Home > Design - Imagery, Banners & Landers > Landing Pages

---

Tested: How Much Does Page Loading Speed Affect CTR? (Results are IN!) (35)

---

We know loading speed is important. But HOW important?

Tomorrow you'll be seeing the first of a series of posts I'll be writing on optimising your landing page and redirect loading speeds. But before we get into that, it's worth knowing - just how big a deal are loading speeds?

Well, I've been doing some research, and the answer certainly looks like it's "really very lots". But we're affiliate marketers, right? We don't go by what we "know" works, we test.

So, I'm running a test over the next few days to settle just how much page loading speed matters.


The Test

Here's how it's going to work. I'm testing an offer in a vertical I'm familiar with - it's Find Me Love, actually, although that's not really important.

I've set up three identical landing pages - identical, that is, aside from a bit of PHP that artificially makes two of them load slower than they otherwise would.


The Fast Lander

Here's the Pingdom result for the first one, which I have optimised within an inch of its life:



That's probably about as fast as you can get a landing page to load - there's maybe another 50ms to shave off there if I was really pushing it.


The Normal Lander

Here's the Pingdom result for the second one, which loads about as fast as an LP that has been well optimised and is running on a fast server:




The Somewhat Leisurely Lander

And here's the result for the third one - still not disastrous, but getting a bit rubbish:



We've probably all gotten a bit lazy at one point or another and deployed something that loads at this sort of speed - I know I have. It's not ideal, but it's not INCREDIBLY slow.


I'm testing using the traffic source ZeroPark, because I wanted something that goes straight to LP hits.

I'm going to be tracking CTR primarily, although I'll post both CTR and conversions - if any. I'll post results when they hit statistical significance!

But before we get there - what do you think the results will be? Fill in the poll above, or comment below!

Update - the results are in!

---

Quick update on this - clicks are coming in slowly but steadily. With any luck, we'll hit statistical significance early next week. If not, I might just pollute the data a bit by pointing a second traffic source at the LPs too

---

Thank you for taking this on.

---

Great intention, looking forward to see the results. Is it mobile or WEB btw?

---

This one's a web test. If there's a demand, I might do a mobile version at a later date.

---

From my understanding, page loading speed time has the biggest impact on mobile. It also makes huge sense to test WIFI and specific carriers separately.
Anyway let's see what WEB insight give us first

---

VERY interesting

---

Definitely interested in seeing it for mobile. I've done some mobile testing myself and see MASSIVE conversion rate increases from better optimized servers and pages.

---

And the results are in!

After 1305 views and 75 clickthroughs, we have our results.

Drumroll please...

The Results


Landing Page: Slowest, which loaded in 2.14 seconds, received an average CTR of 3.6%

Landing Page: Average, which loaded in 0.64 seconds, received an average CTR of 5.5%

And

Landing Page: Fastest, which loaded in 0.13 seconds, received an average CTR of 7.8% - more than TWICE the CTR of the Slowest LP.

Whoa.

So, if you voted for "Absolutely MAHOOSIVE difference between the 3" in the poll, you were right - it turns out that 2 seconds of loading time can equate to more than a 100% increase or decrease in CTR.

Holy crap.

Even I, unstinting evangelist of the faster-loading LP, wasn't expecting that.

If you'll excuse me, I'm off to optimise my landers some more...

Conclusions

• I need to take that landing page round the back and shoot it
• It's not just "slow" landers that make a difference to CTR - the difference between "fast" and "blindingly fast" is at least as large.
• Obviously, these findings will vary to a certain extent by traffic source. However, they echo what big companies have found testing CTR and revenue in all sorts of media.
• If you've got a slowish lander that's already giving you a good CTR, you could put a rocket-ship under your profits with a couple of hours of trimming images and code!

Addendum: The Math Bit

Is this test statistically significant? Yup.

The fastest lander beats the slowest lander with a 92% confidence factor.

The mid-range speed has only a 60% confidence compared with the slowest lander, but that's still enough to get a strong indication.

---

Awesome! So the next question is....how do we make sure our LPs are loading at an extremely fast rate? Aside from actual images/scripts on the page, what else can we do to bump up the speed of our tracking redirects & server load times? (on a dedicated server, not shared hosting).

~Ian

---

What web speed tester did you use? I wanna check mine

---

@aerolan - well, there are some tips in this case study, and watch this space for more server optimisation tutorials coming this week or next!

@philwilrgm - I recommend using Pingdom's checker if you're running a campaign in the areas they have data centers.

---

Fuck me I missed this. Those numbers are crazy! Cheers Caurmen!

---

@caurmen - awesome test! What host are you using for your landing page?

---

Thanks! Those pages are hosted on a Bytemark server, because the test was on UK traffic. For my US case study a little while ago (the 34ms load time one), the host I used there was Linode.

Note - both of those services were unmanaged, and I set the servers up from a fresh operating system install.

---

So what is the best way to achieve maximum loading speed on landing pages? I mean to make it lightening fast that is? And is it possible to do this using cloud hosting?

Thanks!

---

There's no one way to achieve maximum speed - much like affiliate marketing, it's a process of testing and refinement. However, here are a bunch of tips on speeding up your LPs, and I've got a similar list of server configuration tips and tweaks coming up in the next few weeks.

Cloud hosting - very much depends on which cloud host you use. If you're using a hosting service that doesn't optimise well for speed (I believe Amazon EC2 falls into this category) then you're always going to be operating at a disadvantage. I'd strongly recommend choosing your host based on how much you trust their ability to give you the fastest hardware and software possible before anything else.

However, in principle there's no reason an intelligently set up cloud host shouldn't be able to achieve super-fast loading speeds. The pages in this test were actually being run on a cloud host, Bytemark's new BigV service.

---

Would be awesome if we had our own benchmark of different hosting companies. Can get people who are interested on the forum to load up a "control" test page onto their server and monitor the response time for a few days.

---

I would be up for trying this out, i'm curious to see if I'm with the right host.

---

I'm on it! Look for something next month.

---

Dude, fantastic post

---

thanks for this

---

Damn, i have a lot of work to do LOL. Thanks for the post

brbrbrbrbrbrbrbrbr optimizing landers

---

YES, plase do!
And do I need pingdom if I'm doing mobile. Thanks

---

What do you mean? What part of Pingdom? You won't need any part of Pingdom, or really any service, to do mobile - there are tonnes of services you could use for a lot of different things, but this doesn't mean you need them.

---

Any part of pingdom to test speed of my LP, should I use 'em? Thanks

---

Yes, use http://pingdom.com/fpt/ to test your lander load speeds.

---

Do you use different hosting or some cdn for different target markets?

---

It depends on your goals, technical proficiencies and tracking setup.

When starting out I would advise just using a CDN where possible.

---

How are the speeds for mobile loading speeds like? I just tested mine on first load is around 0.7-0.8s. How much is everyone else getting? And what are you using to test? I'm using gtmetrix

---

That's not terrible but a bit higher than what I'd generally aim for. Normally shooting for sub-500ms is a good goal, IMO.

Obviously that depends on the content of your successful lander, though, as well as the geo you're targeting.

---

my lander has no images at all and i have a singapore server and the ads are serving in singapore at the moment. any idea on how to lower the loading times?

---

Is your test of your tracking system link that redirects to the lander? Or just the lander directly? You should inspect the waterfall graph and find out what's taking so long.

If in doubt, offload the entire page onto a CDN (if HTML-only) and try again.

---

It's already on cloudflare, that means it already off loaded? I used the direct link to my lander, not the tracking link from Voluum tracker

---

What is the lander extension?

If .php then CloudFlare won't cache it. If .html, CloudFlare may cache it depending on the settings.

I'm not 100% sure on CF's page caching habits.

---

Home > Design - Imagery, Banners & Landers > Landing Pages