In short
Learning goal: by the end of this lesson you can check the page your visitors land on for clarity, speed and a matching offer, and tell a fix you should just make from a change worth testing.
A landing page is wherever a visitor arrives: your home page, a collection, a product page, or a page you built for an ad or an email. On a new store most visitors arrive from somewhere specific (an Instagram post, a Google ad, a friend's link) with a specific expectation. The landing page's job is to confirm, in the first few seconds, that they're in the right place.
What to measure at this step
| Metric | Definition | Role |
|---|---|---|
| Revenue per visitor of everyone who landed | Revenue ÷ visitors who arrived on the page | Primary goal: the one that decides whether a change helped |
| Orders per visitor who landed | Orders ÷ visitors who arrived on the page | Primary goal when order size won't change |
| Next-page rate | Share of visitors who view a second page | Diagnostic: shows whether the page held attention |
| Product views per visitor | Product pages viewed ÷ visitors | Diagnostic: shows whether visitors found something to look at |
| Load time and page errors | How long the main content takes to appear; script errors | Guardrail: must not get worse |
Notice what's missing: bounce rate as a goal. A page can keep visitors clicking around and still sell less. Judge a landing page by what the visitors who landed on it went on to buy.
The three basics
1. Clarity: what is this, and is it for me?
Look at the first screen on a phone, without scrolling. A visitor should be able to answer three questions:
- What do you sell? A product photo or a short line of words, not a mood image.
- Why this one? One specific reason: handmade in small batches, fits a carry-on, refillable.
- What do I do next? One obvious button, such as Shop the collection or Add to cart.
If you need a paragraph to explain the product, the first screen isn't doing its job yet.
2. Speed: does it appear quickly on a phone?
Google's Web Vitals give a common yardstick. A page has a good loading experience when the largest piece of content appears within 2.5 seconds (Largest Contentful Paint), responds to taps within 200 milliseconds (Interaction to Next Paint), and doesn't jump around as it loads (Cumulative Layout Shift of 0.1 or less), measured at the 75th percentile of visits (web.dev, updated 31 October 2024). Shopify's admin reports your store's web performance on loading speed, interactivity and visual stability, and notes that each feature you add can affect it.
The usual causes on a small store are a very large first image, apps that load scripts on every page, and popups that load before the content. These are fixes, not tests.
3. The offer: does the page keep the ad's promise?
If an ad says "20% off the linen range", the visitor should land on the linen range with 20% off visible, not on your home page. Google Ads describes landing page experience as how relevant and useful your landing page is to people who click your ad. The same idea applies to every link you share: send each one to the page that matches it.
Fix or test?
This is the most useful decision in the course for a small store. A change is a fix when you'd make it whatever a test said:
- The page is slow on a phone.
- A popup covers the button.
- The ad promises a price the page doesn't show.
- An image is broken or the text is wrong.
- Shipping cost is hidden until checkout.
A change is worth testing when reasonable people could disagree and the effect could be large:
- A different first image (product on its own vs in use).
- A different headline promise (price vs quality vs speed of delivery).
- Sending ad traffic to a product page vs a collection page.
At small-store traffic even good tests need large effects. The example shows how large, and the sample-size calculator does the same sum for your page.
A store runs a paid campaign that sends 3,000 visitors a week to one landing page, and 1.5% of them order. It wants to test a new first image, split 50/50.
- Over 4 weeks (6,000 visitors per version), the smallest lift the test can reliably detect at 95% confidence and 80% power is 45.9%: from 1.5% to about 2.2%.
- Over 2 weeks (3,000 per version), it's 67.7%.
Computed with sampleSizePerVariant from Benchmyrk's statistics engine, through the same search the planner uses. The store is illustrative.
So for this store, a test of the first image only makes sense if the two images are really different, such as a product photo against a lifestyle photo, not two crops of the same picture. And everything in the fix list should be done before the test starts, or the test measures a broken page.
Common problems on new stores
- The home page as the landing page for everything. Ads, emails and posts all point to the home page, and the visitor has to find the thing that was promised.
- A slider of three messages. Visitors see the first slide. Pick one message.
- Trust signals hidden in the footer. Returns, delivery times and contact details belong near the button, in a line.
- A newsletter popup on arrival. A visitor who hasn't seen a product yet has no reason to sign up. Show it later, or on exit.
Test ideas, and why they'd work
| Idea | Mechanism |
|---|---|
| Product photo first, lifestyle photo second | Shows what you sell before setting a mood |
| One headline that names the main benefit | Answers "why this one?" in a glance |
| Send ad traffic to the product page it shows | Removes a click and keeps the ad's promise |
| A one-line delivery and returns note under the button | Answers the two questions that stop gift buyers |
A 10-minute exercise on your own store
- Pick the page that gets the most visitors from outside your store (your analytics' landing pages or top pages list).
- Open it on your phone on mobile data. Count the seconds until you see the main image and can tap the button.
- Without scrolling, write down what the page says you sell, why it's different and what to do next. If any answer is missing, that's a fix.
- Open your last ad, email or post that links to this page. Does the page show what it promised? If not, change the link or the page.
- List every app that adds something to this page. Remove one you no longer use.
Do it in Benchmyrk
For the Black Friday version of this checklist, see the fix list in Black Friday for a small Shopify store.
Check yourself
- Your landing page loads in five seconds on a phone. Should you A/B test a faster version? Answer: no. A slow page is a fix. Make it faster; there's no outcome of a test that would make you keep it slow.
- A new headline kept visitors on the site longer, but revenue per visitor didn't change. Did it work? Answer: not as far as you can tell. Time on site and next-page rate are diagnostics. The primary goal is what visitors who landed went on to buy.
- Your page gets 3,000 visitors a week and converts 1.5%. Is a test of two similar headlines worth running for two weeks? Answer: probably not. Two weeks can only detect a lift of about two thirds, and similar headlines won't move orders that much. Test a bigger difference, or just pick the clearer headline.