3 Marketers Went Head-To-Head, Here Are The Results

Show first post

37 replies

Cartrack_Portugal

This article is misleading in my opinion and the answer from Jonathan is not enough. What he is saying is that, even though the conversion rate stayed the same, it’s still a win situation with SmartTraffic because it brought conversions that otherwise wouldn’t convert. How is that possible if the conversion rate is basically the same? How can you guys reach that conclusion with that data?

You send 1.000 visitors to landing page A and another 1.000 to B. Conversion rate is 10%. You get 100 conversions for each landing page and you used SmartTraffic with landing page A. You finish by saying that SmartTraffic is a really good thing. Cmon. You can’t get that info from that data.

Ed_Stanfield
Badge

I see there is a lot of confusion/debate about the conversion rate metrics. I have to agree it is confusing.

I think the most valid comparison would be to include the original design as a variant. You never know what variables may be occurring in the market that may be affecting traffic and conversions, so comparing the 3 new designs to historical metrics isn’t really apples to apples.

Nevertheless, I learned a few things from reading the logic behind the designs and reviewing the 3 new pages. I especially benefited from learning about the script for the fading header.

Jonathan
Badge +2

Some considerations:

  • I don’t think this post is claiming that “smart traffic always works”. It’s about showcasing an experiment and test (which every advertiser should be doing actively) and sharing the findings with everyone. It’s market research. It’s not about winning, it’s about collecting data and using it to further our testing and future iterations. This is a community sharing its findings. So Thank You G Force and Unbounce for sharing!

  • As a general rule, as you scale CVR will always dip for all and any variant.

  • Potentially, if the original variant got ALL the traffic, its CVR may have dipped even further. Of course further testing is needed to validate this theory.

  • Let’s take Facebook’s ad level testing as an example. If you only upload 1 ad in a Facebook campaign, the algo won’t have much to work with. If you upload 5 ads, it will serve the ad it thinks fits most with the user’s behaviour. This is how Facebook tries to get you impressions that are more likely to result in engagement.

  • Smart Traffic tries to do the same thing. It routes your traffic to the page that is most likely to be relevant.

  • The real question is “how do you know CVR would’ve been lower if we didn’t run these challengers?“. In reality the only way would’ve been to run a separate “split test” campaign in Facebook with only the original LP as the destination - and compare that to the multi-variant Smart Traffic test.

  • Furthermore, you now have 4 variants that had more or less the same CVR, but you also have 4 different source of data (heatmaps, recordings, full funnel data, closing rates, lead quality, etc) which will help you develop your strategies in the long term. To me, that’s a great win. You collected heaps of data, without compromising on CVR. Usually, it costs money to collect that sort of data, by compromising our CPL/CVRs for the sake of testing.

  • Another theory is that some variants got traffic that otherwise would not have converted on the original variant. In that case, a 3% CVR is better than 0. That’s how Smart Traffic is supposed to work. To test that theory further testing is required.

G

No problem. I really enjoyed it and I’d do it again, can you be selected twice? 😀

Olivier

Hello @julien_level , just wondering how long are your forms usually to make multi step forms perform better? do you think multi step forms would perform better for a 4/5-field form? merci !

julien_level
Userlevel 5
Badge +4

Hello @Olivier !

It depends on my client but usually it’s the classic field (first name, last name, email, zip code, city and phone number). I would suggest to start first with a question that is more engaging than « first name « for example.

For The contest I added as the opening question: « tell us about your situation ».
It was not required by @gforce01 but I felt could be also useful for the call center and visitors may feel that Homelogic was interested in their situation.

Julien
Olivier

Thanks Julien for the reply, helpful. Should I A/B test multi step vs single step form or is it a waste of time and I should rather ab test something else? The advantage of a single step form to me is that the prospect can see it’s very short so I am still undecisive.

julien_level
Userlevel 5
Badge +4

I understand. I think you should A/B test it

Julien
Olivier

Thanks again Julien, I will. Would you know how to change “1 of 4” to “1 sur 4” on the multi step form?

julien_level
Userlevel 5
Badge +4

yes !

somewhere on the script you have:
},
step: function(state, bar) {
bar.setText(currentField + 1 + ’ of ’ + allFields.length);
bar.path.setAttribute(‘stroke’, state.color);

You just have to replace “of” by “sur”.

Have you added a script to translate error message ? if not i’ll send you our or you can find it on the community.

Julien
B

Hi Guys, hope all is well. I have found the main factor for improving conversion will be landing page and quality score with Google. However, there is something quite big in the type of traffic within the google platform. For example, if you optimise your account for ‘maximise clicks’ this will increase your click rate at a lower cost and usually lower your conversion. if you are able to use Target CPA, this will usually higher your conversion but you pay more per click. Obviously, it’s important to choose the right bidding strategies for your business. Thought i would mention it as in the past i have run some really profitable campaigns on google display and paid 2p per click but only had 3% conversion. I have found conversion rate is important, but not as important as the holy grail, cost per customer acquisition. hope you find this helpful.

digibomb
Userlevel 6
Badge +1

CR aside …

Considering there were more conversions I’m curious if the CPA went up or down? And, how many of those conversions became paying customers? That, for me, would indicate if there was real success in optimization.

Reply


Terms & ConditionsCookie settings