Changes at the top of the page
Why a change visitors see right away can show a moment late, and the choice you make about it in the visual editor.
The part of a page visitors see before they scroll is prime real estate: the headline, the hero, the main button. It's also where a split test is hardest to run cleanly.
The top of a page is what visitors see the moment it opens, before they scroll: the big headline, the main picture, the first button. You can test changes there too. It just needs one extra choice.
Think of a shop window
A shop changes its window display at night, while the street is empty. In the morning, everyone sees the new display. Nobody saw it being changed.
But if you change the window at noon, people walking by see your hands moving things around. That's the top of your page: people are already looking at it.
Why it happens
Needlr reaches your site through Google Tag Manager. Tag Manager loads a moment after your page starts to appear. So a visitor who gets the new version could see the old one for a blink, then the new one. That blink is called flicker.
What Needlr already does for you
- Needlr starts early. Its early tag in Tag Manager starts the moment Tag Manager does, and puts the change in place right away.
- Lower on the page, nobody sees a thing. Anything below the first screen, or anything your page adds later, is changed before it shows.
- A test between two pages never flickers in this way: the old page turns white for a moment while the browser moves on, like a normal page load (more).
When the editor asks you
After you make a change in the visual editor and press Use these changes, step 1 shows what you changed. If the thing you picked was on your first screen, without scrolling, it also says This element is at the top of the page and asks: How should Needlr change it?
- Change it anyway (recommended, and already picked for you). Visitors who get the new version may see it fade in, for about 0.2 seconds. Every visitor is counted.
- Never let anyone see it change. If a visitor could already see the element, Needlr leaves that page view alone and doesn't count it. It does the same in version A and version B, so the race stays fair. But most visitors are then left out, and the impact check may never finish.
- Add one line to your site. A tiny piece of code in your site's <head> hides the page for a split second, never more than one second, until the change is in place. Nobody sees anything switch (how).
Under Other ways, three buttons change your plan instead: Send visitors to another page (redirect), Use a developer check (in code), or Pick something lower to choose another element.
Already added the line? Then there's nothing to choose. Step 1 says Your site has the no-flicker line, so visitors won't see it change.
Adding something new with the Add tab never asks. A new element simply appears with the page.
What you'll see on the check page
- With Change it anyway: a note says the element is at the top of the page, may fade in for about 0.2 seconds, and that every visitor is counted.
- Otherwise: a box says Shown without flicker, with the share of page views where the change was in place in time, and how many were left as is. Views left as is aren't counted, in either version.
- If that share is low, the box turns yellow: results still stay fair, they just collect more slowly.
- If it's very low, the check page warns At this pace this check won't reach a verdict (more).
Common mistakes
- Choosing Never let anyone see it change for the main headline. Almost everyone sees the headline right away, so almost nobody is counted. Keep Change it anyway, or add the line.
- Fixing a slow check by editing it. You can't change a running impact check. Stop it, and set it up again choosing Change it anyway, or add the no-flicker line to your site.
- Pasting the no-flicker line low on the page. It must sit high in the <head>, before the Tag Manager code, or it comes too late.
Words to know
- Top of the page
- What a visitor sees before scrolling. Also called above the fold.
- Flicker
- A blink where a visitor sees the old version, then the new one.
- Fade
- The new version appears softly, in about 0.2 seconds, instead of jumping in.
- Left as is
- A page view where Needlr didn't change the element, because the visitor could already see it. It isn't counted.
What the video says
The part of a page visitors see before they scroll is prime real estate: the headline, the hero, the main button. It's also where a split test is hardest to run cleanly.
Needlr loads through Tag Manager, a moment after the page starts to paint. So a visitor in the new group can catch the old headline for a blink. That's flicker.
Below the first screen, it never shows. Needlr's early tag starts the moment Tag Manager does, so anything lower on the page, or added later, changes before anyone sees it.
At the top, the editor asks you. Change it anyway, the recommended choice: the new group may see a quick fade of about 0.2 seconds, and every visitor is counted.
Never let anyone see it change leaves those views alone, in both groups, so the race stays fair. But most visitors drop out, and the impact check may never finish.
Want zero flicker? Add Needlr's one line to your site's head. The page waits a blink, never more than a second, and nobody sees a switch. Then test where it matters most.