Home > Tracking Campaigns > Voluum

Best practice for evenly rotating landers and direct-links in Voluum? (5)

02-19-2017 09:34 PM #1 chris_climbs (Member)
Best practice for evenly rotating landers and direct-links in Voluum?

Hey guys,

I'm curious to hear other people's setups for rotating landers with direct linking in a Voluum flow or path.

For example, I'm currently setting up a campaign that has 5 offers and 5 landers to be tested, or 25 unique 'funnel-paths'. Obviously without direct-linking, this splits evenly to 4% of all total traffic to each path. I would also like to test the 5 paths that direct linking would create. This would mean 30 unique paths, or about a 3.33% share of the volume per path.

I'm basically wondering if there is an intuitive way to add these 5 direct-linked paths to the other 25, in which Voluum will do the math for me?

Or, should I just take my own calculations, then create two flows: a "direct linked" that gets 16.65% (i.e. 3.33 * 5) of all volume, evenly distributed between five direct links, then a "rotating" flow that gets 83.35% of all volume, split evenly amongst the landers/offers?

I hope that question makes sense; thanks for your help!

02-20-2017 05:45 AM #2 erikgyepes (Moderator)

Hey Chris,

So your goal is to test everything evenly, so each of the combinations gets +/- same amount of traffic?

Is that correct?

How I do this (I'm using flows in Voluum) is just create the DL path with the offers only, set the % to it and below I have the LP path with landing pages and offers.

I think to split it evenly is easy and just set it to 50/50 for both paths.

But once you start optimising and some offers will have more % and you want to follow that pattern in both DL and LP paths it can get trickier.

.
.
.

But are you sure you really want to throw all these 5 landers and 5 offers into a test?

1) it would require bigger budget to get some reasonable data (otherwise it will be just a big mix that doesn't tell you any story)
2) it would be much better to let's say isolate the 2-3 winning landing pages and in the next round figuring out the winning offers (or vice versa)

Ad "best practice": there are no rules and best practices in AM world. Everyone got their own thinking patterns and ways of doing things. Sooner or later you will figure out your very own "best practice"

Erik

02-20-2017 11:26 AM #3 osmiumman (Member)

I wouldn't make it too sophisticated. As Erik says "Everyone got their own thinking patterns and ways of doing things. "

I guess that your landers and offers will all be somewhat related to eachother, and not be totally unrelated. What I mean: each lander could work with each offer, correct?

Then in my opinion you can test all 5 landers with all 5 offers in one single path. You'll probably see pretty soon which landers and offers don't work, so disable them. So I'd just look at the overall data for each lander and offer, so it's just 5 landers and 5 offers to look at, not 5x5 possible combinations.
Again, this works unless there is one (hidden) unique lander -> offer combination that would beat them all but performs bad with the other landers/offers.

As for direct linking: I'd send 20% of the traffic to direct linking (to the 5 offers), because it will probably be worse than using a landing page. You can also start with 50% direct linking and stop it completely very quickly in case it doesn't work.

02-20-2017 11:41 AM #4 matuloo (Legendary Moderator)

Originally Posted by osmiumman

Then in my opinion you can test all 5 landers with all 5 offers in one single path. You'll probably see pretty soon which landers and offers don't work, so disable them. So I'd just look at the overall data for each lander and offer, so it's just 5 landers and 5 offers to look at, not 5x5 possible combinations.
Again, this works unless there is one (hidden) unique lander -> offer combination that would beat them all but performs bad with the other landers/offers.

This is what I'm doing too, even distribution across all possible combinations. But I see certain banner/lp/offer combinations stand out pretty often, if this happens, I isolate these particular combinations and re-test. Pretty often, they fail, but in some cases it's the start of a perfect campaign.

02-20-2017 03:39 PM #5 chris_climbs (Member)

Thanks guys for all of your thoughtful responses!!

Originally Posted by erikgyepes

But are you sure you really want to throw all these 5 landers and 5 offers into a test?

1) it would require bigger budget to get some reasonable data (otherwise it will be just a big mix that doesn't tell you any story)
2) it would be much better to let's say isolate the 2-3 winning landing pages and in the next round figuring out the winning offers (or vice versa)

Yea, they are all pretty similar "generic content download"-type carrier billing offers with pretty low payouts (0.10-0.20), so my initial thought was to test a grip of them together with a handful of "generic" landers to see if I could tease out a good combo. I guess my inclination was moving toward a larger initial test due to both the low payout, and because my last few camps got 0 conversions even after cycling similar offers through. I just want some data to work with!! :P

Originally Posted by osmiumman

I guess that your landers and offers will all be somewhat related to eachother, and not be totally unrelated. What I mean: each lander could work with each offer, correct?

As for direct linking: I'd send 20% of the traffic to direct linking (to the 5 offers), because it will probably be worse than using a landing page. You can also start with 50% direct linking and stop it completely very quickly in case it doesn't work.

Yes, they should, so I'll give this simpler strategy a shot. I think the idea of just giving it some arbitrary percentage (15/20/whatever) to the direct-linked path makes sense, as I have the feeling it wont perform as well anyway (based on my experiences direct linking similar offers in a different GEO). I was thinking a little too hard about it, it seems

Thanks everyone!!

Home > Tracking Campaigns > Voluum