How do you benchmark an online store's conversion funnel?
Short answer: Benchmarking a webshop funnel means recording each step's current conversion rate, from a fixed data source over a fixed window, before anything is changed — then comparing each step against a benchmark for that category and traffic source, not a single blended average. Six steps, roughly an afternoon, and it has to happen first.
Why does this come before the writing, not after?
Benchmark before producing content because a baseline recorded afterwards is not a baseline. Without a before figure, weak copy, a broken page and unqualified traffic all produce the same flat result and cannot be told apart. The measurement is what converts content production from a recurring expense into something that can be judged. A benchmark also tells you which step not to work on: a step already at its category benchmark cannot be moved by better copy, because copy closes the distance to a benchmark rather than pushing past one.
What counts as a conversion?
A conversion is the specific action being counted, agreed explicitly before measurement starts — a completed purchase, a quote request, a subscription start, a booked call. It should never be assumed to be "purchase" by default. Two rules prevent most of the damage: a lead is not a conversion (counting a B2B quote request as one inflates the rate and hides where quotes fail to become orders), and decide once whether the count is gross or net of cancellations and refunds, then hold that convention for every future measurement.
Blended vs per-step · illustrative
A healthy blended rate can hide a leaking step
The blended figure sits inside the general 2–3% e-commerce range and looks fine. Step by step, the cart→checkout drop is where the money leaks — invisible until you measure each step separately. Rates shown are illustrative.
Sessions, users or interactions — which denominator?
Take the figures from data, not from memory: analytics, the shop platform or the payment processor, never the founder's sense of how things are going. It matters which denominator you use, because sessions, users and interactions are three different denominators and produce three different conversion rates from identical data. A platform's reported rate is often per ad interaction rather than per visitor, which is not comparable with a site-wide session-based figure. Record which denominator is in use alongside the number itself.
Should you benchmark blended rate or per step?
Benchmark each funnel step separately. A blended site-wide rate can sit at the category average while one step leaks badly, because a strong step compensates for a weak one in the blend. Step-level measurement is what makes a leak visible; a blended figure is what keeps it hidden. Record each step with the volume entering it, not only the percentage — the volume is what turns a percentage-point gap into money later.
What should you compare your rate against?
Compare each step against a benchmark for the same category and the same traffic source, not a global average. Standard web e-commerce converts at roughly 2% to 3% overall, but paid social and paid search convert at materially different rates, and category variation is wider than the gap between a good shop and a poor one. Where no category benchmark exists, label the comparison as a general e-commerce average rather than presenting it as category-specific — a clearly-labelled assumption is usable, an unlabelled one is a landmine for the next reader.
How long should the baseline window be?
Use a window long enough to cover normal weekly variation and at least one full purchase cycle — 90 days is the common default. Record the exact date range alongside the figures, and use the same length for the follow-up so the two are comparable. And you cannot set a valid baseline after changes have begun: a baseline recorded after the fact measures the changed state, not the original. If changes are already live, the honest position is to record current figures as a new baseline, state plainly that no valid before-measurement exists, and judge the next round rather than this one.
What should you write down so the baseline survives?
The baseline needs to outlive the conversation it was measured in. Record, per step, six fields: the rate, the volume entering, the date range, the data source, the conversion definition, and the denominator. A baseline missing any one of them cannot be safely compared against in eight weeks, because nobody will remember which convention was used.
What benchmarking does not tell you
A benchmark identifies distance from a norm; it does not identify cause, and it does not by itself justify work. A rate measured across a seasonal peak is not comparable with one measured outside it — compare against the same period last year, or state that the read is contaminated. And a change on a small number of orders is usually noise: below roughly a few hundred conversions in the window, ordinary week-to-week variation will exceed most real improvements, so judge direction over several periods rather than declaring a result from one.
FAQ
How do you benchmark a webshop's conversion funnel?
Record each step's current conversion rate, from a fixed data source over a fixed window, before anything is changed — then compare each step against a benchmark for that category and traffic source rather than a single blended average. Six steps, roughly an afternoon, and it has to happen first.
Why benchmark before producing content rather than after?
Because a baseline recorded afterwards is not a baseline. Without a before figure, weak copy, a broken page and unqualified traffic all produce the same flat result and cannot be told apart. The measurement is what converts content production from a recurring expense into something that can be judged.
Should you benchmark blended conversion rate or per step?
Per step. A blended site-wide rate can sit at the category average while one step leaks badly, because a strong step compensates for a weak one in the blend. Step-level measurement is what makes a leak visible; a blended figure is what keeps it hidden.
What should you compare your conversion rate against?
A benchmark for the same category and the same traffic source, not a global average. Standard web e-commerce converts at roughly 2% to 3% overall, but paid social and paid search convert at materially different rates, and category variation is wide. Label any general figure as a general average.
Can you set a baseline after you have already made changes?
No. A baseline recorded after changes have begun measures the changed state, not the original one, and any improvement calculated from it is unprovable. Record the current figures as a new baseline, state that no valid before-measurement exists, and judge the next round.
How do you benchmark a B2B webshop?
Run the identical measurement with two changes: count a quote request as a lead rather than a conversion, and use reorder rate in place of repeat purchase rate. Because B2B traffic is low-volume and high-value, conversion percentages are noisy and should be reported as such.
Method applies to B2C and B2B webshops with the conversion-definition change noted above. General e-commerce conversion range (~2–3%) reported as a broad average from published benchmarks; category and traffic-source figures vary widely and should be sourced per case. Figures as of August 2026.
