Traffic but No Sales? Find the First Decision Your Visitors Couldn't Finish, Before You Buy Another App

Write down what you actually know, and what you don't

Pick one time period, long enough to contain a useful number of visits but short enough that you remember what changed.

Then make a simple table:

Don't invent missing numbers. Write "unknown" and note what you could check by hand.

Here's the key idea. A visit means someone arrived. A product view means someone found a product. A basket means someone was interested. None of them tells you why a sale didn't happen.

Separating them stops the most expensive mistake: treating every visit as proof of the same problem.

Before blaming the page, check what the link promised

A visitor from a viral video is not the same as someone who searched for "stained glass suncatcher".

So look at the actual route, the specific video, not just "social". Ask four questions:

• What was this person probably looking for?
• What did the video promise they'd find?
• Does the page they land on deliver that quickly?
• Could someone understand the offer without already knowing the studio?

Saoirse's answers were uncomfortable.

The video showed one large, rainbow-coloured suncatcher, about 30 cm across, catching the light. Viewers wanted that one.

The link in the profile went to the shop's homepage. The homepage showed twelve small suncatchers, all about 10 cm across. The large rainbow piece wasn't there, because it was a one-off commission that had already been sold.

So 900 people arrived looking for something that didn't exist on the site. About 120 of them looked around anyway. Two bought something different.

"Bad traffic" isn't an explanation; a mismatch is

It's tempting to say the video brought the wrong audience. That's partly true, and it isn't useful.

The useful version is specific: the video promised a large rainbow piece, and the page offered small single-colour pieces with no mention of the large one. That's a mismatch you can fix, either by making large pieces available to order, or by linking the next video to something people can actually buy.

Understanding comes before persuasion, so check it next

For the visitors who did reach a product page, the next question is whether they could understand it.

Read one product page as someone who's never seen stained glass up close:

• What is this item?
• How big is it, really?
• How does it hang?
• What colours will I see, and does it look like the photo in my window?
• What arrives in the parcel?
• How fragile is it in the post?

Glasshouse's page said "Small sunburst suncatcher, €35" with a beautiful close-up photo. A close-up of a 10 cm piece can look like a 30 cm piece. There was no measurement until halfway down the description.

Put a mark beside the first question you had to stop and work out yourself. That's your starting point for the page.

Choice problems hide inside attractive pages

Some visitors understand the product and still can't choose.

Glasshouse sold the same design in four colourways with names like "Ember", "Tide", "Moss" and "Dusk", shown as small colour dots. In daylight through a window, stained glass looks very different from a product photo on a white background.

The fix was a photo of each colourway hanging in a real window, and a plain line: "Ember: amber and red. Tide: blue and clear."

If your options look similar, have names only you understand, or appear only after several taps, the visitor is being asked to do detective work.

Trust is the gap between liking something and being willing to pay for it

Someone can love the suncatcher and still pause, because they can't predict what happens after paying.

For a fragile, handmade product, the worries are specific:

• Will it break in the post?
• Is it made to order, and if so, how long will it take?
• What happens if it arrives cracked?
• Who's behind this shop?

Glasshouse's delivery information was on a separate page. The product page didn't mention packing or breakage at all.

So the page got one line beside the button: "Packed in foam inside a rigid box. If it arrives cracked, send a photo and we'll make you another."

Put the answer next to the decision that raises the question, not in a footer link.

Basket and checkout are separate problems with separate fixes

Keep them apart in your notes:

• People view products but rarely add to basket: look at discovery, understanding and choice
• People add to basket but don't start checkout: look at the full cost, delivery charges and anything surprising
• People start checkout but don't finish: test the payment process yourself and look for errors

Glasshouse had 9 baskets and 2 orders. When Saoirse added a €35 suncatcher to her own basket, she saw why some stopped. The delivery charge, €9.50 for a tracked and padded parcel, appeared for the first time at checkout. That's over a quarter of the price, and nothing on the page had warned about it.

She didn't lower the charge, because it reflected the real cost of sending glass safely. She added it to the product page: "Tracked delivery in protective packaging: €9.50 in Ireland."

Don't assume an abandoned basket means someone disliked the product. But if the same surprise appears in your own test, fix it before you pay for more visitors.

The one-hour diagnostic: one route, one product, one blocked decision

Pick one route that has actually brought visitors, even modest numbers.

Copy the promise. Write down exactly what the post, email or search result showed or said.

Open the landing page. Does it show what was promised, or make the visitor search? If it makes them search, stop there and record the mismatch.

Follow the path as a stranger. Phone in hand, pretend you don't know the shop. Note the first moment you need to interpret a name, hunt for an answer or calculate a cost.

Don't fix anything while you test. The point is to see the path as it is.

Label the first blocked decision, not the most dramatic one

Use one label for the first place the route breaks:

• Discovery: the visitor can't find what was promised, or wasn't looking for what you sell
• Understanding: they can't tell what it is, how big it is or what's included
• Choice: they can't pick an option with confidence
• Trust: they can't predict delivery, quality or what happens if something goes wrong
• Basket: the total or conditions change their mind
• Checkout: the transaction is slow, confusing or broken

Choose the earliest blocked step. Glasshouse had problems at discovery, understanding, trust and basket. But discovery came first. No amount of checkout polish helps 780 people who never found a product they wanted.

Write one repair sentence

"Visitors arriving from [route] can't [decision] because [observable problem]. This week I'll [one change] and check [one signal]."

Saoirse's: "Visitors from the window video can't find the piece they saw, because the link goes to a homepage without it. This week I'll make a 'Large suncatchers, made to order' page and link the next video straight to it, and check how many visitors reach a product page."

That's testable. "Improve conversions" isn't.

Four common patterns, and where each one points

Your numbers can't explain why people leave, but certain patterns suggest where to look first:

• Many visits, very few product views: usually a discovery problem. Links going to a homepage, or a post promising something the site doesn't show.
• Many product views, very few baskets: usually understanding or choice. Missing size, unclear contents, confusing options, no delivery information near the button.
• Many baskets, few checkouts started: usually a cost surprise. Delivery appearing for the first time, a minimum spend.
• Many checkouts started, few completed: usually a transaction problem. An error, a payment method that doesn't work, a field that breaks on a phone.

These aren't rules. They're where to point your attention first.

Ask the visitors what they were looking for

Analytics show where people went. They rarely show what people wanted. Sometimes the simplest way to find out is to ask.

Saoirse added one question as a pinned comment under the video: "If you came to the shop from this video, what were you hoping to find?"

Within a day, over forty people answered. Most said "the big rainbow one". A few asked about custom colours. Several asked whether she posts to the US.

That's not a scientific survey. But it confirmed the discovery problem in people's own words, and it surfaced a delivery question she hadn't considered.

Check how fast the page loads on mobile data

A slow page is its own kind of blocked decision. Most video viewers were on phones, often on mobile data.

Saoirse loaded her homepage with Wi-Fi switched off. The large photo of the studio window took several seconds to appear, and the product grid below it took longer still.

The photo files were very large, straight from the camera. Resizing them made the page appear much faster. If your first screen takes more than a few seconds on mobile data, look at your image sizes before anything else.

New visitors and returning visitors need different things

A returning customer may buy from a fairly thin page, because they already trust you. A first-time visitor from a viral video knows nothing about you.

Almost all of Glasshouse's video traffic was new. So the page needed to work for a complete stranger: what it is, how big, how it arrives, who makes it.

Judge the page as the visitors you actually have would see it, not as a loyal customer would.

Small numbers wobble, so look at the counts behind any rate

Two orders from 900 visits sounds alarming as a rate. But the counts tell a fuller story: most of those 900 never saw a product at all.

When numbers are small, percentages exaggerate. Going from two orders to four is "100% growth". It's also two people.

For small shops, direct evidence often tells you more than rates do: the questions people send, a test you run yourself, someone unfamiliar trying to find a product while you watch.

Pick a check that fits the size of the fix

A small change needs a small, relevant check. One page edit can't prove a theory about your whole shop.

• A route fix: check whether visitors now reach the intended product
• A missing fact: watch whether the same question keeps arriving
• A choice problem: ask someone unfamiliar to choose, and explain why
• A cost surprise: check whether baskets reach checkout more often

Keep a four-line change log: date and product, the problem tested, the change made, and the result plus what's still uncertain.

When a new tool is actually justified

Analytics tools, heatmaps, session recordings and checkout apps all have real uses. The trick is naming the job before buying.

Write the requirement as one sentence: "I need to see [specific event] for [specific route], so I can decide [specific repair]."

Saoirse couldn't finish that sentence for a countdown timer or a "people viewing" badge. Neither would have helped someone who couldn't find the piece they came for. So she didn't install them.

Most questions a paid tool answers can be partly answered for free: watching one person use the page, testing your own basket, reading customer messages. If those still leave a specific gap, that's when a tool earns its price.

Sometimes the honest answer is that most of the traffic was never going to buy

A viral video brings lots of people who enjoyed the video. Many of them were never shopping for a suncatcher.

That's not a failure. It's just what that kind of attention is. A clear page will convert more of the people who were interested, but it won't turn every viewer into a buyer.

So keep your expectations honest. The useful question isn't "why didn't 900 people buy?" It's "of the people who wanted something, how many could find it, understand it and trust it?"

Fix the live route while the attention is still arriving

A popular post keeps sending visitors for days or weeks. So the first fix doesn't have to wait for a perfect new page.

The same afternoon she found the mismatch, Saoirse changed the link in her profile to the "Large suncatchers" page, even though it was only half finished, and pinned a comment: "The big rainbow piece in this video was a one-off, but I take orders for large pieces. Details here."

That simple, honest fix redirected the rest of the video's traffic to something people could actually buy, while she finished the page properly.

What happened next at Glasshouse

Saoirse made three changes, one at a time:

• A "Large suncatchers, made to order" page, with sizes, prices, a four-week making time and real window photos. The next video linked straight to it.
• Measurements and a hanging photo near the top of every product page.
• The delivery charge and packing note beside the button.

The next video sent fewer visitors, about 300. But most of them reached a product page, and eleven people ordered a large piece.

That doesn't prove every change worked, and the second video might simply have been a better fit. But the route now delivered what the video promised, and that was the problem she'd found.

Your next step this week

Take the last post, email or video that sent people to your shop. Copy exactly what it showed or promised.

Open the page it links to and ask: does this page deliver that promise within a few seconds?

Then follow the route as a stranger, and write down the first moment you had to stop and think. Label it: discovery, understanding, choice, trust, basket or checkout.

Make the smallest change that helps at that exact point, and choose a check that fits it. One route, one blocked decision, one repair.

Keep going

Product Pages / Practical guide

Product Page Checklist: 12 Checks to Run Before Your Page Goes Live, Starting With the Fact That Rules Buyers Out

Most product page checklists count the parts of a page. These twelve checks test whether a buyer can understand, choose and order without guessing, using one made-up dog harness to show every check finding a real problem.

Product Pages / Practical guide

What Buyers Should See First on a Product Page: What It Is, Which Version, the Price and the Next Step, Before Any Story

A new visitor decides in a couple of seconds whether your page is worth scrolling. The top of the page doesn't need to say everything. It needs to say what the product is in ordinary words, which version is selected, what it costs, what to do next, and the one condition that could change the decision.