Conversion rate is the ratio of users who reach a specific goal to the relevant user pool. A simple formula, but when calculated incorrectly, it produces misleading results. The most common mistake: combining all traffic sources and devices into a single number (blended CR).
Calculation
Suppose an ecommerce site receives an average of 1,000 visitors per day from various sources. 100 of these visitors complete a purchase: 100/1,000 = 0.1, a 10% conversion rate1.
The same calculation through an ad campaign: ads are shown to 1,000 people, resulting in 200 clicks. 10 of these 200 people complete a purchase: 10/200 = 0.05, a 5% campaign conversion rate.
The critical point: the denominator must always be the relevant user pool. A form page’s conversion rate should be calculated based on users who reached the form page, not all users who viewed the homepage. The homepage button click and form submission are separate processes and should be tracked with separate conversion rates.
Micro and Macro Conversions
A macro conversion is the primary goal: purchase, lead form submission, subscription start. Micro conversions are intermediate steps toward that goal: add to cart, newsletter signup, video view, 90% scroll2.
Micro conversions are used to diagnose funnel friction, not as primary KPIs. A macro or micro conversion definition can be created for any action expected to show growth over time.
Evaluating conversions by source/channel makes it possible to calculate return on investment for each channel and manage resources accordingly. To overcome channel-level attribution limitations, a Marketing Mix Modeling (MMM) approach can also be considered.
Why Blended CR Is Misleading
Numbers like “average ecommerce CR of 2.5-3%” are blended across industry, traffic source, and store size3. When display ad spend increases, overall CR drops even if nothing changed on the site.
CR by Traffic Source
| Source | Average CR |
|---|---|
| 5%+ | |
| Organic search | 3%+ |
| Paid social | 0.5-1.5% |
CR by Device
| Device | Average CR |
|---|---|
| Desktop | 3-4% |
| Mobile | 1-2% |
Source: Skailama, Build Grow Scale ecommerce benchmark reports4 5.
Comparison should be made against your own historical trend, not generic benchmarks. Two companies reporting the same conversion rate may be using completely different tracking methodologies3.
Cart Abandonment Rate
According to Baymard Institute’s meta-analysis covering 50 studies, the average cart abandonment rate is 70.22%6. Abandonment reasons:
- 48% unexpected additional costs (shipping, taxes, fees)
- 26% mandatory account creation
- 21% long/complicated checkout process
Mobile abandonment rate is 80.02%, desktop 66.41%. Baymard estimates $260 billion (US and EU) in losses recoverable through better checkout UX6.
Key Events in GA4
Transition from UA to GA4
Universal Analytics (UA) stopped processing data on July 1, 2023. UA had 4 goal types (destination, duration, pages/session, event) and a once-per-session counting model. GA4 completely changed this structure.
GA4 renamed “conversions” to “key events” in 20237. In GA4, only events can be marked as key events, providing a more flexible structure.
Counting Methods
GA4 offers two counting methods for key events8:
| Method | Behavior | Default for |
|---|---|---|
| Once per event (recommended) | Counts every time the event fires | New key events |
| Once per session | Counts only once per session | Key events migrated from UA |
If a user triggers a key event 5 times in the same session: once per event counts 5, once per session counts 1. This difference directly affects the Session key event rate metric.
Counting method changes apply to future data only, not retroactively8.
GA4 and Google Ads Integration
GA4 key events can be imported into Google Ads as conversions9. The process:
- Mark the event as a key event in GA4
- In Google Ads: Goals > Summary > Create conversion action > select GA4 property
- Conversions created from GA4 default to Secondary; switch to Primary for bidding use
Discrepancy causes: A 20-30% conversion difference between GA4 and Google Ads is normal10. Key reasons:
- Click-date vs event-date attribution: Google Ads attributes conversions to the click date; GA4 to the event date. A user who clicks on July 19 and purchases on July 20: Ads reports July 19, GA4 reports July 20.
- Counting method mismatch: Google Ads “Every” (all conversions per click) vs “One” (one conversion per click). May not align with GA4’s once per event.
- Cookie durations: Google Ads cookies last 90 days; GA4 cookies up to 2 years1112. Conversions after 90 days appear in GA4 but not Ads.
- Invalid click filtering: Some GA4 conversions are filtered when imported to Ads due to invalid click detection.
- Ad blocker effect: GA4 tracking script is blocked for approximately 31.5% of users globally. These conversions may reach Google Ads through Enhanced Conversions.
GA4 has a hidden Conversion Differences report. Access it by appending
/advertising/key-event-differences to the GA4 property URL. Not active on
all properties13.
Micro-conversion tracking in GA4 works in compliance with Consent Mode V22.
AI/LLM Traffic Impact on Conversions
AI-powered search and shopping experiences (ChatGPT, Google AI Overviews, Perplexity) are reshaping the conversion funnel.
LLM-Sourced Traffic
According to 13 months of data analysis (January 2025 - February 2026), LLM-sourced traffic accounts for under 2% of total referral traffic (range: 0.15%-1.5%). However, H1-H2 2025 comparison shows 80% average growth, with some companies seeing up to 300% increases14.
The notable finding is the conversion rate: LLM referral traffic converts at 18% CR, higher than paid shopping, SEO, and PPC. However, volume is 25x lower than SEO/direct14.
Attribution Problem
70.6% of LLM traffic is classified as direct in GA415. According to an analysis covering 181.6 million GA4 sessions:
- 22% of ChatGPT sessions fall into “(not set)” medium
- 32% of Perplexity sessions fall into “(not set)” medium
- Claude and Gemini: 100% correctly attributed as “referral”
ChatGPT only began appending utm_source=chatgpt.com to citation links in June 202516. Google AI Overviews and mobile app LLM referrals still send no attribution data. Actual AI-sourced traffic may be 2-3x higher than measured data suggests15.
Agentic Commerce and Funnel Collapse
AI agents making purchases on behalf of users (agentic commerce) are collapsing the traditional funnel structure. Protocols like Google’s Universal Commerce Protocol (UCP) are accelerating this shift17:
- Traditional funnel: impression > click > visit > cart > checkout > purchase
- Agent funnel: query > agent recommendation > one-step purchase
This collapse makes it difficult for merchants to calculate traditional conversion rates, run A/B tests on checkout pages, and execute cart abandonment recovery strategies.
According to Adobe’s 2025 holiday season data, users shopping with AI assistance showed 31% higher conversion rates and 254% more revenue per visit17. Ecommerce sites using conversational AI report 15-35% higher conversion rates18.
Factors Affecting Conversion Rate
- Technical problems: errors, slow loading, incorrect redirects, device incompatibility
- Tracking setup errors
- Payment methods: inadequate or incompatible options
- User experience issues: product pages, checkout steps, account management
- Strategic errors: target audience mismatch, pricing, lack of post-sale support
Conversion Rate Optimization (CRO)
CRO requires systematic methodology, not random A/B tests. CXL’s ResearchXL framework provides an industry-agnostic process: identifying site problems and converting them into testable hypotheses19.
The most common A/B testing mistake: ending tests too early. Statistical significance requires at least 2 full weeks and 1,000 conversions per variation20.
Footnotes
- Conversion marketing. Wikipedia ↩
- Micro Conversions in GA4. Analyzify ↩ ↩2
- Ecommerce Conversion Rate Benchmarks. Nector.io ↩ ↩2
- Ecommerce Conversion Rate by Industry 2026. Skailama ↩
- Ecommerce Conversion Rate Benchmarks 2026. Build Grow Scale ↩
- 50 Cart Abandonment Rate Statistics 2026. Baymard Institute ↩ ↩2
- GA4 Conversions vs Key Events. 2Point Agency ↩
- Change the counting method of key events. Google Analytics Help ↩ ↩2
- Create conversions from Google Analytics events in Google Ads. Google Ads Help ↩
- 8 Key Reasons for Data Discrepancies Between Google Ads and GA4. Dataslayer ↩
- [GA4] Cookie usage on websites. Google Analytics Help ↩
- Data retention. Google Analytics Help ↩
- GA4 Google Ads Discrepancies Report. Search Engine Land ↩
- Jason Tabeling. (2026). What 13 months of data reveals about LLM traffic growth and conversions. Search Engine Land ↩ ↩2
- The AI Referral Gap. Workshop Digital ↩ ↩2
- LLM Traffic Attribution. Alhena.ai ↩
- Agentic Commerce Trends and Statistics. MetaRouter ↩ ↩2
- The Future of AI in Ecommerce: 40 Statistics. HelloRep.ai ↩
- CRO Process. CXL (Peep Laja) ↩
- Conversion Rate Optimization Guide 2025. Convert ↩
- 01 Blended CR is misleading; actual performance is invisible without segmenting by traffic source and device
- 02 GA4 key events have two counting methods: once per event (recommended) and once per session (legacy)
- 03 A 20-30% conversion discrepancy between GA4 and Google Ads is normal, caused by click-date vs event-date attribution
- 04 70.6% of LLM-sourced traffic is classified as direct in GA4; actual AI traffic may be 2-3x higher than measured
- 05 Average cart abandonment rate is 70.22%; 48% is caused by unexpected additional costs (Baymard Institute)
- 06 The denominator in every conversion calculation should be the relevant user pool, not total site traffic
+ How is conversion rate calculated?
Conversion rate = (number of users who reached the goal / relevant user pool) x 100. The critical point: the denominator should be the number of users who reached the relevant step, not total site traffic. A form page's conversion rate should be calculated based on users who reached the form page, not all homepage viewers.
+ What is the difference between micro and macro conversions?
A macro conversion is the primary goal: purchase, lead form submission. Micro conversions are intermediate steps toward that goal: add to cart, newsletter signup, video view, 90% scroll. Micro conversions are used to diagnose funnel friction, not as primary KPIs.
+ What are the key event counting methods in GA4?
Two options: Once per event (recommended) counts every time the event fires. Once per session (legacy) counts only once per session. Key events migrated from UA default to once per session. The counting method directly affects the Session key event rate metric.
+ Why do GA4 and Google Ads conversion data differ?
Google Ads attributes conversions to the click date; GA4 attributes to the event date. Additionally, counting method differences, cookie durations (Ads 90 days, GA4 2 years), invalid click filtering, and ad blocker effects create discrepancies. A 20-30% difference is normal.
+ What is a good conversion rate?
It varies by industry, traffic source, and device. The ecommerce average is around 2.5-3%, but this is a blended number. Email traffic: 5%+, organic search: 3%+, paid social: 0.5-1.5%. Desktop: 3-4%, mobile: 1-2%. Comparison should be made against your own historical trend, not generic benchmarks.
+ How is AI/LLM traffic affecting conversion rates?
LLM-sourced traffic is under 2% of total traffic but converts at 18% CR, higher than traditional channels. The problem: 70.6% of this traffic is classified as direct in GA4. ChatGPT, Perplexity, and similar sources send attribution data inconsistently.