Why Is My Conversion Rate Low? Benchmarks and a Noise Test

Why is my conversion rate low? Treat the question as a claim to test: measure the rate the way you think it's measured, compare it with the right benchmark, rule out noise, split it by traffic source, and only then look for the funnel step that leaks.

What does a low conversion rate mean? Fewer visitors buy than in comparable stores. It doesn't tell you why, and the reasons that sound obvious (friction, weak copy, missing trust) are the last things to check.

In this guide:

• The three tool definitions that make one store show three different rates
• Shopify percentile cutoffs, and who they leave out
• A noise test you can run on your own session count
• A source split rule, a step-by-step funnel read and one fully worked store

A conversion rate depends on which tool computed it, because Google Ads divides conversions by ad interactions, GA4 divides sessions with a key event by sessions, and Shopify divides sessions that completed checkout by sessions.

The generic conversion rate formula is conversions divided by the visits that could convert. For a store that counts one purchase per order, that is orders divided by sessions. Each tool then picks its own denominator:

• Google Ads: conversions divided by the ad interactions that can be tracked to a conversion in the same period. If you count "Every" conversion, Google says the rate can pass 100%.
• GA4 session key event rate: sessions in which a key event happened, divided by total sessions.
• Shopify: sessions that completed checkout, divided by sessions.

Three different denominators mean three honest answers to the same question. The benchmark in the next section uses the GA4 form, sessions with at least one purchase over all sessions. So compare it with your GA4 rate, never with your Google Ads rate.

Two mechanisms move the number without any change in customer behavior.

Shopify's session update. Shopify's session measurement rollout ran from September 21 to 23, 2026, and identified bot sessions are now filtered out of session reports by default. Shopify's own help page says that when your session count changes, your conversion rate can change even when orders or sales haven't.

Its analytics field notes add that the rate can look lower than expected because of bot traffic, and suggest the Human filter. Human or bot, a session sits in the denominator until a filter removes it.

GA4 consent modeling. GA4 reports no events from visitors who decline consent unless the property meets Google's behavioral modeling thresholds. Those are at least 1,000 events a day with analytics storage denied for at least 7 days, and at least 1,000 daily users with it granted on at least 7 of the previous 28 days.

A small store usually misses both. Visitors who decline consent are then not reported, so your GA4 rate describes the visitors who accepted.

Here's the three-step check before you touch any page:

1. Match the numerator. Confirm that the orders in your store admin match the orders behind the rate.
2. Clean the denominator. Filter to human sessions, so bot sessions don't dilute the rate.
3. Match the definition. Read your rate in the same form as the benchmark you compare it with.

Among Shopify stores with at least 90 purchases in 90 days, a conversion rate is low when it sits in the bottom 20% of stores, below 0.5%, and the common 2% line is not that cutoff. Once the rate is confirmed low, what to test first on Shopify sets the order of work by monthly sessions.

The numbers come from Littledata's November 2026 panel of 450 Shopify stores. It measures GA4 sessions with at least one purchase divided by all sessions over 90 days. It only includes stores with a connected GA4 property, at least 1,000 sessions and 90 purchases in that 90-day window, which is about one order a day.

That floor matters. The panel is Shopify-only and excludes small stores, so these are not "ecommerce averages". The average ecommerce conversion rate by industry looks like this on Shopify:

Download CSV (CC BY 4.0)

Littledata puts beauty and skincare at the highest median conversion rate (2.8%) and home and furniture at the lowest (1.1%). Across all 450 stores the mean is 1.9%, the middle 60% of stores sits between 0.5% and 2.8%, and the top 10% reaches 3.6% or more.

Now the correction. The under 2% rule sits at the panel mean of 1.9%, above the 1.3% median. Half the panel sits below 1.3%, which is under 2%, so the rule labels at least half of these stores low. A store at 1.3% sits in the middle of the panel.

The question "why is my conversion rate low on ecommerce" has three different answers. For a low conversion rate Shopify store, find yours first:

1. Your store is below the panel floor. The 0.5% cutoff is not your bottom fifth. At 100 sessions a day, a store needs at least 1.0% to qualify, and at 300 a day at least 0.33%, because 90 purchases must fit inside 90 days of its sessions. Use the 1.3% median as a loose reference and let the noise test in the next section decide.
2. Your rate fell from a known level. Your own trailing rate tells you what changed. The panel only tells you whether the earlier level had headroom. Checks 1, 3 and 4 still apply.
3. You have no history and clear the floor. The panel is your yardstick.

If you run a SaaS product and your rate is a signup or trial rate, SaaS conversion rate benchmarks cover a different population from this Shopify panel.

A conversion rate is only evidence of a problem once enough sessions have passed: at the 1.3% median, 200 visitors give a 7.3% chance of zero sales by luck alone.

That's the "200 visitors, no sales" worry in one number. We got it by treating each visit as an independent chance at the median rate and compounding the chance of no sale over 200 visits. The chance of zero sales at the median rate shrinks fast with volume:

• 100 visitors: 27.0% chance of zero sales.
• 200 visitors: 7.3%.
• About 230 visitors: the first point where the chance falls under 5%.

So if you're wondering "why am I not getting conversions" after a few hundred sessions, the first answer is that a median store would often look the same.

Orders matter as much as sessions. A 95% interval is the range your true rate probably sits in given what you observed, and you can compute one for any store.

Take two illustrative stores, with Wilson score intervals. One records 10 orders in 2,100 sessions, which is 0.48% with a 95% interval of 0.26% to 0.87%. The other records 2 orders in 200 sessions, which is 1.0% with an interval of 0.27% to 3.57%.

The first store is below the median even at the interval's top end, so its rate is a real signal. The second store's interval contains the median, so its 1.0% is indistinguishable from a median store.

Here's the rule: compute the 95% interval for your rate and compare its upper end with the median. If the interval contains the median, wait or pool more days before you change anything. If you want to know whether your traffic is enough to run a test at all, our cro best practices guide covers that threshold.

A conversion rate is a traffic problem when one source converts far below the others on the same landing page, and a page problem when every source sits below your benchmark.

The same split explains the high traffic low conversion pattern. More traffic from a source that doesn't convert lowers the blended rate while the page stays the same. So split before you rewrite anything. Use these three rules:

1. One source is low. Compare each source's rate with the rate of the other sources combined. If the others' 95% interval contains the store median and this one source sits below that interval, the problem is a mismatch between the traffic and the offer. Fix the ad, the audience or the promise on the same landing page.
2. Every source is low. If each source sits below your benchmark, the page is the problem. Move to the funnel read in the next section.
3. The mix shifted. A blended rate can fall while no channel changed, because the share of each channel moved.

The mix shift is easy to underestimate, so here is the arithmetic. Suppose paid social converts at 0.14% and every other source at 1.14%, both unchanged. When paid is two thirds of sessions, the blended rate is 0.48%. When paid drops to one third, it's 0.81%. Same page, same channels, a 0.33-point gap that comes only from who showed up.

Split by device next. Littledata's panel shows a median conversion rate of 1.5% on mobile and 1.2% on desktop, so a store whose mix moved toward one device can shift its blended rate the same way.

A new store adds a wrinkle. With no brand recognition and traffic that comes only from ads, visitors arrive cold. That's our judgment, not a measured effect, but it's why a new store should read its per-source rates against each other first and the panel second.

Once the source test names a landing page, landing page optimization is the next topic; this post stops at finding it.

In an illustrative funnel built from Shopify panel medians, more cart sessions leave between cart and checkout than inside checkout, so comparing your own step rates with the panel's medians finds the leak faster than guessing causes.

Here's how we built it. A median cart abandonment of 79% means 21 of every 100 cart sessions buy. A median checkout completion of 47% means those 21 are 47% of the sessions that start checkout, so about 45 start checkout. That leaves about 55 who leave between cart and checkout and about 24 who leave inside checkout.

Treat it as a direction, because nobody measured this split. The three medians come from different store subsets, because cart metrics use only stores that track add-to-cart events at least as often as purchases.

A median of a ratio isn't the product of medians either. Rebuilding the median rate as 6.9% times 21% gives 1.4% against the reported 1.3%, about 11% high. Run on 100 sessions, the funnel gives 6.9 carts, about 3 reaching checkout and 1.4 purchases.

Now read your own funnel against the panel. Classify each step as below its bottom 20% cutoff, below the median or fine:

Download CSV (CC BY 4.0)

The lowest step names the cause family. The table below maps each symptom pattern to a suspect cause and the one metric that confirms it. It's our judgment built on the sourced cutoffs, so read "suspect" literally:

The table covers the cause families the funnel can point to. Trust signals such as return policies, reviews and visible delivery times sit at the checkout row; shipping cost sits at the cart-to-checkout row. They look alike from the outside, and the confirming check is where the sessions end.

Two blocks need their own numbers. First, speed. Google says a good Largest Contentful Paint is 2.5 seconds or less at the 75th percentile of page loads. Use that threshold and no other.

Second, device. The panel shows add-to-cart at 5.4% on mobile and 10% on desktop, and checkout completion at 63% on mobile against 32% on desktop. We can't say why the panel shows that, so split your own steps by mobile and desktop before you assume which device is the problem.

Fixes follow the flagged step. Guest checkout, express payment buttons and shipping costs shown up front are worth trying only when checkout completion is under the 25% cutoff. If more than one step is low, conversion funnel optimization shows which stage to fix first.

Once you know which step to watch, session recordings and heatmaps show what visitors do there. Our guide to conversion rate optimization tools helps you pick them.

An illustrative store at 0.48% conversion is low on check 2, is not noise on check 3, and in check 4 paid social carries the gap while the other sources can't be told apart from the median at this volume, so the first fix is the ad-to-page match and checkout waits.

Every input below is illustrative; the benchmarks are real. The illustrative store has 2,100 sessions in 7 days (300 a day) and 10 orders.

Walk the five checks in order:

1. Measurement. The 10 orders match the store admin and the sessions are human, so 10 divided by 2,100 is a clean 0.48%.
2. Benchmark. 0.48% is under the 0.5% bottom 20% cutoff. At 300 sessions a day over 90 days, the store logs 27,000 sessions and about 130 purchases at that rate, which clears the panel's 1,000 and 90, so the panel comparison applies.
3. Noise. The 95% interval runs 0.26% to 0.87%, wholly below the 1.3% median. This is not noise.
4. Traffic split. Paid social has 1,400 sessions and 2 orders, 0.14%. The other 700 sessions have 8 orders, 1.14%, with an interval of 0.58% to 2.24% that includes the median. Paid social carries the gap; the other sources are indistinguishable from the median at this volume.
5. Leaking step. The whole store's add-to-cart rate is 1.9%. Paid social produced 12 cart sessions in 1,400 (0.86%) and the other sources 28 in 700 (4.0%). The 4.0% clears the 3.5% cutoff by 0.5 points on only 28 carts, so call it not flagged rather than proven healthy.

Here's the numbers table:

The leak sits before the cart, in paid social traffic. That decides the fix order:

1. Fix the ad-to-page match first. Align the promise in the paid social ads with the first screen of the landing page.
2. Pool more days and rerun check 4. The other sources are not flagged, but 28 carts is thin evidence.
3. Touch checkout last. Checkout work and cart recovery come after add-to-cart per source clears the cutoff, because cart recovery only helps when a cart exists and paid social produced 12.

When the leak is a Shopify-specific one, the fixes live in Shopify conversion rate optimization, which this post doesn't cover.

So a low conversion rate has to survive four tests before it's a page problem: the number is what you think it is, it's low against the right benchmark, it isn't noise at your volume and it's low in every source. Your next step is to pull the last seven days of sessions and orders by source from your own admin and run check 1 today.