· 20 min read

Shopify Store Not Making Sales? Diagnose It in Four Layers

No sales is four separate problems, and only one of them is conversion. Here is the order to check them in, plus the 30-minute triage that separates them.

Empty checkout counter in a small shop with no customers waiting

The first thing most store owners change when the orders do not arrive is something visible. A new theme. A lower price. One more app. Three weeks later the store still has no sales, and now it also has a different theme, which means the next diagnosis starts from a new baseline and the old one is gone for good.

No sales is rarely one problem. It is four problems stacked in a fixed order, and only the last one is about persuasion. The other three are distribution, targeting, and arithmetic. Checking them out of order is how people spend a month rewriting product descriptions to fix a store that nobody is visiting.

This guide is the order to check them in, the ratios that do most of the work, a 30-minute triage you can run today, and an honest account of what a quiet dashboard cannot tell you.

Four Problems Wearing One Zero

Every store with no sales is sitting in one of these, or in a combination of them:

  1. Nobody is arriving. The store works, the audience does not exist yet.
  2. The wrong people are arriving. Traffic exists, but it has no reason to buy what is on the page.
  3. The right people arrive and do not buy. This is the layer everyone assumes first, and it is the third one.
  4. They buy and the order loses money. Sales exist, cash does not.

The order is not arbitrary. It is sorted by cost to test and by falsifiability. Layer 1 takes ten minutes and requires no changes to your store. Layer 3 takes a week of edits and is very hard to evaluate afterwards, because traffic moves at the same time you change the page. Do the cheap checks first, every time.

One more rule before the layers: write down the number before you touch anything. Sessions, add-to-carts, checkouts, orders, average order value, and what one order costs you in product and shipping. If you change three things and one number moves, you have learned nothing. The point of a diagnosis is to leave with one hypothesis, not a renovation.

First, Split “Never Sold” From “Stopped Selling”

These are different diseases and they get different treatments, so settle it in the first minute.

A store that never sold has no history. There is no baseline, no conversion rate to compare against, and often no traffic at all. The question is whether the store has been put in front of enough of the right people once.

A store that sold and stopped has a baseline, which makes the search much narrower. Something changed: a channel, a product, a price, a theme release, a payment method, a shipping rate. If the drop is recent and sharp, the fastest route is a formal drop diagnosis rather than a general one — our guide to diagnosing a Shopify traffic drop in 30 minutes covers the measurement artifacts and the four-way split that this article deliberately skips.

A store that sells seasonally is not broken in February. Before treating a slow month as a failure, check the same month last year, not last month.

A store where one product stopped and others did not is not a store problem at all. It is a product problem, or an inventory problem, and no amount of homepage editing will fix it.

The rest of this guide assumes the hard case: a store that has been live for weeks or months, gets some traffic, and produces orders at a rate that does not pay for the work.

Layer 1: Is Anyone Arriving at All?

Ten minutes, no changes, one output: a weekly session count you trust.

Start by separating three numbers that get used interchangeably and behave nothing alike. Sessions are visits. Users are people, and one person can be several sessions. Pageviews are pages, and a pageview count can double while your audience stays flat because you added a gallery. For this diagnosis, sessions is the number to use.

Then look at where they come from, because the total hides the interesting part:

  • Direct and brand search. Someone typed your name or tapped a bookmark. On a new store this is mostly you, your test orders, your friends, and your own phone. It is not demand.
  • Paid social. Volume is easy to buy and intent is close to zero. A thousand sessions from a cold feed is a different asset from a hundred from a search result.
  • Organic search. Slow to build, high intent when it lands on the right page. If this line is flat at zero after three months, the store has a content and indexing problem, not a conversion problem.
  • Referral and marketplaces. Usually a single link doing all the work. Check whether it is still live.

The judgment is blunt. If the store is seeing double-digit sessions in a week, there is not enough traffic for any conversion number to mean anything — the confidence interval around a rate computed from 40 sessions is wider than the difference you are trying to detect. A store in that state does not have a conversion problem, it has a distribution problem, and the next four sections will not help.

For a store you do not own, the same question has to be answered from outside. Panel-based estimates are modeled rather than measured, but they reliably separate a store doing thousands of monthly visits from one doing a few hundred, which is the only distinction that matters at this stage. Our guide to checking a Shopify store’s traffic for free explains what those panels can see and where they go quiet.

The trap in “we get traffic”

Almost every owner who says “we get traffic” means “we get some traffic”. The two statements have completely different implications. Before moving to Layer 2, write the actual number down, next to the number you assumed. The gap between them is usually the diagnosis.

Layer 2: Are the Right People Arriving?

If sessions exist and orders do not, the next question is not “why is the page bad”. It is “who is on it”. Five mismatches account for most of this layer, and each one is visible in the analytics you already have — no purchase needed.

MismatchWhat it looks like in the dataWhat it does to conversion
IntentHigh sessions, low engagement time, high bounce on the landing pageVisitors were never shopping; they were scrolling
GeographyTraffic concentrated in countries you do not ship to, or cannot serveThe visitor cannot complete the purchase at all
LanguageSessions from a market the store is not written inComprehension drops before persuasion starts
DeviceMost sessions on mobile, most revenue on desktopThe mobile experience is where the money dies
PromiseAd promises one thing, the landing page sells anotherThe visitor’s question is never answered

The last one deserves its own paragraph, because it is the most common and the least visible. A cold ad that says “the last organizer you will buy” and a landing page that opens with “Welcome to our store” are two different products as far as the visitor is concerned. Message match is not a copywriting nicety; it is the thing that decides whether the click becomes a session with intent or a bounce with a story attached.

Two free ways to test this layer without spending on ads:

Read the competitor’s angle first. Open the Meta Ad Library, filter by a page in your category, and read the first line of every ad that has been running for more than a month. Those lines are what the market responds to, and they are public. Our walkthrough of reading competitor ads for free covers the filters and the longevity test that separates a test from a winner.

Split your own conversion rate by channel. A blended conversion rate is an average of several different businesses. If paid social converts at 0.4% and organic search at 2.8%, the blended 1.1% describes nothing and hides the fact that one channel is working. Keep the split, even when the samples are small, and read it as an ordering rather than a precise number.

Layer 2 fails quietly because everything looks healthy from the top: traffic is up, page speed is fine, the product is good. The only symptom is that the audience and the offer were never introduced properly.

Layer 3: Right People, No Purchase

Now the assumption everyone started with. Even here, the first move is not persuasion — it is elimination, because a broken checkout looks exactly like a weak offer from a dashboard.

First, prove a stranger can buy

Place a real order from a device and network that is not yours, with a card you have not used on the store, shipping to a normal address. This one test catches an embarrassing share of “no sales” cases:

  • Is the product in stock, in the variant shown in the ad?
  • Does the shipping rate appear before the customer commits, or only at the last step?
  • Are the payment methods you expect actually offered in that country?
  • Does the discount code field reject a valid code?
  • Does the order confirmation arrive, and does the order appear in admin?

If any of those fail, stop. You are not looking at a conversion problem, you are looking at a transaction that cannot complete.

Then read the funnel as four ratios

Do not read the conversion rate. Read the three ratios inside it, because each one fails for a different reason and suggests a different fix.

RatioFormulaWhen it collapses, look here first
View to cartAdd-to-carts ÷ product page viewsFirst screen, price and value, reviews, photos, variant confusion
Cart to checkoutCheckouts ÷ add-to-cartsShipping cost revealed late, taxes, forced account, a coupon field that erases the discount
Checkout to orderOrders ÷ checkoutsPayment methods, risk rules, timeouts, stock running out mid-session
OverallOrders ÷ sessionsOnly useful after the three above are read separately

The shapes matter more than the levels, and you do not need an industry benchmark to read a shape. A funnel that is weak at the first ratio and normal afterwards is a page problem. A funnel that is healthy until the last ratio is a payment or shipping problem, and it will not respond to better copy no matter how many times you rewrite it.

For the page layer specifically, the conversion slots are finite and inspectable. Our teardown of twelve product page slots walks through each one — first screen, persuasion layer, decision layer, close — and what a missing slot looks like to a shopper. Work through them in order rather than redesigning the whole page.

The four-question version

If you only have five minutes, ask these in order:

  1. Can they tell what it is in three seconds? If the first screen needs a scroll to explain the product, the answer is no.
  2. Can they tell what it costs, including shipping? Surprises at checkout are a tax on trust.
  3. Do they believe it works? Reviews with photos, a specific return policy, and a real address do more than another badge strip.
  4. Can they finish? Payment method, country, device, stock.

Most “conversion problems” are one of those four failing, and none of them require a new theme.

Layer 4: The Order That Costs You Money

A store can clear all three previous layers, produce orders every day, and still be dying. This is the layer nobody checks because the dashboard is finally green.

The number that matters is not revenue. It is contribution per order:

Contribution = price − product cost − shipping you pay − payment fees − platform fees − returns allowance

Then compare that number with what one order costs to acquire. If acquisition costs more than contribution, growth makes the hole bigger, faster. That is arithmetic, not strategy, and it is why “we just need more traffic” is the most expensive sentence in ecommerce.

Two guides go deeper than this article can: the dropshipping margin calculator guide rebuilds the cost stack line by line and turns it into a per-order ceiling, and the break-even ROAS calculator converts that ceiling into the minimum return an ad has to produce before it is worth running.

Three leaks are worth checking before you assume the ads are the problem:

  • Stacked discounts. A sitewide sale, an automatic bundle discount, and a welcome code can all apply to the same order. The customer sees a deal; you see a contribution that went negative.
  • Free shipping thresholds set by feel. A threshold below your average order value gives away margin on orders that would have converted anyway. A threshold far above it produces abandoned carts instead of bigger ones.
  • Returns with no allowance. If 8% of orders come back and you never priced that in, your real contribution is lower than the spreadsheet says, and the gap grows with volume.

None of these show up as a conversion problem. All of them show up as a store that is busy and broke.

The 30-Minute Triage

Run this in order, write the numbers down, and stop at the first layer that fails. Fixing a later layer while an earlier one is broken is how a month disappears.

MinuteLayerWhat you recordPasses when
0–3SplitNever sold, stopped, seasonal, or single productThe problem has one shape
3–131. DistributionSessions per week, split by channel, new vs returningEnough traffic for a rate to mean something
13–202. AudienceConversion rate by channel, top countries, device splitThe converting channel exists and the page matches it
20–273. TransactionTest order, then the three funnel ratiosA stranger can buy, and only one ratio is weak
27–304. EconomicsContribution per order vs acquisition costEach order leaves money behind

Twenty-seven minutes of reading before a single edit. That ratio is the whole method.

Worked Example: Three Zeros, Three Different Stores

The numbers below are hypothetical, constructed to show how the same symptom produces three unrelated diagnoses. They are not benchmarks, and you should not compare your store to them.

Store A — three weeks old, 84 sessions a week. About 210 product page views, 9 add-to-carts, 3 orders. The conversion rate is 3.6%, which sounds respectable and means nothing: three orders cannot distinguish a good store from a lucky week. The diagnosis is distribution. The correct action is to put the product in front of more of the right people — not to rewrite the product page, which has been seen by roughly eighty people.

Store B — 4,200 sessions, 6,300 product views, 74 add-to-carts, 44 checkouts, 26 orders. View to cart is 1.2%, cart to checkout is 59%, checkout to order is 59%. The last two ratios are healthy and stable; the first one is where the store dies. Two-thirds of the visitors who reach a product page never put anything in a cart. That points at the first screen, the price framing, the reviews, or a variant selector that confuses people — in other words, at the page, not at the checkout, and not at the traffic.

Store C — 2,800 sessions, 4,200 product views, 286 add-to-carts, 232 checkouts, 71 orders. View to cart is 6.8% and cart to checkout is 81%, both strong. Checkout to order is 31%, which is the collapse. People want the product, they get all the way to payment, and three out of four do not finish. That shape is a transaction failure: a payment method missing in the buyer’s country, a shipping rate that appears at the last step, a risk rule rejecting cards, or stock disappearing mid-session. If that ratio returned to something like the store’s own earlier behaviour, orders would roughly double without a single extra session — which is why this layer is checked before anything creative.

Same zero on the dashboard. Three different fixes. The only thing that separated them was the shape of the funnel, read in order.

What a Competitor Can Tell You That Your Dashboard Cannot

Your analytics describe your store against its own past. They cannot tell you whether “this is what normal looks like in this category”, because they have never seen another store.

Public competitor data fills exactly that gap, and only that gap. From the outside you can read a store’s traffic order of magnitude, its catalog size, how many ads it is running, and how old its longest-running ad is. What you cannot read is its conversion rate, its margin, or its orders — none of which are published anywhere, whatever a revenue estimator claims. The distinction matters: use competitors for magnitude and direction, never for a precise target.

The practical use is calibration. If three stores in your category run 40 to 90 ads at a time and you run 4, the gap is not a creative problem. If they carry 300 products and you carry 12, your product page is not competing on assortment and never will. If their longest-running ad is four months old, that angle has been paid for long enough to be a known winner. Our free 30-second store analysis covers the fields that are actually readable in that time.

Apply WINNING for 20% off TrendTrack →

Verified 2026 · 20% off Starter, Pro and Business for your first 3 months

When you need that comparison across thousands of stores instead of the one in front of you, the paid platform is what makes it systematic. The TrendTrack promo code is WINNING - 20% off all plans for the first 3 months if you want to test the database against your own category before committing to a year of research.

Four Fixes That Almost Never Work

These are the interventions people reach for first, and the reason each one usually fails is the same: it treats a layer that has not been shown to be broken.

1. A new theme. A theme changes how the store looks. It does not change who arrives, what the ad promised, whether the payment method exists in the buyer’s country, or what an order costs. If Layer 1 or 2 is the problem, a new theme buys you a new baseline and no new information. If Layer 3 is the problem, a new theme can add a new bug — theme migrations break variant selectors, discount fields, and shipping calculators with impressive reliability.

2. A sitewide discount. Discounting converts a demand problem into a margin problem. On a store that already has thin contribution, a 20% sitewide cut can turn a break-even order into a loss, which means more orders and less money. If the offer is the issue, change the offer’s framing or the bundle, not the price of everything.

3. One more app. Every app adds scripts, and scripts add weight. A store with nine conversion apps is slower than the same store with three, and speed sits inside Layer 3 whether or not anyone measured it. Adding the tenth app to fix a Layer 1 problem is how stores end up with a slow page and no traffic.

4. A new product. Swapping the product does not fix distribution, targeting, checkout, or arithmetic. It resets the clock, which feels like progress because the dashboard is empty in a new way. Change the product when the evidence says the product is wrong — flat add-to-cart despite qualified traffic, high returns, a category with no advertisers — not when the evidence says nobody has seen it.

What This Diagnosis Cannot Tell You

Three honest limits, because a method that claims everything is a method you should not trust.

It cannot tell you whether demand exists. A perfect funnel on a product nobody wants still produces zero. Layers 1 through 4 measure whether the machine works, not whether the market cares. Demand is tested by putting an offer in front of strangers and watching what they do, which is a spend decision, not an analytics one.

It cannot give you a benchmark conversion rate. Published averages blend categories, price points, and traffic sources into a number that describes no actual store. A 1% rate can be excellent for a $900 furniture item sold cold and terrible for a $12 accessory sold to returning customers. Read your own ratios against each other and against your own history.

It cannot see inside someone else’s checkout. For a store you do not own, Layer 3 is largely invisible: you can place a test order, and you can read the page, but you cannot see the analytics that would tell you where visitors drop. Outside diagnosis stops at behaviour, which is why the ten free checks for whether a store is legitimate focus on what is published rather than what is assumed.

FAQ

Does TrendTrack have a promo code?

TrendTrack promo code is WINNING - 20% off all plans for the first 3 months. It applies to Starter, Pro and Business, works with monthly and annual billing, and stacks on top of the annual discount — an annual subscription with the code saves about 33% in year one.

Why is my Shopify store getting traffic but no sales?

Traffic without orders usually means the audience and the offer were never matched, or the purchase cannot be completed. Split your conversion rate by channel first, because a blended rate hides the fact that one source converts and another never will. Then place a real test order from another device and network to prove a stranger can actually pay.

How much traffic do I need before conversion rate means anything?

Enough that a single order does not move the number by a visible amount. In practice, a store seeing double-digit sessions in a week cannot distinguish a good page from a bad one, because the range of plausible rates is wider than the difference you are trying to measure. Fix distribution first and treat early conversion rates as anecdotes rather than evidence.

Should I lower prices or fix the product page first?

Neither, until you know which ratio is failing. If visitors reach the product page and rarely add to cart, the page or the price framing is the suspect, and a discount is one of several possible fixes. If they add to cart and abandon at checkout, a lower price will not repair a shipping calculator or a missing payment method.

How do I know if my checkout is broken?

Place a test order as a stranger would: different device, different network, a card you have not used on the store, and a normal delivery address. Watch whether shipping appears before the final step, whether your payment methods are offered for that country, whether a valid discount code is accepted, and whether the confirmation email arrives. Any failure there is a transaction bug, not a persuasion problem.

Can I diagnose a store I do not own?

Partly. From outside you can read traffic order of magnitude, catalog size, page structure, ad volume, and ad age, which is enough to rank stores by momentum and to calibrate what normal looks like in a category. You cannot read orders, conversion rate, or margin, because none of those are published on any public Shopify surface.

Will a new theme fix low sales?

Only if the theme itself is the specific failure, which is rare and testable. A theme change does not alter who visits, what an ad promised, whether checkout works in the buyer’s country, or what an order costs you. It does reset your baseline, and it routinely introduces new bugs in variant selectors, discount fields, and shipping calculators.

Final Word

A store with no sales is not a mystery, it is an ordered list of four questions. Is anyone arriving. Are the right people arriving. Can they buy. Does the order leave money behind. Answer them in that order and most stores resolve in under half an hour, usually at a layer the owner had not considered.

The expensive mistake is not picking the wrong fix. It is picking a fix before the diagnosis, then losing the baseline that would have told you whether the fix worked. Write the numbers down first, change one thing, and give it long enough to answer.

    Share:
    Back to Blog

    Related Posts

    View All Posts »