Website monitoring

Conversion monitoring

On this page

Why conversion monitoring is critical
The two monitoring areas
Tools for conversion monitoring
How to monitor conversion step by step
Common conversion problems that monitoring catches
How to get the most out of conversion monitoring

A website can lose conversions even when its traffic is unchanged, so track conversion regardless of how the traffic is developing. Measure it both as a number and as a share of the visitors to the website.

Build reporting systems that immediately signal if the conversion metrics change or if a critical event occurs. A small drop in the conversion rate can cost enormous sums if it is not detected quickly. Technical problems in the checkout or in forms can stop 100 per cent of conversions overnight.

Conversion monitoring has two parts: monitoring results (are the conversion rate, volumes or traffic distribution changing?) and monitoring functionality (are the checkout, forms and payment working?).

Why conversion monitoring is critical

Small drops can cause large losses of revenue

Example

Scenario: an online shop with 10,000 visitors a week, a conversion rate of 2 per cent and an average order value of SEK 500.
Normal revenue: 10,000 × 2 per cent × SEK 500 = SEK 100,000 a week.
Problem: The conversion rate falls to 1.5 per cent (minus 25 per cent).
New revenue: 10,000 × 1.5 per cent × SEK 500 = SEK 75,000 a week.
Loss: SEK 25,000 a week (SEK 1.3 million a year).
With monitoring: The problem is detected on day 1 and fixed on day 2. Loss: SEK 3,500.
Without monitoring: The problem is detected in week 4 when the monthly figures are reviewed. Loss: SEK 100,000.

Technical problems can stop all conversions

Common problems

The checkout form does not work
The payment gateway is down
JavaScript errors that block forms
The “Complete purchase” button does not work

Without monitoring: Problems are discovered only when customers complain or when no sales have come in for several days.
With monitoring: An alert within minutes. The problem can be fixed before the revenue losses become large.

Segmented monitoring shows where the problems are

An aggregated conversion rate hides problems.

Example

Overall conversion rate: 2.5 per cent (down from 2.7 per cent). A quarter of visits are on mobile.

Broken down by device

Desktop: 3 per cent (unchanged)
Mobile: 1 per cent (down from 1.8 per cent, minus 44 per cent)

Insight: The mobile checkout has problems. Without segmentation, this is missed.

Proactive monitoring prevents escalation

Example

Day 1: The conversion rate falls by 20 per cent (an alert is sent).
Action: Investigate and fix.
Day 2: The conversion rate is back to normal.

Without monitoring

Days 1-7: The conversion rate continues to fall (now minus 30 per cent).
Day 8: The problem is discovered when the monthly figures are reviewed.
Day 9: The problem is fixed, but a large loss has already occurred.
Days 10-30: The sales that were lost cannot be made up.

The two monitoring areas

1. Monitoring internal results

Purpose: Monitor the conversion metrics and identify changes in behaviour.

The number of completed conversions has changed

What is monitored: The total volume of conversions per day, week and month.

Threshold

A 20 per cent change from one day to the next (for daily monitoring), a 15 per cent change from one week to the next (for weekly monitoring)

Example: “Conversions down from 50 a day to 30 a day (minus 40 per cent).”

Causes

Traffic has fallen (fewer visitors = fewer conversions)
The conversion rate has fallen (just as many visitors, but fewer convert)
A technical problem (people try to buy but cannot)

Actions

Check the traffic (has it fallen?)
Check the conversion rate (has it fallen?)
If the conversion rate has fallen: Investigate why

The number of completed conversions in sub-areas has changed

What is monitored: Conversions per product category, per landing page and per campaign.
Example: “Conversions from the ‘Electronics’ category down from 20 a day to 8 a day (minus 60 per cent), other categories unchanged.”

Causes

A problem specific to that category (broken product pages, products out of stock)
Traffic to the category has fallen

Actions: Investigate that category specifically.

Completed but failed conversions have occurred

What is monitored: Conversions that reach the thank-you page even though the payment failed.
Example: “Ten conversions reached the thank-you page but no payments were registered in the payment system.”

Causes

Problems with the payment gateway
A timeout between the checkout and the payment system
Data synchronisation errors

Actions: Check the payment gateway and investigate the technical fault.

The number of started conversions has changed

What is monitored: The number of users who start the checkout process.

Threshold

A 20 per cent change.

Example: “Checkouts started down from 100 a day to 60 a day (minus 40 per cent).”

Causes

Traffic has fallen
Fewer users reach the checkout (a problem earlier in the funnel)
The call to action (CTA) leading to the checkout is less visible or less appealing

Actions

Check the traffic
Check the traffic flow (are people reaching the product pages? Are they clicking “Add to basket”?)

The abandonment rate for started conversions is increasing

What is monitored: The share of users abandoning the checkout increases.
Formula: Abandonment rate = (Started minus Completed) / Started

Threshold

An increase in the abandonment rate of more than 10 percentage points.

Example: “Checkout abandonment up from 70 per cent to 85 per cent (plus 15 percentage points).”

Causes

The checkout process is too long or complicated
Unexpected costs (delivery, fees)
A technical problem (a form does not work)
Slow loading
A missing payment method

Actions

Test the checkout manually
Review heatmaps and session recordings
Identify where people drop out
Optimise

The conversion rate has changed

What is monitored: The conversion rate overall and per sub-area.
Formula: Conversion rate = Conversions / Visitors

Threshold

A 15 per cent change.

Example: “Conversion rate down from 2.5 per cent to 2.0 per cent (minus 20 per cent).”

Causes

The quality of the traffic has deteriorated (more irrelevant visitors)
The website or the offer is less appealing
Technical problems
A competitor has a better offer

Actions

Segment (which traffic source or which device has problems?)
Investigate specific segments

The share of started conversions relative to the number of visitors has changed

What is monitored: The share of visitors who start the checkout.
Formula: Started conversions / Visitors
Example: “Checkouts started down from 5 per cent of visitors to 3 per cent (minus 40 per cent).”

Causes

Fewer people click “Add to basket” (product pages less appealing)
Fewer people reach the checkout (a problem on the basket page)

Actions: Analyse the funnel from the product page via the basket to the checkout.

The share of started conversions from predefined key pages has changed

What is monitored: For important landing pages, how many people start the checkout from there?
Example: “Checkouts started from /campaign-page down from 10 per cent to 5 per cent (minus 50 per cent).”

Causes

The page has been updated and is now weaker
The CTA to the checkout is less clear
The offer is less appealing

Actions: Review the page and A/B test improvements.

The number of completed conversions from a traffic source has changed

What is monitored: Conversions per traffic source (organic, paid, direct, social, email).
Example: “Conversions from Google Ads down from 20 a day to 8 a day (minus 60 per cent), other sources unchanged.”

Causes

Traffic from the source has fallen (campaign paused, budget used up)
The conversion rate from the source has fallen (weaker targeting, the wrong landing page)

Actions

Check the traffic from the source
Check the conversion rate from the source
Optimise the campaign/landing page

The number of started conversions from a traffic source has changed

What is monitored: Checkouts started per traffic source.
Example: “Checkouts started from Facebook down from 30 a day to 12 a day (minus 60 per cent).”

Causes

Traffic has fallen
The quality of the traffic has deteriorated (the wrong audience)

Actions: Review and optimise the campaign.

The share of completed conversions from a traffic source has changed

What is monitored: The conversion rate per traffic source.
Example: “Conversion rate for email down from 5 per cent to 2.5 per cent (minus 50 per cent).”

Causes

The email campaign is reaching the wrong segment
The landing page does not match the email message
The offer is no longer relevant

Actions: Analyse the email campaign and the landing page.

The share of started conversions from a traffic source has changed

What is monitored: The share of visitors from a traffic source who start the checkout.
Example: “Checkouts started from organic traffic down from 6 per cent to 3 per cent (minus 50 per cent).”

Causes

Organic traffic is reaching the wrong pages (rankings for irrelevant search phrases)
Pages are less appealing

Actions: SEO analysis, optimise the pages.

The number of completed conversions per device has changed

What is monitored: Conversions per device (desktop, mobile, tablet).
Example: “Conversions from mobile down from 30 a day to 15 a day (minus 50 per cent), desktop unchanged.”

Causes

Mobile traffic has fallen
The mobile conversion rate has fallen (problems with the mobile checkout)

Actions: Test the checkout on mobile and optimise it.

The number of started conversions per device has changed

What is monitored: Checkouts started per device.
Example: “Checkouts started from mobile down from 50 a day to 25 a day (minus 50 per cent).”

Causes

Mobile traffic has fallen
Fewer people reach the checkout on mobile (a problem in the funnel)

Actions: Analyse the mobile funnel.

The share of completed conversions per device has changed

What is monitored: The conversion rate per device.
Example: “Mobile conversion rate down from 1.5 per cent to 0.8 per cent (minus 47 per cent), desktop unchanged at 3 per cent.”

Causes

Problems in the mobile checkout (design, loading, functionality)

Actions: Prioritise mobile optimisation.

The share of started conversions per device has changed

What is monitored: The share of visitors per device who start the checkout.
Example: “Checkouts started by mobile visitors down from 4 per cent to 2 per cent (minus 50 per cent).”

Causes

The mobile pages are less appealing
The CTA to the checkout is less visible on mobile

Actions: Optimise the mobile experience before the checkout.

2. Monitoring internal events

Purpose: Monitor technical problems that prevent conversions.

Parts of the conversion chain are not working

What is monitored: Specific steps in the checkout process are not working.
Example: “Step 3 of the checkout (choice of payment method) returns a 500 error.”
Impact: 0 per cent can complete the checkout from this step.
Actions: Fix the fault immediately. Critical.

The checkout is not working

What is monitored: The entire checkout process is down.
Impact: 0 per cent convert. Revenue stops completely.
Actions: Top priority. All resources are deployed.

The connection to the payment provider is not working

What is monitored: The integration with the payment gateway (Stripe, Klarna and others) is broken.
Impact: Users can fill in the checkout but the payment fails.
Example: “Payment gateway timeout. Users get an error message when paying.”

Actions

Contact the payment provider (are their systems up?)
Check your own logs (are there errors in the integration?)
Roll back the latest deployment if it caused the problem

Out-of-stock items can be selected for purchase

What is monitored: Products with no stock still appear as available to buy.
Impact: Users add the item to their basket, reach the checkout and get an error message saying it is unavailable. They get frustrated and drop out.
Example: “Product X out of stock but the ‘Buy’ button is still active.”

Actions

Sync the stock status correctly
Hide the “Buy” button for sold-out items
Show “Out of stock” clearly

Basket functions are broken

What is monitored: It is not possible to update the quantity, remove a product or enter a discount code.
Impact: Users get stuck in the basket and cannot continue to the checkout.
Example: “The discount code field does not accept codes. Users get an error message.”
Actions: Test the basket manually and fix the faults.

Pages or functions are broken on certain devices

What is monitored: The checkout works on desktop but not on mobile (or vice versa).
Impact: If the mobile checkout is broken, the purchases from all mobile traffic are lost.
Example: “The checkout button on mobile is not clickable (a CSS problem).”
Actions: Test on all devices and prioritise fixing the fault on mobile.

Tools for conversion monitoring

Analytics and alerts

Google Analytics 4

Custom insights for changes in the conversion rate (evaluated hourly, daily, weekly or monthly)
Real-time conversion tracking
Funnel analysis

Adobe Analytics

Enterprise-level conversion tracking
Advanced segmentation

Error tracking

Sentry

JavaScript errors in real time
Alerts for errors in the checkout
Stack traces for debugging

Rollbar

Back-end error tracking
Alerts for API errors (payment gateway and others)

LogRocket

Session replay with error tracking
See exactly what users did when the error occurred

Funnel and session tracking

Privacy: Session recording requires consent, and fields containing personal data and payment details must be masked. Microsoft Clarity requires a consent signal for visits from the EEA, the UK and Switzerland.

Contentsquare (formerly Hotjar)

Funnel visualisation
Session recordings (see where people drop out)
Heatmaps

Microsoft Clarity

Free session recordings
Rage clicks (users clicking many times in frustration)
Dead clicks (users clicking on non-clickable elements)

Uptime and synthetic monitoring

Pingdom

Transaction monitoring (simulates the checkout process every five minutes)
Alerts if the transaction fails

Datadog Synthetic Monitoring

Simulate user flows
Alerts if the checkout flow breaks

A/B testing and CRO

Tests in existing systems

Modern websites are complex, so external testing tools can rarely be relied on
A test of a checkout change is usually built into the company’s existing systems

Separate analysis

Data from the test is extracted for analysis outside the test environment
Funnel analysis

Payment gateway monitoring

Stripe Dashboard

Payment success rate
Failed payment reasons

Klarna, PayPal and others

Their own dashboards with success rates

How to monitor conversion step by step

Step 1: Define critical conversion goals

What counts as a conversion?

E-commerce

A completed purchase

Lead generation

A form filled in
A demo booked
A PDF downloaded

SaaS

Registration
A trial started
A subscription activated

B2B

Contact form
Quote request

Step 2: Set up conversion tracking

Google Analytics 4

Key events configured
E-commerce tracking (for online shops)
Event tracking for micro-conversions

Google Tag Manager

Triggers for conversions
Custom events

Step 3: Configure alerts

Google Analytics custom insights

The conversion rate falls by more than 15 per cent from one day to the next
Conversions fall by more than 20 per cent from one day to the next

Custom alerts (data via the GA4 API, with the alerting logic built separately)

The checkout abandonment rate increases by more than 10 percentage points
More flexible thresholds
Segmented alerts (per device, per traffic source)

Step 4: Set up error tracking

Sentry

Install in the front end (JavaScript)
Install in the back end
Alerts for errors in checkout-related code

Alert configuration

An immediate alert at more than ten errors per minute in the checkout
A daily summary of all errors

Step 5: Implement synthetic monitoring

Pingdom Transaction Monitoring

Create a script that simulates the checkout
Run it every five minutes
Alerts if the transaction fails

Example script

Go to a product page
Click “Add to basket”
Go to the basket
Click “Go to checkout”
Fill in the checkout form (test data)
Click “Complete purchase”
Verify that the thank-you page is displayed

If any step fails: Alert.

Step 6: Review daily

Daily routine (five minutes)

Check Google Analytics: Conversions yesterday compared with last week
Check the error tracking: Any new errors?
Check the synthetic monitoring: Any failed transactions?

If anything looks odd, investigate it immediately.

Step 7: Weekly in-depth analysis

Weekly routine (30 minutes)

The trend in the conversion rate (the past four weeks)
Funnel analysis (where do most people drop out?)
Segmented analysis (device, traffic source)
Session recordings (watch a few abandoned purchases, with consent and with personal data masked)

Identify patterns and optimisation opportunities.

Common conversion problems that monitoring catches

Problem 1: The checkout button is not clickable on mobile

Alert: “Synthetic monitoring: Checkout transaction failed on mobile.”
Cause: A CSS error after the latest deployment.
Detected: ten minutes after the deployment with monitoring. Days later without monitoring, when customers complain.
Action: Roll back the deployment, fix the CSS and deploy again.
Impact with monitoring: ten minutes of mobile checkout downtime.
Impact without monitoring: Three days of downtime and a loss of SEK 100,000.

Problem 2: The conversion rate falls by 24 per cent for no obvious reason

Alert: “Conversion rate down from 2.5 per cent to 1.9 per cent (minus 24 per cent).”

Investigation

Traffic unchanged
Segment: Desktop fine, mobile conversion rate down from 1.8 per cent to 0.9 per cent (minus 50 per cent)
Test the mobile checkout: Slow loading (five seconds or more)
Check recent changes: New high-resolution images uploaded

Action: Optimise the images and add lazy loading.
Impact with monitoring: Detected on day 1, fixed on day 2. Minimal loss.
Impact without monitoring: Detected in week 4, a large loss.

Problem 3: Payment gateway timeout

Alert: “Sentry: 50 errors per minute in payment processing.”
Cause: The payment gateway (Stripe) is having temporary problems.

Action

Contact Stripe (they confirm a problem on their side)
Show users a message: “Payment temporarily unavailable, please try again in 10 minutes”
When Stripe has fixed the fault: Remove the message

Impact with monitoring: Users are told that the problem is temporary.
Impact without monitoring: Users think the checkout is broken, give up and buy from a competitor.

How to get the most out of conversion monitoring

Always segment

Aggregated data hides problems.

Segment by

Device (desktop, mobile, tablet)
Traffic source (organic, paid, direct, social, email)
New or returning visitors
Geography (if relevant)

Prioritise critical alerts

Not all alerts are equally important.

Critical (immediate action)

The checkout completely down
The payment gateway down
Zero conversions for more than two hours

Important (action within 24 hours)

The conversion rate falls by more than 15 per cent
Checkout abandonment increases by more than 10 percentage points

Informational (action within a week)

The conversion rate falls by 5 to 15 per cent
Minor errors in the checkout (affecting fewer than 5 per cent of users)

Test the checkout regularly

Even without alerts, test manually

Desktop checkout (every week)
Mobile checkout (every week)
Different payment methods (every two weeks)

Catch problems that synthetic monitoring misses.

Document all problems and actions

Problem log

Date
Problem (what happened?)
How the problem was detected (an alert? Manually?)
Cause
Action
Time to fix
Impact (estimated loss)

Use

Learn from past problems
Identify recurring problems
Justify investment in monitoring (show how much money has been saved)

Compare with deployments

When problems arise, check

Was there a recent deployment?
Did anything change on the website?

If so: That is probably the cause. Roll back and investigate.

How to go from alert to action when conversion falls

Process

Alert sent → “Conversion rate falls by 25 per cent”
Verify → Is it accurate? (Check Google Analytics manually)
Segment → Which segment has the problem? (Device, traffic source)
Diagnose → What is the cause? (Test manually, check the error logs, check the latest changes)
Fix → Fix the problem
Monitor → Is the conversion rate recovering?
Document → Log the problem and the solution

Contact us about conversion monitoring

Carnaby helps companies set up robust conversion monitoring, from alerts and error tracking to synthetic monitoring, so that problems in the checkout and conversion flows are detected and fixed before revenue is lost.

Want to know where you stand today?

Call us, or fill in the form and we will get back to you.