How to Set a Realistic Conversion Rate Target
A realistic conversion rate target is defined as a range built from your own twelve-month baseline plus only what the specific fixes you intend to ship can plausibly deliver. Record the median and the normal weekly spread for one segment, then state the target with a review date and a guardrail metric, usually revenue per visitor. The CROBenchmark dataset of 7,000+ websites [CROBenchmark Report 2026, Omniconvert] gives context but never the number itself. This guide shows the five rungs of the Target Ladder and what a target cannot do.
- Avoid industry averages, because they describe other stores' traffic and cannot set your target.
- Record your baseline's variance, or normal noise reads as failure.
- Segment by device and by new against returning visitors before setting any number.
- Size the target from the specific fixes planned, then discount for the ones that slip.
- Pair every target with a guardrail metric, or the rate can rise while revenue falls.
This guide sets out how to build a target from your own numbers, why it should be a range rather than a point, and what to measure alongside it so a rise in rate cannot disguise a fall in revenue. If the prior question is what counts as good in the first place, start with a good conversion rate.
What is a realistic conversion rate target?
Why borrowed benchmarks set the wrong conversion rate target
Consider two stores selling the same category at the same price. One gets 70% of its sessions from paid social on mobile, mostly first-time visitors. The other gets 55% from email and direct, mostly people who have bought before. The second will convert several times higher, and no amount of optimisation work at the first store closes that gap, because the gap is composition rather than quality.
The Target Ladder: five rungs to a number you can defend
A guardrail metric is defined as a second number you refuse to let fall while you chase the target, usually revenue per visitor. Baymard Institute puts average cart abandonment near 70% [Baymard Institute, 2025], a reminder that one rate rarely tells the whole story.
- Measure the baseline and its variance. Twelve months of weekly conversion rate, for one segment at a time. Record the median and the spread between the quietest and busiest weeks. Most stores discover their normal week-to-week swing is wider than the improvement they were about to promise, which is the single most useful thing this exercise produces.
- Segment before you set anything. Split by device, and by new against returning. A blended target can be hit by a change in device mix while both underlying rates stay flat, and it can be missed the same way. Two numbers are harder to fake and easier to act on.
- Size the fixes you intend to ship. Write down the specific changes planned for the period, attach an expected effect to each, then discount the total for the ones that will not ship on time. Some will not. A plan that assumes all four land is not a plan.
- State the target as a range, with a date. A band drawn from your own measured spread, with the review date fixed and the traffic mix it assumes written next to it. If acquisition changes the mix mid-quarter, the target moves and everyone knows why.
- Set the guardrail metric. Name what must not fall while the rate rises. Usually revenue per visitor, sometimes margin per visitor. Without it, a discount can deliver the target and cost you the quarter.
The ladder is ordered by dependency rather than by effort. The first rung is an afternoon with an analytics export and it determines whether anything above it means anything. The third rung is where most of the argument happens, because it is where a target stops being an aspiration and becomes a list of work somebody has to own.
How much movement is actually available
The table below sets out the situations we see most often and what each one can reasonably support in a quarter. The figures are relative improvements against the store's own baseline, not absolute percentage points, which is the distinction that causes the most confusion in target meetings.
| Store situation | Typical quarterly headroom | Where the movement comes from |
|---|---|---|
| Never audited, visible checkout friction | Large | A handful of named blockers, mostly in checkout and on mobile |
| Audited once, fixes partly shipped | Moderate | Completing the backlog that was already written |
| Running a structured test programme for a year | Small | Incremental wins, compounding rather than dramatic |
| Mature programme, mobile still untouched | Moderate, mobile only | The segment nobody set a separate target for |
| Traffic mix shifting toward paid and new visitors | Negative without work | Composition, not quality; the target must account for it |
| High returning-customer share, thin new-visitor experience | Moderate | First-visit trust and information gaps |
The last row of that table is the one worth reading twice. A store whose acquisition team is successfully buying more new visitors will see its blended conversion rate fall while nothing about the store gets worse. If the target was set without naming the traffic mix it assumed, that quarter ends in an argument nobody can win with the data they have.
What the target is for, and what it cannot do
It cannot replace the diagnosis. A number does not tell anyone what to change, and a team given a target without a list of findings will spend its first weeks producing one. Running the diagnosis first is the cheaper order, and where the funnel has not been examined at all, our companion guide on how to audit it covers the method step by step.
And it cannot protect revenue by itself. This is the failure that looks most like success. Conversion rate rises, the quarter is declared good, and revenue is flat because average order value fell by the amount the discount cost. Statista's eCommerce reporting shows how widely order values move between categories and periods, which is exactly why the rate cannot be read alone. Set revenue per visitor as the guardrail at the same moment you set the target, not after the first surprising month.
What to do this week
- Export twelve months of weekly conversion rate, split by device. Record the median and the gap between your quietest and busiest weeks. That gap is your noise floor and no target should be narrower than it.
- Run a free audit against your own funnel so the target has findings behind it. The CROBenchmark audit produces the list of named blockers that step three of the ladder needs.
- Write down the four fixes you will actually ship, with an owner and an expected effect for each, then cut the total by whatever proportion of your last quarter's plan did not land.
- State the target in one sentence: the range, the segment, the date, the traffic mix assumed, and the guardrail metric. If it does not fit in a sentence, it is not yet a target.
Where the findings need settling rather than debating, Omniconvert Explore is the CRO platform that pairs heatmaps, recordings and surveys with the A/B test that resolves them, averaging a 23.2% conversion uplift across 70,000+ experiments. Where the harder problem is deciding which finding to work first across a long backlog, Omniconvert Commander (formerly Nexus) is an AI for eCommerce growth engine that unifies commerce data, ranks experiments by True Profit, and generates campaigns and creative you approve before they go live.
Which signal to believe when the two disagree is the prior question, answered in benchmarks vs your own baseline.
Teams often set a round-number target, three percent because it sounds serious, and miss it in the first quiet week. The miss is blamed on the team, although the normal weekly spread was never measured, so noise was read as failure.
AliveCor used Omniconvert to run a structured A/B testing programme and achieved +21% conversion rate, +5% revenue per visitor, and 94% statistical relevance across their experiments. [Omniconvert, AliveCor case study] Pairing a target with a guardrail metric is how a structured programme avoids buying conversion with margin.
FAQ: conversion rate targets
Should I set my conversion rate target from an industry average?
No. An industry average describes a population of stores with different traffic mixes, price points, and proportions of returning customers, and none of those match yours. Use the average to sanity-check whether your baseline is unusual, then set the target from your own twelve-month baseline and the fixes you intend to ship.
How much can a conversion rate realistically improve in a quarter?
On a store with unresolved friction, a relative improvement in the low tens of percent is achievable in a quarter, and it comes from a handful of specific fixes rather than from general effort. On a store that has already run a serious programme for a year, the same quarter buys much less, because the cheap problems are gone. The honest answer depends on which of those two stores you are, and the audit tells you.
Why should the target be a range instead of a number?
Because weekly conversion rate varies for reasons nobody controls, and a single number turns normal variance into a monthly argument about whether the team is failing. A band drawn from your own measured spread absorbs that noise, so the review discusses the work rather than the weather.
Should mobile and desktop have separate targets?
Yes, and this is the most valuable split most stores are not making. Mobile and desktop convert at different rates for structural reasons, so a blended target can be hit by a shift in device mix while both underlying rates stay flat. Two targets make the improvement visible and stop traffic composition doing the work.
What should I do if we hit the target but revenue did not move?
Treat it as a failed quarter and check the guardrail. A conversion rate can rise because average order value fell, because a discount pulled forward demand, or because paid traffic was cut and the remaining visitors were better qualified. Revenue per visitor is the metric that catches all three, which is why it is set alongside the target rather than after it.
How often should the target be revisited?
Once a quarter, and on a date agreed before the quarter starts. Targets revised in the middle of a bad month tend to be revised downward for reasons that feel compelling at the time, and targets never revisited at all drift out of contact with the store as its traffic mix changes.
The bottom line
A conversion rate target is only as good as the arithmetic underneath it. Pull your own twelve months, record how much the number moves on its own, split it by device and by whether the visitor has bought before, and only then add what your named fixes can plausibly deliver. State the result as a range with a date, and set revenue per visitor beside it so a rise in rate cannot hide a fall in revenue. A target built that way tells the team what to build, survives a quiet week without a crisis meeting, and can be defended in a room where somebody asks where the number came from. A target copied from an industry average answers none of those, and the quarter it costs is not recoverable.