Benchmarks vs Your Own Baseline: Which Should You Trust?

Benchmarks and baselines get treated as two answers to one question, and they are two answers to different questions. A benchmark tells you where you sit relative to other stores. A baseline tells you what your own store did before you changed something. Only the second can tell you whether an action worked, which is why the two disagreeing is far less troubling than it feels. Last updated: September 2026.

Omniconvert has measured conversion behaviour across the CROBenchmark dataset of 7,000+ websites in 15+ industries, against 248+ audit criteria, over 13 years in eCommerce, and has run Omniconvert Explore across 70,000+ experiments. The single most consistent finding in that data is unhelpful to anyone selling a benchmark: the spread within any category is far wider than the gap between categories, and most of that spread is explained by traffic mix rather than by how well a store is built.

That one fact settles most of the argument, and the rest of this article works through what follows from it. Below: what each signal actually measures, why category averages are so wide, the four cases where a benchmark wins, how to build a baseline you can defend, and what to do when the two point in opposite directions. If you want the category ceiling first, our guide to a good conversion rate covers it.

What each signal actually measures

Think about what has to be true for a comparison to support a decision. You need the two things being compared to differ in only the way you care about. A baseline satisfies that approximately: same store, same catalogue, same customers, different week. A benchmark satisfies it not at all, because the comparison store differs in every respect at once, including several you cannot observe.

This is why a benchmark can be interesting and still not actionable. Learning that your conversion rate sits below a category average tells you that a difference exists. It cannot tell you whether the difference is your checkout, your prices, your product range, your delivery proposition or simply that your traffic skews toward people earlier in a decision than theirs does.

The baseline has the opposite properties. It is narrow, it says nothing about whether you are good, and it is the only thing that can tell you whether last month's change helped. For a team whose job is to improve something, that trade is not close.

Why category averages are so wide

Traffic mix is the dominant variable and it is invisible in every published benchmark. Returning visitors convert at multiples of first-time visitors. Branded search converts far above cold prospecting. Email to an engaged list converts above almost everything. A store's blended conversion rate is therefore mostly a statement about where its visitors came from.

Two consequences follow. First, a store can raise its conversion rate by buying less top-of-funnel traffic, which improves the metric and may reduce the business. Second, a store scaling acquisition successfully will usually see conversion rate fall, because it is adding visitors who are earlier in their decision. Judged against a benchmark, growth looks like decline.

Price point compounds it. Higher-priced goods convert lower and consideration periods run longer, so a category average that mixes price tiers is describing a distribution rather than a typical store. Baymard Institute's checkout research, which has documented average cart abandonment near seventy percent along with its recurring causes across years of testing [Baymard Institute], is more useful than a conversion average precisely because it reports mechanisms rather than a single figure.

The four cases where a benchmark legitimately wins

The last row is the one that costs real money. A store scaling paid acquisition will reliably see blended conversion fall while total revenue rises, and a team benchmarking blended conversion rate will read that as deterioration. The recommended remedy, spend less on cold traffic, would work on the metric and damage the business.

The third row is the practical compromise, and it is how the two signals are meant to coexist. Your baseline says what is reachable from here. The benchmark says roughly where the ceiling in your category sits, which stops a target being set above anything anyone has achieved. Neither is doing the other's job.

Building a baseline you can defend

Start with the period. It has to be long enough to contain a full purchase cycle for your category, including the lag between first visit and order, or you will be comparing a period that captured its own demand against one that inherited demand from before it.

Then split it, because a blended figure hides the two variables that move it most. Device matters because mobile and desktop routinely differ by a factor that swamps most changes anyone ships. Source matters for the reasons above. A baseline that is one number per device per major source is harder to argue with and much harder to accidentally misread during a mix shift.

Clean it next. Internal traffic, bot traffic and any test or staging activity all inflate sessions and depress conversion. Then reconcile: the conversion event should agree with your order table within a percent or two, and where it does not, everything built on it is unreliable. This is the same first pass any competent audit runs, and our sibling guide on how to audit it covers the sequence in full.

Finally, state it as a range. Weekly conversion rates bounce for reasons nobody controls, and a baseline expressed as a single decimal invites a conversation about whether this week beat it, which is noise. A range with a review date is honest about the variance and much more useful to plan against.

When they disagree, and what to do

The disagreement people find hardest is sitting comfortably at or above a category average while revenue softens. The temptation is reassurance, and it is misplaced. An average calculated across other businesses cannot see your pricing, your competitors' new proposition, or a change in what your traffic is worth. Your own declining trend is direct evidence about your store, and the benchmark is indirect evidence about other people's.

The opposite case is easier and gets mishandled more often. Sitting below a category average while improving steadily is a healthy position, and it frequently triggers an unnecessary strategic response. Before acting on the gap, check whether your traffic mix explains it, since a store with a higher share of cold prospecting should sit below a mixed average and is not underperforming by doing so.

Bain and Company's retention work with Fred Reichheld holds that a five percent improvement in retention can raise profits by twenty-five to ninety-five percent [Bain and Company], which is a reminder that the number worth improving is often not the one being benchmarked. Where the diagnosis needs to reach past the on-site funnel, Nexus by Omniconvert is an AI for eCommerce growth engine that unifies commerce data, prioritises experiments by True Profit and generates campaigns you approve before they go live, and Omniconvert Explore covers the testing, surveys and recordings that turn a gap into a tested hypothesis.

For the statistical reasons underneath all of this, we set out why industry averages mislead and what they are still good for.

FAQ: benchmarks versus baselines

Should I trust an industry benchmark or my own baseline?

Your own baseline, in almost every case where the two disagree, because it is the only one measured on your traffic, your catalogue and your prices. A benchmark is drawn from stores whose traffic mix you cannot see, and traffic mix drives conversion rate more than most on-site factors do. Use the benchmark to ask a question and the baseline to answer it.

What is an industry benchmark actually good for?

Two things. It tells you whether a number is unusual enough to investigate, and it gives you a rough sense of the ceiling in your category so you do not chase an impossible target. Both are framing uses. Neither justifies a decision on its own, and a benchmark should never be the reason a specific change is made.

When is a benchmark better than a baseline?

When you do not have a usable baseline. A new store, a new market, a new category or a business whose traffic mix has changed fundamentally all lack comparable history, and in those cases a benchmark is the only external reference available. It is a temporary substitute that should be retired as soon as three or four clean months exist.

Why do industry conversion benchmarks vary so widely?

Because they aggregate stores with different traffic mixes, price points, catalogue sizes and definitions of a conversion. A store with mostly returning email traffic and a store with mostly cold paid traffic can differ several times over on conversion rate while both being run competently, and an average across the two describes neither.

How do I build a baseline I can defend?

Take a recent period long enough to cover a full purchase cycle, split it by device and by traffic source, exclude internal and bot traffic, and confirm the conversion event reconciles with your order table. Then state it as a range rather than a single figure, because a baseline expressed as one number invites comparisons it cannot support.

What should I do when the benchmark says I am fine but revenue is falling?

Believe the revenue and ignore the benchmark. Sitting at a category average while your own trend declines means something changed in your traffic, your pricing or your competition, and an average calculated across other businesses cannot see any of it. A benchmark can only ever tell you where you sit, never what moved.