whoever built it. That is a broken feedback loop waiting to happen, and a broken loop degrades bidding weeks before anyone spots it in a report.

One dependency people miss: Google needs the ad_user_data consent signal granted before it uses the data you send. Enhanced conversions and consent mode are one system, not two projects.

Treat consent mode as tracking, not paperwork

Consent Mode v2 has been mandatory for EEA and UK traffic since March 2024, and it carries four signals: ad_storage, analytics_storage, ad_user_data and ad_personalization. The last two decide whether Google can use your data for advertising and whether you can build remarketing audiences at all.

You have two ways to run it, and the choice has a real cost.

Basic consent mode blocks Google tags from loading until someone interacts with the banner. Nothing reaches Google before consent, and your modelled conversions come from a general model.

Advanced consent mode loads tags with consent defaulted to denied, then sends cookieless pings that carry consent state and event signal without identifiers. That earns you an advertiser-specific model, which is materially better than the general one.

Run advanced unless your legal position rules it out. Modelling needs a data collection threshold and at least seven full days of implementation before the uplift reporting appears, so give it time before you judge it.

The settings that quietly distort your numbers

Defaults are set for the average advertiser. You are not one.

Setting	Default	What I check	Why it matters
Counting	Every conversion	”Every” for sales, “One” for leads	”Every” on a lead form counts the same person filling in three enquiries as three leads
Attribution	Data-driven	Leave it on data-driven	Only data-driven and last click remain. First click, linear, time decay and position-based were retired
Click-through conversion window	30 days on Search and Display	Match it to your real sales cycle	Settable from 1 to 90 days depending on source. Google recommends at least 7
Value	Often blank	Pass dynamic value, or a sensible fixed one	Without value, target ROAS has nothing to bid on

Two notes worth carrying. Counting changes apply to future conversions only, so you cannot fix history by flipping a setting. And a conversion window shorter than your sales cycle does not just under-report, it starves the bid strategy of the exact conversions you most want more of. If most of your enquiries close in week four, a 30-day window is cutting it fine.

Check it on a schedule, not after a bad month

Tracking breaks silently. A developer ships a new checkout, a consent banner gets swapped, a tag stops firing, and the first symptom is a bidding strategy quietly pulling back.

• Weekly: glance at daily conversion volume for a flat line or a cliff. That shape is a broken tag, not a bad week.
• Monthly: open the Goals page. Confirm one primary per goal, no duplicates, and check the tag diagnostics report.
• Quarterly: compare Google Ads conversions against orders or enquiries in your own system. They will never match exactly. They should move together.
• After any site change: test a real conversion end to end before you trust the numbers.

Get a second pair of eyes on the data

I have never audited an account where the tracking was perfect, including accounts run by good people. It is the one part of paid search where the errors compound quietly and everything downstream inherits them. If you want someone to check whether your conversion data is telling the truth before you scale spend against it, take the free Google Ads audit. We go through the Goals page, the tags and the counting settings first, then the campaigns. If the setup needs rebuilding, that is part of how we run Google Ads management.

FAQ

How do I know if my Google Ads conversion tracking is working? Complete a real conversion on your own site, then check the conversion action in Google Ads a few hours later. Google reports conversions against the click date, so the number appears against the day of the ad click, not the day of the purchase. Use the tag diagnostics report to confirm the tag is firing and receiving data.

Should I use GA4 conversions or the Google Ads conversion tag? Use one for any given action, never both. Importing a GA4 key event while a Google Ads tag tracks the same event double counts it and misleads Smart Bidding. GA4 import suits ecommerce reporting consistency, the Google Ads tag suits lead gen and enhanced conversions for leads.

Why do Google Ads and GA4 show different conversion numbers? They count differently by design. Google Ads attributes to the click date and includes view-through and cross-device conversions using its own attribution, while GA4 uses session-based reporting and its own attribution settings. Expect a gap. Investigate when the gap changes shape rather than when it exists.

Do I need enhanced conversions if I already have consent mode set up? Yes, they do different jobs. Consent mode governs what you are allowed to send and models the gaps. Enhanced conversions improves matching for the data you are allowed to send. Together they recover a meaningful share of the conversions that cookie restrictions would otherwise hide.

What conversion window should I set? Match it to how long your customers actually take. Segment your conversions by days to conversion, find where the bulk of them land, and set the window past that point. A 30-day default under-reports any business with a long consideration period.