Google Ads Conversions — What to Measure and How to Set It Up

Jakub Pytlik | April 19, 2026 | 8 min read

What a conversion is and why it's the foundation

A conversion is any valuable customer action you record in Google Ads. A purchase in the store, a filled form, a call lasting longer than a minute, an account registration, a booking, a whitepaper download — all of these can be conversions. They share one trait: they tell Google whether the campaign is doing what it was meant to.

Without conversions, Google Ads has no way to optimize for the business. It looks only at clicks and impressions, which don't answer whether you're earning. That's why an account without properly configured conversions is like a car driving through fog without a compass — you can burn fuel, but you won't get where you want.

Conversion types — macro vs micro

That's the key split, and it shapes the whole strategy.

Macro conversions — these are money

Actions that translate directly into a sale or its closest step. This is what you actually optimize the campaign for.

• E-commerce: purchase, transaction
• Lead gen B2B/B2C: filled contact form, demo request, call > 60 sec
• Local services: online booking, call > 60 sec
• SaaS: account registration + activation (e.g. adding the first integration)

Micro conversions — these are signals

Actions earlier in the funnel, showing interest but not yet a sale.

• Add to cart (e-com)
• Newsletter sign-up
• Whitepaper / catalogue / case study download
• Click on "pricing" or "reviews"
• Spending > 2 min on the site
• Watching > 75% of a video

Why distinguish them

Two reasons:

1. Optimizing for the right metric. If you set everything as "primary conversion", the system optimizes for the sum — and may catch students filling whitepapers instead of customers buying for 5,000 PLN. Only macro should be primary, micro as secondary (signal).
2. Diagnostics. Micro conversions show where the funnel leaks. Lots of clicks + lots of add-to-cart + few sales = a checkout problem (payment, shipping costs). Lots of clicks + few add-to-cart = a product page problem (photos, copy, price).

Step 1 — how to set up a conversion

Three main methods, each for a different situation:

For most e-commerce stores I recommend a combination of GTM + GA4 import:

• GA4 captures everything (purchases, micro conversions) via the dataLayer
• GTM exposes the conversion to Google Ads (better attribution than a pure import)
• From GA4 you import only micro conversions as signals (secondary)

For lead gen (forms, calls, bookings) — skip the GA4 import. Configure conversions directly via GTM with a Google Ads Conversion Tracking tag. The GA4 delay (3–24h) matters with leads where you want to react quickly.

How to set up a store purchase, step by step

1. In Google Ads: Tools → Conversions → New conversion action. Pick "Website" → type "Purchase / Sale".
2. Set the name ("Purchase — store"), category "Purchase", enable "Use different values for each conversion" (transaction value).
3. Copy the conversion snippet or use GTM: tag "Google Ads Conversion Tracking" → trigger on the purchase dataLayer event.
4. DataLayer variables: {{transaction_id}} as the Order ID (prevents duplicate counting), {{value}} as the value, {{currency}} = PLN.
5. Test: do a test purchase → check in Google Ads that the conversion was recorded with the right value. Time: 1–4h, max 24h.

Step 2 — conversion value (the key element)

This is the single biggest difference between a well-configured and a badly configured account. Without conversion value, Smart Bidding optimizes for the number of conversions — not the actual revenue.

A practical example: a clothing store has two product lines — t-shirts (200 PLN) and dresses (600 PLN). Without conversion value the system treats both sales the same. It optimizes for the cheaper queries → grabs more t-shirt purchases. Result: ROAS drops, even though the order count rises. Once you add value, the system understands that a dress is 3× more important and allocates the budget differently.

Three ways to set the value

• Dynamic transaction value (recommended for e-com) — value from the dataLayer reads the actual basket total. Each transaction has its own amount.
• Fixed value per conversion type — e.g. demo request = 150, contact form = 80, call > 60 sec = 100. Works for lead gen and services.
• Point scale per CRM stage (B2B) — newsletter = 10, demo = 100, MQL = 300, SQL = 1,000, Closed-Won = deal value. Requires offline conversions / OCI.

The worst setup: every conversion with value "1". The system can't tell them apart. That's common in old accounts that have never been updated.

Step 3 — Enhanced Conversions and offline conversions

Two features that help Google see more of your conversions — not just the ones caught by a cookie.

Enhanced Conversions — recovering tracking after cookies

The customer clicks the ad, comes back later via a bookmark, and buys on a different device — classic tracking won't see it. Enhanced Conversions send hashed data (email, phone) from the purchase form to Google. Google matches it with logged-in user accounts and attributes the conversion to the correct click, even after the cookie has expired.

From my experience the uplift is typically +5–15% recorded conversions with no site changes — the scale depends on how many customers come back via other devices and how clean your form is (email, phone in fixed fields). Setup: in Google Ads → Conversions → pick the conversion → enable Enhanced Conversions → choose the source (Google Tag, GTM or Google Tag Gateway — the last is advanced and requires server-side setup). Time: 30–60 min.

Offline conversions — call, CRM, brick-and-mortar store

Most companies have sales that Google can't see from the website. Three main types:

• Call conversions — a call from the ad or from a phone number on the site counts as a conversion after a set duration (typically 60 sec). Enable in Google Ads → Conversions → Phone calls.
• CRM offline import — export stage changes (MQL, SQL, Closed-Won) with the GCLID from HubSpot/Salesforce/Pipedrive. Google attributes them to the original clicks. Essential for B2B with a long decision cycle.
• Store visits + in-store purchases — a visit to a brick-and-mortar store or an offline purchase (e.g. local pickup). Requires Google Business Profile and enough traffic scale.

Google describes the Enhanced Conversions mechanism in the documentation Enhanced Conversions setup. Conversion Modeling (the model that fills tracking gaps caused by cookies / consent mode) — Conversion modeling. The "+5–15% conversions" uplift comes from accounts I run (the scale depends on form quality and how many customers come back via other devices).

Step 4 — attribution and conversion window

Attribution answers the question: who do we credit a conversion to when the customer interacted with the ad multiple times? The customer clicks the ad on Monday (Search), browses for a week, buys on Saturday after clicking Performance Max — who gets that sale?

Attribution models in 2026

Since 2023 Google has had data-driven attribution (DDA) as the default for new conversion actions. That's usually the best choice — the algorithm splits the conversion between clicks proportionally to their actual influence on the customer's decision.

Other models (last-click, first-click, time decay, position-based) are still available, but in 95% of cases DDA gives the most realistic picture. Since 2024 Google has significantly lowered the volume requirements for DDA and no longer publishes a hard threshold — from my experience the algorithm starts to work sensibly at around 100+ conversions per month. At very low volume (5–30 conversions per month) DDA is active but the data is too shallow — last-click then tends to be a more stable choice.

Conversion window — the time after the click

That's how many days after the click Google counts a conversion as coming from that click. The default is 30 days, but you can adjust:

• Impulse e-commerce (cosmetics, accessories) — 7–14 days is enough
• High-ticket e-commerce (furniture, electronics) — 30–60 days
• B2C lead gen (courses, financial services) — 30–90 days
• B2B — 60–90 days for MQL, longer for SQL/Closed-Won (via offline conversions)

Important: the GCLID expires after 90 days — that's the limit for very long cycles (enterprise B2B). Workaround: Enhanced Conversions for Leads based on a hashed email or customer identifier, which work beyond the GCLID limit. Requires implementation in the CRM (HubSpot/Salesforce have native integrations), but for 6–12 month cycles it's the only sensible way to keep the full pipeline tracked.

The most common conversion mistakes

1. Everything as primary conversion. 12 conversion types, all counted together, the system optimizes for the sum. It catches the cheapest (usually micro), not the most valuable (macro). Fix: only real macro go to "include in Conversions", the rest as secondary.
2. No transaction deduplication. The customer refreshes the page after the purchase → conversion counted twice. Fix: always set transaction_id as the Order ID — Google removes duplicates by itself.
3. Value = 1. The system can't tell a 50 PLN transaction from a 5,000 PLN one. It optimizes for volume, not revenue. Fix: dynamic value from the dataLayer for e-com, or a point scale for lead gen.
4. Counting conversions "every" instead of "one per click" for forms. A customer fills the form twice = two conversions. Fix: set "one per click" for lead forms, "every" only for purchases (where you want to count every transaction).
5. No Enhanced Conversions. 5–15% of conversions never land in reports. Money you don't see. Fix: enable in 30 minutes, takes effect immediately.
6. No offline conversions in B2B. The lead lands in CRM, gets worked for 3 months, you close — Google never finds out which click produced the deal. Fix: OCI from HubSpot/Salesforce.
7. 30-day conversion window for everything. B2B with a 90-day cycle loses most conversions in attribution. Fix: tune the window to your real decision cycle.

From my experience on accounts I run, I most often see mistakes 1 and 3 combined — the account has 8 conversion types flagged as primary, all with value 1. The system optimizes for the sum, grabs the cheapest queries, ROAS drops 30–50% month over month. Fix: 30 minutes in the panel, I usually see uplift in the first week after the change.

How to check whether conversions are working

Level 1 — is there any data

• In Google Ads → Conversions you see stats for the last 7 days (not zeros)
• Status "Recording" (green icon) on every conversion
• The transaction test passed (do a test purchase → conversion visible in 1–4h)

Level 2 — is the data complete

• Conversion value populated (the "Conv. value" column is not empty)
• The conversion count in GA4 matches Google Ads (a 15–20% gap is normal, especially after Consent Mode v2 — anything bigger is a problem signal)
• Enhanced Conversions: in the conversion details you see "Match rate > 50%"
• No duplicates (purchases < transactions in GA4)

Level 3 — is the system optimizing for the right things

• Only macro conversions are flagged "include in Conversions"
• Each campaign has roughly 15–30 value-based conversions in 30 days — per campaign (Google no longer publishes a hard threshold — from my experience Smart Bidding behaves erratically below this range)
• Data-driven attribution active as default (with 100+ conversions/month; on lower volume last-click tends to be more stable)
• Conversion window adjusted to the category (not left at 30 days)

If level 1 is OK but 2 or 3 is failing — that's the classic: campaigns look stable but the money goes where it shouldn't. The first thing to check after taking over an account.

About the sources: Mechanisms of conversions, Enhanced Conversions, Conversion Modeling and attribution as described in the Google Ads Help — Conversions documentation. Practical uplifts and category bands — from 11 years of running Google Ads campaigns at JP Ads. Numbers from my own practice are flagged "from my experience".