Home > General > Affiliate Marketing Forum

---

Is this a solid strat for quick testing? (17)

---

Yo Stackers!

I would like to have the feedback and opinion on a straight forward and very basic strategy I believe will allow me to launch campaigns almost every day.

I am trying to develop a quick strat for testing app installs. My goal is to launch a campaign everyday for some quick data.

I was thinking of doing the following provided I have done some research and created some decent angles:

1) I have a standard landing page template. I change the image and the text.
2) 4 - 6 banners for each angle - ideally 3 banners 320x50 and 3 banners 300x250
3) Split test different networks
4) Split test 3-4 traffic sources
5) Ditch the offer if it's not in green or very close to green

Pros of doing this:
- Quick launch of campaigns
- Quick data
- Covers important split testing procedures
- Easy to fulfill with my specific time schedule
- Point 5 allows me to cut campaigns out of the picture without having to use up to much time with analyzing data.
(Anything below -50% will be out of the equation)

Cons of doing this:
- Will not go in depth with researching. Research is performed in appannie, testimonials, my own experience using the app, affiliate manager's suggestions and anything I find quickly off the net.
- Not much room to test different angles.
- More likely to cut out a campaign because of bad research and preparation instead of bad offer.

I won't cut out a campaign if it's -50% on the first round. I'll give those a 2nd round with new banners/landers and optimizing some of the -50% banners/landers. Again, this won't take up to much time as the campaigns are already set and I just replace pictures, text and some new banners. But, it shouldn't take up anymore time unless it's in green or at least improves.

There's an exception for every campaign but the whole point is to launch as many campaigns as possible per day to get data rolling in and increase your chances of finding a winner to scale. Though, if the quick research looks promising and I feel it should do well with more effort, that will be considered an exception.

Sooo...to sum it up:
This is all about light research and frequent campaigns over heavy research and fewer campaigns .... with exceptions

I would appreciate your thoughts on the above i.e. if it's viable, if people use this , if it's common or not. Please do let me know how I can improve it or if it's completely useless

Looking forward to hearing from you soon.

Aris

---

Developing a system is a very good idea. Campaigns are significantly a numbers game.

One thing I would counsel you to change: make sure that you test each angles with more than one ad ( I use 8 most of the time ) and a dedicated lander for that angle.

Why? If you're using a generic lander for all angles, it'll negatively affect the angle as the funnel flow isn't good. Worse, it'll affect different angles by different amounts - meaning you can't trust your results.

If you use one banner per angle, meanwhile, it'll mean you can't tell if the angle's bad or the banner is bad!

Hope that helps. The plan's sensible overall - just gotta watch the wrinkles!

---

My apologies for the misunderstanding. I edited the post to make it clear.

When I said : "2) 3 banners for 2-3 angles" I actually meant 2-3 banners for each angle and dedicated to it. The lander would of course have an image and text that reflects the angles individually and each angle would have a lander. I meant a generic template, not lander. The layout is the same but I change images and text.

I was creating 8 banners per angle as well but banner's take up a lot of time since I don't have any banners of yet with statistical significance to remake (inventory). All of them are made from scratch and unique, hence takes up time I don't really have. So the initial idea was to make 3 banners (320x50) and 2-3 banners (300x250) per angle .

Thank you for the feedback

Aris

---

Seems like a great list. The only thing you still need to add is the dollar value you are going to spend on every test.
My general rule for any new campaign test is to spend between 50$ and 100$ before deciding to either PUSH or CUT the campaign. As a network we don't run campaigns ourselves because we don't want to compete with our 4.3K affiliates but that is generally what we recommend to our new affiliates as a $-treshold for a test

---

Yeah I didn't say anything about that because launching campaigns everyday or every other day will reduce my budget rapidly. I have rough estimates but nothing solid.

I am aiming for only tier 2 countries and low payouts right now so that should help out with the spend.

Thanks for the tip

Aris

---

Sure thing! If you wanna brainstorm a bit more about this just add me on skype. Hope you are gonna make big numbers any time soon!

---

I have a question about creating mass campaigns, when i am uploading banners on my traffic network i am using creaid to allocate each banner to specific landers and then setting up the rules for creaid in Thrive for the tracking, this takes a lot of time if every banner has a different creaid and i am testing 3-4 traffic sources, is there a way to speed up the process? do you set one campaign for every set of banners+lander to avoid messing up with creaid? and if you do aren't the campaigns being outbidding eachother?

Thank you for the input btw

---

I'm pretty sure you should've created another thread for this as it might require some more in depth explanations.

I use Voluum and I create a campaign for every set of banners made for a 1 specific angle and it's lander.

For example, if my offer is for Canada and I have 4 angles, I will also have 4 landers (1 for every angle) and they will look like this in Voluum:

Traffic source - Canada - Offer name - Angle 1 - mobile/site
Traffic source - Canada - Offer name - Angle 2 - mobile/app
Traffic source - Canada - Offer name - Angle 3 - wifi/site
Traffic source - Canada - Offer name - Angle 4 - wifi/app

I would do the same if it had more geo's and nothing can over lap or outbid as they are targeting different things i.e. mobile/site only for the first campaign (mobile carrier/ traffic from websites on mobile)

I hope that's what you were asking for

Aris

---

Do you have 4 random angles or is each angle tailored specifically to it's respective connection/content?

I think your strategy is good and your logic is sound. Good luck!

---

Since you're planning to use ads of two very different sizes, you may have to split the sizes into separate campaigns if you want to make use of automated bidding (depending on the traffic source). This is because tools such as Go2Mobi's auto bidder make use of data from similar ads to calculate the optimal CPM, and if you mix different sizes then you can end up overbidding for 320x50 and underbidding for 300x250 placements.

You said that you want to split-test multiple angles, multiple networks and 3-4 traffic sources. If you throw in tests for carrier vs WiFi, site vs app, two creative sizes, etc. this adds up to a lot of campaigns for each offer. With so many campaigns it may be tempting to collect data quickly, but it will be important to run each campaign long enough to make decisions that are statistically valid. So you'll have to make sure that the budget is sufficient (taking into account that some traffic sources have higher daily minimum limits), or else adjust the plan beforehand.

---

Angles are always tailored for the offer, not the connection type. The connection type applies to all campaigns so I can simply test all connection types for each campaign.

---

Yes they will be grouped up into 2 groups. The 320x50 in 1 group and 300x250 will be in another as separate campaigns. The budget is the ONLY thing standing in my way a bit so I'll have to take it slow and launch a few campaigns and see where I am standing. I need to get a few thousand impression on each banner for sure but not over do it. Affiliate networks get kind of pushy as well as they want traffic! The beginning is rough in everything you do in life, hopefully it won't be insanely bad though.

There is where im hoping (but not relying on) luck to hopefully kick in a bit and help out the situation

Thank you

Aris

---

I see a lot of variations. I'd probably focus in on a single source vs trying so much different stuff. The combination of all those and to reach statistical relevance is going to be steep and confusing.

Narrow down to 5-10 angles. 1 source and 5-10 ads for each. Based on your assumed conversion rate calculate how many clicks you need for each variation.

That's

Source + angle + ad

Also need to consider various settings like wifi vs carrier vs what carriers.

For each variation you need to get to statistical relevance and this costs money. Do a little media planning before hand based on what you think conversion rates and click costs will be so you know what you're in for to really test this strategy thoroughly.

---

Yeah, Smaxor makes a good point. Trying to test all this with multiple traffic sources may become overwhelming.

I'd say try it, but if you're starting to drown in data, cut down to the most promising traffic source and test there.

And I'd definitely recommend running some napkin math as he suggests to get an idea of what your costs will be. What sort of offers are you thinking to test with this?

---

I am focusing on app installs only.

I have the budget right now to run with 2 traffic sources and so I want to take advantage of that.

I did take Smaxors advice into consideration and the costs where overwhelming indeed, but this post was more of an idea and planning ahead for the near future.

I'll run 2 this time around and see how it goes. I am eager to test as much as possible but those damn budgets are always in the way

Thank you for the help guys

---

I would drop the two different banner sizes. Adds unnecessary complexity, especially if you are testing multiple sources.

I think the basic outline of the system is fine.

I mean, the main positive is that you actually have a system.

Most affiliates don't bother, and their efforts are all over the board as a result.

That said, it's always a good idea to reduce the variables you are working with if you are starting from scratch.

Not just for budget purposes, but for sanity purposes too.

---

Well that's something I didn't expect anyone to tell me

I am a one man army but trying to work on this like I'm 5 and I am aiming high! Since AM is a numbers & testing game, that's what I was aiming to do. Hence trying different traffic sources / banner sizes. But from your post, I am getting another picture right now and maybe things aren't so complex as I think they are when it comes to launching campaigns.

Are you saying I should be focusing on 1 traffic source, 1 banner size and done? Scale only when you have statistical significance or drop it if you don't?
What if the green is at another traffic source and on another banner size? Does that not mean i should be testing as many traffic sources and banner sizes as possible?

I'm starting to understand that there is 2 systems I can follow here and one is a headache, the other is quick and cheap.

1st way: Everything I stated above in my first post. Test the hell out of things with various traffic sources and banners sizes.

2nd way: Follow the plan above but keep it at 1 banner size and 1 traffic source. Use the extra money on testing landers/more banner variants, and scrap if it isn't working. Don't move on to other banner sizes, don't move on to more traffic sources until I can scale.

2nd way is ideally what a quick test strat in mind would be but if it's not optimal or never gonna find me that $1k per day campaign, I don't want to do it.

Thanks a lot for your reply, highly appreciated.

Aris

---

Home > General > Affiliate Marketing Forum