A good conversion rate counts real enquiries or sales. It holds up against your own last comparable period and brings in work worth having. Anyone quoting one healthy percentage for every website is guessing.
Below are benchmark pitfalls, our band for paid service enquiries, lead cost maths and what to fix first.
What does conversion rate actually measure?
Start with the completed action, then identify the unit that was eligible to perform it. That's the whole definition. Plenty of arguments about conversion rate are two people using different units.
For a session based rate, count the sessions with at least one selected outcome. Then divide by the eligible sessions. Dividing every event occurrence by sessions is different, because one session can fire the same event three times.
The example is made up. The definition is one session based measure, not every platform metric that carries the words conversion rate. The platforms differ in ways that matter when you compare.
Website and session rates
A website enquiry rate needs two decisions written down. Whether the unit is a session or a user. Whether the outcome is an accepted form, an answered call or a confirmed booking. A button click is an intermediate action.
In GA4, the Traffic acquisition report defines session key event rate. That rate is the percentage of sessions in which a user triggered a key event. The user version of the rate uses a different denominator. So specify which key event, which report and which filters.
Making that key event mean an accepted enquiry is a setup job. Our guide to GA4 for small business covers it.
Landing page goals
A landing page rate belongs to a page cohort, a platform and a configured goal. A signup page and a high value sales enquiry page are not interchangeable. Vendors do not all count unique people, sessions and repeated conversions the same way. When a builder reports a rate, find out what it divided by what before you repeat it.
Online store purchases
Shopify's behaviour reports define online store conversion rate as the percentage of sessions that resulted in a purchase. Each funnel step is calculated as sessions for that step divided by all sessions.
The same reports have open and closed funnel settings. The open funnel counts a visitor who reached checkout without adding to cart. The closed funnel counts only sessions that completed every step in order. So the same store can show two different step percentages.
If you calculate checkout completion among checkout starters instead, label that denominator. It's not the all sessions figure. Refunds and margin need their own commercial checks. Nothing in a conversion rate shows either.
Ad interaction rates
Google Ads calculates conversion rate as conversions divided by ad interactions. The denominator counts only interactions that could be tracked to a conversion in the same period. The same page explains that counting more than one conversion action can push the rate above 100%. So can counting every conversion rather than one per interaction.
That doesn't mean more than every visitor bought something. It means the numerator counts things the denominator doesn't. Each conversion action's Count setting decides how many. The One option counts a single conversion per ad click and suits a lead. The Every option counts each conversion after an ad interaction and suits a sale.
Our guide to the Google Ads metrics that matter covers which of the account's numbers to read first.
Lead to customer close rate
A close rate is customers divided by a defined lead cohort, with enough time for sales to close. It belongs after the website enquiry, never in the same benchmark column as a session rate. The common mistake is multiplying a website rate by another period's close rate and calling it a forecast.
Whatever the platform, validate the accepted outcome before judging it. A call click is not proof the call connected. A booking tool visit is not a completed booking. A form submit event can fire on a failed attempt.
Remove duplicates and spam. Distinguish channel actions from unique leads when the same person calls and submits a form. Check the recording itself. Our guides to Google Ads conversion tracking setup and call tracking for Google Ads show how.
Rate | Numerator | Denominator |
|---|---|---|
Website enquiry rate | Sessions or users with an accepted enquiry | Eligible sessions or users |
GA4 session key event rate | Sessions containing the selected key event | All sessions in the report scope |
Landing page rate | The platform's configured goal completions | The page cohort, as the platform counts it |
Online store purchase rate | Sessions with a purchase | All sessions |
Checkout completion rate | Sessions that completed checkout | Sessions that started checkout |
Google Ads conversion rate | Recorded conversions, as counted in the account | Eligible ad interactions |
Lead to customer close rate | Customers | A defined lead cohort, after closing time |
Which benchmarks are useful for your website?
Only the ones whose population, outcome and denominator you can name. Many published figures come from one vendor's own customers, with its own goals and counting rules. That makes them context at best. None is a target.
There's no universal figure for a whole website either. Visits have different tasks. So define the cohort and the action first, then look for a study that matches them. Check any figure like this before you use it:
Source and report date: [named provider and publication]
Population: [page type, industry, platform and sample]
Window: [when the data was collected, if stated]
Outcome and denominator: [what is counted and divided]
Summary: [median, mean or pooled rate]
Fit: [how this differs from your business]
A common mistake is setting two benchmarks side by side when their methods differ. A store's purchase rate beside a landing page median compares a purchase with a form fill. It would look as if one kind of business simply performed better.
Three words decide what a benchmark figure means. A median is the middle reported rate: half the sample sat above it and half below. It promises nothing about yours. A mean, the average of the businesses' rates, can be dragged by a few outliers. A pooled rate, all outcomes divided by all opportunities across the sample, weights big sites more heavily.
Do not average channel percentages unless you weight them by the correct base. 10% of 50 visits and 2% of 5,000 visits do not average to 6% of anything.
Check whether the sample is Australian before you lean on it. Calling global vendor data "Australian benchmarks" doesn't make it one. Several things affect comparability:
country
acquisition mix
offer
price
how well customers already know you
how hard the conversion is
So we use our own rates for one common case: service lead generation from high intent paid traffic. We treat 5% to 6% of clicks becoming enquiries as standard. With enough conversion history and a landing page built for the campaign, 8% to 9% is realistic.
Below 3% is worth investigating. Large project trades and finance can sit lower. Those rates are our account experience, not a published benchmark.
SourceElev8d account experience, not a published benchmark
Use the band to decide where to look first. A rate inside the band still needs its counts, lead quality and cost per qualified enquiry checked. Once you have your own comparable history, that beats any band, ours included.
How do you compare rates fairly?
Use these seven checks to compare like with like. Make the comparison useful before you make it impressive. If the first two fail, the other five don't matter.
Same outcome and unit
Purchase with purchase, accepted lead with accepted lead, session with session
Comparable visitors
Same channel, device, intent, service, geography and new or returning mix
Counts as well as rates
4 from 50 and 400 from 5,000 are both 8%. State the counts
Consistent reporting
Duplicates, internal traffic, spam, bot filtering, consent, missing events and handoffs
Enough outcome time
Allow for purchase or qualification delay before comparing cohorts
Traffic mix and seasonality
More low intent visits lower the overall rate with no segment getting worse
Percentage points and relative uplift
From 4% to 6% is 2 percentage points or a 50% relative increase. Say which
Keep the platform's counting rules visible when you match outcomes. Do not slice a small sample into fragments to find a comparable segment. Note promotions, stock, prices and seasons beside the traffic mix, because they move the rate the same way.
The counts matter because the same 8% carries very different certainty at the usual 95% confidence level. A result of 4 from 50 is consistent with a true rate anywhere from about 3% to 19%. But 400 from 5,000 pins it to about 7% to 9%.
Measurement changes break comparisons without touching customers. Shopify's session measurement update rolled out from 21 to 23 September 2026. It changed how Shopify's own analytics counts sessions and filtered identified bots out by default. So a store's rate can rise or fall while its orders and sales stay the same.
Shopify's advice works for any tracking change. Mark the break and use the new data as the baseline. Then compare periods with the same filter settings.
GA4 and Google Ads report the same journey with different definitions and processing too. Our guide to GA4 and Google Ads reporting shows where the two part ways.
Traffic mix is the quietest of the seven. So here it is as arithmetic.
Chart data:
| Period A | Period B | |
|---|---|---|
| Higher intent | 8 | 8 |
| Other traffic | 1 | 1 |
| Overall | 7.3 | 4.5 |
SourceExample arithmetic, not client data
Period | Higher intent visits | Other visits | Enquiries and overall rate |
|---|---|---|---|
A | 900 at 8% = 72 | 100 at 1% = 1 | 73 of 1,000 = 7.3% |
B | 500 at 8% = 40 | 500 at 1% = 5 | 45 of 1,000 = 4.5% |
Both segments converted at exactly the same rate in both periods. Each segment's rate uses its own visits and the overall rate uses all 1,000. The share of lower intent traffic grew from 100 visits to 500. So the overall figure fell from 7.3% to 4.5%. Nothing on the page got worse.
A campaign that widened its targeting can do this. A new social channel or a seasonal shift in who is searching can too. The fix, if one is needed, is in the traffic, not the layout.
How does conversion rate affect lead costs?
Directly, and in one direction. That's why the number matters even though it isn't the final metric. Take a Google Ads campaign with 200 clicks at A$12 per click. That's A$2,400 of spend. Let the conversion rate mean leads per ad click for that same cohort.
Click to lead rate | Leads from 200 clicks | Spend | Cost per lead |
|---|---|---|---|
2% | 4 | A$2,400 | A$600 |
4% | 8 | A$2,400 | A$300 |
6% | 12 | A$2,400 | A$200 |
8% | 16 | A$2,400 | A$150 |
10% | 20 | A$2,400 | A$120 |
Cost per lead equals cost per click divided by the click to lead rate, as a decimal. The formula only works when the spend, the clicks and the leads share the same scope.
Drop a session based GA4 rate into a click based cost formula and the number means nothing. Sessions and clicks are different counts of different things.
The chart holds cost per click constant to show the arithmetic. It is not a forecast that a copy change will deliver the higher rates.
Then follow the raw cost per lead through to what the business paid for. Cost per qualified lead and cost per acquired customer are the numbers that decide whether the campaign works. A cheaper lead can be unsuitable, duplicated or never answered.
Our guide to a good cost per lead in Google Ads weighs it against lead quality. The spend side is in our guide to how much Google Ads costs. It covers the costs behind the headline click price too.
For a store, the equivalents are:
customer acquisition cost
order value
gross margin
returns
repeat purchases where you can evidence them
A discount can raise conversion rate and lower profit in the same week. Set targets against your unit economics and your capacity to deliver, not against a vendor median.
Where should you look first when the rate is low?
Fix what the evidence points at. Start with the measurement and end with the page.
Measurement goes first for a reason. Our audit sample is 180 Google Ads account audits from August 2021 to September 2026. Take the clicks but no leads problems we diagnosed in it.
We tracked where the first confirmed fault sat. In about 1 in 2 cases, it was measurement, a broken or wrong destination or enquiry delivery. It wasn't in the page's content.
Observed issue | Useful next investigation |
|---|---|
A sudden reporting drop | Validate collection, consent, filters and handoffs, then traffic, then the business records |
Unqualified visitors or enquiries | Inspect targeting, search intent, message match, offer fit and service area |
Relevant visitors cannot understand the offer | Clarify the outcome, suitability, inclusions, price context and next step |
Visitors need reassurance | Place genuine evidence beside the decision it supports. No ratings or results you cannot back up |
The action is difficult or broken | Check mobile usability, speed, form errors, payment and booking paths and tap to call |
Enquiries arrive but sales do not | Review response time, answered calls, qualification, attendance and follow up |
A store loses purchase intent | Review product information, stock, delivery, returns, payment options and checkout errors |
Three habits go with that table. Fix verified faults before you experiment. Improve qualification rather than maximising volume, because the problem is often after the page, in the response. And do not transplant an enquiry page template into a store, where the purchase conditions do the converting.
Work it in order. A low rate can have more than one cause. So a page fault is the last thing you assume, not the first. Our guide to why a website is not converting runs the same order across the whole site.
When the traffic is the problem, the usual suspects are search terms the offer was never for. That's where a working list of negative keywords earns its keep. The other suspects sit in the account. Our guide to why Google Ads is not working works through them in order.
When the page is the problem, five levers decide whether a relevant visitor acts. Each fits in a sentence.
Copy that names the job, the audience and the place, in the words the ad used. An offer with inclusions, conditions and a price context the business can honour. Proof placed beside the claim it supports.
Usability that works on a phone with the keyboard and the cookie banner open. An action that says what happens next and then does it.
No lever guarantees that fewer fields, a new hero or a full rebuild improves the business result. A rebuild in particular needs a demonstrated structural problem, not a slow month.
A verified form problem is the subject of our guide to enquiry form design.
We'll check what your tracking records against what reaches your inbox. Then we'll tell you which fix we'd make first.
How do you test and set a realistic target?
Decide how you will decide before you change anything. Define the primary outcome and the eligible population. Record the baseline counts and the commercial quality of what you already get. Then choose one specific hypothesis. Confirmed defects get fixed without a test.
If a change needs a proper test, decide the sample size and stopping rule before it starts. Then run the versions side by side, with visitors assigned at random. That's because a before and after comparison can't separate your change from a shift in traffic or tracking. Our guide to building a landing page that converts covers the page itself and the test design.
On a low volume site, fixing verified failures and reviewing the enquiries often beat an underpowered A/B test. And "inconclusive" is a valid result, not a win for the bigger percentage.
Set the target from your own recent comparable baseline and an improvement that would be economically meaningful. Give it quality guardrails, a capacity check and a review date. Don't set it from a median, a doubling or a seven day test. The same goes for a fixed number of conversions somebody once posted.
Outcome and population: [exact action and denominator]
Baseline: [dates, counts and rate]
Business need: [qualified leads, customers or profit]
Hypothesis: [change, evidence and expected mechanism]
Comparison: [randomised test or labelled observation]
Decision plan: [sample needs, delay and stopping rule]
Guardrails: [quality, cost, accessibility and capacity]
Review: [owner, date and limitations]
Before you act on any rate, run the checks below. They are the minimum for a number you intend to make a decision with.
Measurement
The outcome is an accepted action, counted once, with duplicates, spam and internal traffic handled
Matched comparison
The same outcome, unit, platform rules, period and visitor mix on both sides
Counts
The numerator and denominator are stated and large enough to carry the decision
Quality
Qualified enquiries or orders are reviewed beside the rate
Economics
Cost per qualified enquiry or margin is known, not only the percentage
Documented next action
One change, an owner, a review date and a labelled comparison method
If the checks pass and the journey still underperforms, then proper conversion rate optimisation is worth paying for. It starts from the same validated numbers rather than from a benchmark someone found.
What we recommend at Elev8d
Define the rate before anyone judges it. For a lead business, we report one number. It's accepted enquiries divided by the clicks or sessions that could have produced them. Accepted means forms that reach a person and calls that connect. Then we send a labelled test enquiry and make sure it lands before we read a single trend.
Next we compare the rate with the business's own last comparable period, counts included. For paid search enquiries we also hold it against our planning band. Below 3%, we work the decision tree from the top, tracking first. Inside the band, we start with lead quality and cost per qualified enquiry, not the design.
We change that order in three cases. A rate well above the band sends us to the conversion actions first. We look for double counting and button clicks recorded as enquiries. Large project trades and finance get judged on cost per qualified lead, because their rates can sit lower. And a store gets planned from its own purchase rate and margin, never from a service band.
We test only when the volume can settle the question. On a quieter site we fix verified faults and read the enquiries instead. Say your ads get clicks and the enquiries never arrive. Then start with our guide to Google Ads clicks but no leads.
FAQ
Is a 12% conversion rate good?
It depends on what was counted. Twelve newsletter signups per hundred sessions and twelve accepted quote requests per hundred clicks are different achievements. For quote requests from paid search, 12% sits above the band we plan with. So check how the account counts them before you celebrate. If the count holds up, ask what the enquiries were worth.
What rate should a new site or campaign plan on?
Borrow the closest named benchmark as a planning assumption, not a promise. For enquiries from high intent paid search, that's our planning band. For a store, use a benchmark only if its cohort looks like your setup. Swap the assumption for your own counts as soon as you have a comparable period. Judge the early months on cost per qualified enquiry or margin rather than the percentage.
Why can a higher conversion rate produce fewer sales?
Because the rate measures the action, not the value. A page that invites everyone to enquire can lift the rate. It can also fill the inbox with people the business can't serve. A stricter page converts fewer visitors, but into better enquiries. Follow the rate through to booked or closed work before you call the higher number the better one.
Next steps
Write down your definition
Name the outcome, the unit and the platform for the one rate you report. Then check that the outcome is an accepted action. Most of the confusion ends there.
Compare against your own history first
Pull the last comparable period with its counts. Mark any measurement changes and traffic mix shifts. Only then look at a named external benchmark, using the benchmark check template above.
