What “conversion” means
A conversion is whatever the business has decided counts as a successful outcome from marketing. For e-commerce, almost always a purchase. For B2B, usually a qualified lead or demo booking. For SaaS, a signup or paid upgrade. For apps, an install or in-app event.
Once defined, the conversion event becomes the optimisation target everywhere downstream:
- Meta, Google, TikTok algorithms learn to find more people likely to convert
- Attribution models distribute credit for the conversion across touchpoints
- ROAS = revenue per conversion × conversions ÷ spend
- CAC = spend ÷ conversions
The quality and completeness of conversion tracking determines the quality of every one of those.
Primary vs secondary conversions
Ad platforms usually let you designate a primary conversion event (the thing you actually optimise for) and several secondary events (signals you track for context).
- Primary, purchase. The algorithm bids for users likely to do this.
- Secondary, add to cart, view content, begin checkout. Used for retargeting audiences and signal enrichment, not direct optimisation.
Setting the wrong event as primary is a common mistake. Optimising for “add to cart” produces lots of cart abandoners. Optimising for “purchase” produces customers.
Conversion rate
The headline efficiency metric:
Conversion Rate = Conversions ÷ Traffic
Benchmarks vary wildly:
- E-commerce average: 2-3% across all traffic
- Top decile e-commerce: 5-8%
- B2B SaaS demo request: typically 1-3% of qualified traffic
- Lead-gen landing pages: 5-15% (single-purpose pages)
The right way to read conversion rate is per channel, per device, per cohort. Average conversion rate hides too much.
Conversion quality
A conversion is only useful if it represents real business value. Two failure modes:
- Fake conversions, bot clicks, test orders, mis-fired events. Pollute the optimisation signal.
- Low-value conversions, completed but refunded, abandoned post-conversion, fraudulent. Look like wins but cost money.
Modern ad platforms reward signal quality. A pipeline that reliably ships clean, high-value purchase events outperforms one that pushes 30% more events but with 30% noise.
What kills conversion tracking
The same forces that kill all tracking:
- ITP, ETP, ad blockers strip client-side events
- Pixel failures drop conversions silently
- Identity mismatches route conversions to the wrong user / journey
- Consent rejections prevent tracking entirely for opted-out users
Recovering this data needs server-side tracking, CAPI, and event ID deduplication. Pixel-only conversion tracking in 2026 misses 30-40% of actual conversions.
Common mistakes
- Optimising on secondary events. Get the primary event right. Secondaries are context.
- Reporting conversion rate without dimension. Average across all traffic is rarely actionable.
- Letting fake conversions in. Bot-driven add-to-carts inflate the funnel and confuse the algorithm.
FAQ about Conversion
What is a conversion in marketing?
A conversion is a tracked user action that signals business value, purchase, signup, lead, install. It is the optimisation target of every paid-media campaign and the primary input to attribution and ROAS reporting.
What is the difference between a primary and secondary conversion?
Primary conversions (typically purchase) are what ad platforms optimise their bidding around. Secondary conversions (add to cart, view content) are tracked for context and audience-building but are not the optimisation target.
Why are my reported conversions inflated?
Most often: bot traffic, test orders, or fired-but-not-completed events polluting the conversion signal. Use server-side filtering and only fire conversion events on confirmed purchases (post-payment-capture), not on order-attempt.