Use case · Conversion rate

Explain conversion rate change by product and traffic source

Overall conversion fell — but is your funnel actually worse, or did more low-intent traffic arrive? WaterfallBridge separates true performance from mix.

The question

"Our site-wide conversion rate dropped this month. Did the site get worse, or did the drop come from a shift in where our traffic came from? Which source is responsible?"

Why traditional methods struggle

Conversion rate is orders ÷ sessions — a ratio that can fall even when every source held steady, purely from mix.
Simpson's paradox is common: each source can improve while the blended rate declines.
A new traffic source (e.g. a referral or display push) brings low-intent sessions that dilute the average.
Manual analysis can't cleanly split per-source performance from traffic mix.

How WaterfallBridge breaks it down

Set the Result to Conversion Rate, with SumY = Orders (numerator) and SumN = Sessions (denominator), broken down by the Key = Traffic source (or Product). With last period as baseline and this period as comparison, WaterfallBridge isolates each source's real conversion change from the mix effect and the newly added source.

Example input data

Illustrative figures. Referral is a new source this period.

Traffic source	Period	Sessions (SumN)	Orders (SumY)	Conv. rate
Organic	Baseline	1,000	50	5.0%
Organic	Comparison	900	36	4.0%
Paid	Baseline	500	15	3.0%
Paid	Comparison	900	27	3.0%
Referral new	Comparison	300	6	2.0%
Total	Baseline	1,500	65	4.33%
Total	Comparison	2,100	69	3.29%

Sessions rose +600 and orders rose +4, but conversion fell −1.04 pp (4.33% → 3.29%).

Example output: the conversion bridge

Contribution of each source to the −1.04 pp conversion change (illustrative).

Driver	Contribution to conv. rate	Effect
Baseline conversion rate	4.33%	start
Organic — real drop (5.0% → 4.0%)	−0.65 pp	down
Paid — performance held (3.0%)	0.00 pp	flat
Referral — added (low 2.0% intent)	−0.30 pp	down
Mix shift (Paid grew vs Organic)	−0.09 pp	down
Comparison conversion rate	3.29%	end

The conclusion

The decline is part real, part mix. Organic genuinely worsened (5.0% → 4.0%, ~0.65 pp) — that's a real funnel/SEO issue worth investigating. The rest is dilution: the new low-intent Referral source cost ~0.30 pp and the shift of volume toward lower-converting Paid cost ~0.09 pp. So roughly two-thirds of the drop is a true Organic problem and one-third is mix. Fix Organic first; judge Referral on cost, not conversion.

"Conversion is down" can mean very different things. The bridge tells you how much is performance and how much is mix.

FAQ

How do I explain a conversion rate change by traffic source?

Define conversion rate as orders divided by sessions, set orders as SumY and sessions as SumN, and break it down by traffic source or product. WaterfallBridge separates the real per-source conversion change from the mix effect of where your traffic came from, and accounts for any new traffic source added during the period.

Explain your own conversion change

Try the demo with sample data, or ask an AI agent to analyze your file in one sentence.

Try with sample data → Ask AI to analyze your file →