The Conversion Rate Formula, and Why the Blended Rate Lies

Conversion rate fell from 2.90 percent to 2.50 percent in the month the prospecting budget went up, and the brief that came out of the review asked for a landing page redesign. Both audiences on that site converted better that month than the month before. Returning visitors went from 5.0 to 5.5 percent, new prospecting traffic from 1.5 to 1.75 percent, and the business made $3,170.40 more contribution after ad spend. The headline number fell because the mix of visitors changed. Nobody converted worse.

The conversion rate formula is one division, and nobody gets the division wrong. The trouble is that it runs over a crowd, and the crowd changes shape every time the media plan does. This guide covers the conversion rate calculation itself, the funnel form that tells you where to work, and the check that separates a site problem from a traffic problem before anyone commissions a redesign.

The formula, and the two counts it divides

Conversions over visits, times a hundred. That is the whole of how to calculate conversion rate, and the conversion rate calculator does it in two fields. Its page also carries a table of how far a rate can move by chance, which on a few hundred conversions is further than most dashboards admit.

Conversion rate = Conversions / Visits x 100

Conversion rate = 580 orders / 20,000 sessions x 100 = 2.90%

Both counts need a definition that stays fixed. Conversions can mean orders, converting sessions, or every tracked event including newsletter sign-ups. Visits can mean sessions, users or ad clicks. The same 580 orders over 14,500 users is 4.00 percent, and neither figure is wrong. What is wrong is quoting the user-based number against last quarter's session-based one and calling the gap progress. Choose a pair, write it at the top of the report, and keep it.

That is as far as most explanations of the formula go. The useful part starts when you notice that the 2.90 percent is an average of several very different rates.

The funnel version: four rates multiplied

A purchase conversion rate is the product of every step between landing and paying. Split the same month into its steps and the single number turns into four.

Conversion rate = view rate x add-to-cart rate x checkout rate x completion rate

2.90% = 50% x 14.5% x 50% x 80%

Twenty thousand sessions produce ten thousand product views, 1,450 add-to-carts, 725 checkouts and 580 orders. Because the steps multiply, a ten percent relative lift at any single step lifts the whole rate by ten percent, to 3.19 percent. So far that argues for working on whichever step is cheapest to move.

Points are a different story. One extra percentage point at add-to-cart, 14.5 to 15.5, takes the overall rate to 3.10 percent. One extra point at checkout completion, 80 to 81, takes it to 2.94 percent. Same point, under a fifth of the effect. Teams that set targets in points rather than ratios tend to end up polishing the step that already works, because that is where a point feels achievable.

How a blended rate falls while every segment improves

Now split the same 20,000 sessions by audience instead of by step, and add the following month, when prospecting spend went up.

Returning visitors improved by ten percent. Prospecting improved by a sixth. The site as a whole fell by about 14 percent, from 2.90 to 2.50. Statisticians call this Simpson's paradox, and it stops looking like a paradox once you write the blended rate the way it is built.

Blended rate = (share of visits x segment rate), summed over segments

Month A: 0.40 x 5.00% + 0.60 x 1.50% = 2.90%
Month B: 0.20 x 5.50% + 0.80 x 1.75% = 2.50%

The blend is a weighted average and the weights moved. Prospecting went from 60 percent of visits to 80 percent, and it converts at roughly a third of the returning rate, so every extra prospecting session drags the average towards 1.75 no matter how well it performs against its own history. Run it backwards and the same arithmetic flatters a cut. Trim prospecting to 4,000 sessions at 1.50 percent and the blended rate climbs to 3.83 percent while orders fall from 580 to 460. A dashboard that celebrates that month is rewarding the business for shrinking.

Separate the rate from the mix before anyone reacts

The fix takes one line. Hold last month's shares fixed, apply this month's segment rates, and you get the rate the site would have posted if the traffic had not changed shape.

Mix-adjusted rate = 0.40 x 5.50% + 0.60 x 1.75% = 3.25%

Rate effect: 3.25% − 2.90% = +0.35 points
Mix effect: 2.50% − 3.25% = -0.75 points

The two effects sum to the 0.40 point fall on the dashboard. The rate effect belongs to whoever owns the site, and it was positive. The mix effect belongs to whoever owns the media plan, and it was a direct and expected consequence of buying more cold traffic. A redesign brief built on the headline would have spent money on a page that was improving.

Two segments catch most of this in practice. Choose the split whose rates differ the most, usually returning against new, or brand search against everything else. Splitting further thins the counts until the segment rates are noise, which is the same sample size problem the A/B test calculator exists to flag. A segment with thirty conversions a month does not have a conversion rate yet, only a rough range.

The rate that pays is the segment rate against break-even

Knowing that prospecting improved does not tell you whether it pays. For that, compare its own rate with the rate its click price requires. The argument is set out in full in what a good conversion rate is; the short form is below, using the site's standard example of an $86.00 order at 42 percent contribution, which is $36.12 an order. The AOV formula guide covers why that per-order figure should come from contribution rather than revenue.

Break-even conversion rate = CPC / Contribution per order

Break-even conversion rate = $0.60 / $36.12 = 1.66%

At a $0.60 click, prospecting was under water in Month A and above the line in Month B.

Add the returning visitors, who carry no media cost in this example, $14,448.00 of contribution in Month A and $15,892.80 in Month B, and contribution after ad spend goes from $13,749.60 to $16,920.00. That is 23 percent more money in the month the conversion rate fell by 14 percent. The cost per order tells the same story from the other side: $40.00 against a $36.12 ceiling, then $34.29 under it, which the CPA calculator will confirm from spend and orders alone.

One simplification carries this table. The click price is held at $0.60 in both months so the rate is the only thing moving. Real scaling usually raises the click price, and when it does, the break-even rate has to be checked on the extra clicks at their own price, not on the monthly average. The CPC calculator gives you that price from spend and clicks for any slice of the month.

What to do with this

The next time a conversion rate lands in a report, split it into two segments before sending it, compute the mix-adjusted rate at last month's shares, and print the rate effect and the mix effect beside the headline. If the rate effect is positive, the site is not the problem, whatever the headline says. Then take each paid segment on its own and set its rate against its click price divided by your contribution per order, which you can build in the contribution margin calculator. That comparison, not the blended figure, decides whether the budget moves.

Benchmarks and figures here are illustrative. They come from one worked example built so that every intermediate reproduces the printed result, not from a survey, and the only traffic that should set your targets is your own.

Related reading: what a good conversion rate is and the CPC your click-through rate commits you to and what the average order hides.