Home > Mobile >

---

My Process for Testing Offers and Finding Profitable Campaigns (46)

---

After receiving several positive comments and a request to repost my information in a separate thread, here it goes. I do suggest reading through the original thread which sparked the inspiration of what I'm about to share but I have also expanded it a bit in this thread.

Original thread: http://stmforum.com/forum/showthread...o-they-convert

I'm going to share my thoughts and approach to testing offers to find a profitable campaign.. doing so in the shortest amount of time and while minimizing losses. In no way am I an expert and there is certainly more than one way to approach testing offers but I hope my experiences can help others. I do have a fair amount of experience in AM but have been out of the game for several years. In mid-July of this year, I started to get back into launching campaigns and my starting point has been with mobile pops. Prior to July, I had never reached more than $10-15/day profit with mobile pops.

With the help and advice of Amy, matuloo and others.. including the vast amount of information spread across STM, I have been able to build two mobile pops campaigns that each hit the $300/day in profit mark and several smaller campaigns. While this isn't huge success compared to some, in my mind its great progress from July to present day because I now have a strong understanding of a process to build a profitable campaign.

Main Takeaways in this Thread

Tip 1 - Test fast and efficiently (the more campaigns you launch, the better. have a process for setting a test budget)
Tip 2 - Don't limit yourself, be flexible and test everything. Break out of your comfort areas and don't rely exclusively on ripping. (test more offers, landers, geos and get translations, test your own angles, etc)
Tip 3 - Break your "Testing" into two phases.. an Initial Test Phase and then an Actual Test Phase
Tip 4 - TEST EVERYTHING. With your first goal being to find a converting offer and then mass test GEOs and Landers.

So here we go... from what I've learned on my journey so far, sticking to just 1-2 geos in the beginning is not a good idea. Not that it can't work but I originally kept my focus on just a few select geos that I felt would be a good starting point and never made much progress.

My tipping point was when I began testing several geos for one vertical, picking the one that had the best results and continued with it.

Recently, I have been sticking to 3-5 geos and working mostly between those ONLY because I've had success with them previously and understand how to make them work. But, I still test other geos regularly.

For the best chances of finding a profitable campaign, I recommend these options:

Option 1:

Step 1) Pick 1 Vertical (like sweepstakes, antivirus, etc).
Step 2) Pick 3-5 GEO's.
Step 3) Grab 2-3 of the most popular landers on AdPlexity for each GEO you are testing.
Step 4) Gather and test EVERY OFFER you can find for that vertical in those geos.
Step 5) That should give you atleast 5 campaigns to test broadly.
Step 6) Review the stats in detail, if there are any combinations that look promising, then test it further with more landers and offers if possible. (a promising combination is a good performing OS or Carrier or a few high volume placements that are doing well)
Step 7) If none of these campaigns show potential, then repeat this whole process but choose 3-5 new GEOs for that vertical. Keep repeating until you find a winner or until you've tested as many geos as you can.

Option 2:

Step 1) Instead of picking just 1 vertical, you can pick 2 GEO's and test every type of vertical/offer you can find for them.
Step 2) Group and test the offers by Vertical. So gather all of the Sweepstakes offers and test them for those 2 GEOs, gather all of the Antivirus offers, gather all of the gaming offers, gather all of the app install offers, gather all of the pin submits, gather all of the 1 click billing offers, etc.
Step 3) Gather 2-3 landers for each offer type per GEO.
Step 4) Test and repeat in as many geos as necessary until you find something that works for you. Follow the steps outlined in the previous Option 1

TIP* - Your first steps SHOULD NOT focus on testing just 1 carrier, 1 geo, 1 os, etc. Although that can work, you need to be sure of your strategy and when you have little experience this isn't a good approach.

TIP* - The most important part of this entire process is finding an offer that converts well.

I start my testing very broad, sometimes I setup a test campaign that targets all OS types and both wifi/carrier traffic.. and othertimes I setup two test campaigns for the same offers/geo while targeting all OS types and split wifi traffic and carrier traffic into their own campaigns. The best approach is probably to split wifi/carrier into their own campaigns but there are some cases it's not needed unless you really want to squeeze every penny of profit you can from a campaign. Run a bunch of traffic to the offer/landers, review the stats and decide how I want to proceed. Sometimes if I see a certain segment converting well from the start, I will change my targeting to just focus on the best segments (but I only look at an OS or Carrier level, nothing more granular than that). This is a good approach if you are in a high volume geo and/or trying to be efficient with your budget.

TIP* - Change the way you think about testing. Break the "Testing Phase" into 2 parts.

Initial Testing Phase - The whole goal of this phase 1 of your testing is simply to find offers that will convert more than once.

Actual Testing Phase - The whole goal of this phase 2 of your testing is to take the offers that you've found to convert and focus on running more traffic to them and the best target segments, test more landers and test more offers. This allows you to give more budget to test just the best combos, revealing whether or not the campaign has any real potential.

In the Initial Testing Phase you need to test every offer you can find for your vertical/geo with the goal of finding offers that convert more than once. Expect to lose -75% or more of your testing budget (with occassional exceptions). Set a specific test budget for this phase. Anything that doesn't make conversions (preferably more than 1 conversion) after your test has completed, cut them.

A general budget formula that I use for this phase is:
(Average Offer Payout) x (Number of Offers) x (Number of Landers) x (3-5 times) = Test Budget

Next, the Actual Testing Phase is to further test the offers that you found to be converting. This is where you actually determine if the campaign has potential or not. The Initial Phase just helps you identify what GEO / Offers / Landers / Vertical you should test more. The Testing Phase is where you isolate what segments, GEO's and offers are converting and spend more budget on traffic to see if any of them really have real potential. The purpose of this mindset is, if you stopped testing after your initial test, you very likely spent much of your budget on offers/landers/geos that do no perform well and the best offers didn't have much of a chance to show their real potential. This approach ensures you fully test the best offers you find by sending an adequate amount of traffic from only the best segments. Expect to be losing -50% or more of your budget, but your best offer/lander may be doing better. At this stage, is when I take offers and landers to statistical significance before cutting them.

Next, if something has real potential then you further test those segments, offers/landers, etc. Test more offers, test more landers, test more geos for that offer if it accepts it, etc. Otherwise, go back to the Initial Phase and test other campaigns. At this phase you should be losing -25% to -50% overall but since you should have found a solid offer by this point and and testing many landers, gathering data and finding what placements and targeting is or is not working at all. Your best offer/lander should be atleast -25% or better.

Finally, if a campaign takes you to this step you are now optimizing a campaign to make it profitable. You should already be near breakeven or better. At this point, you cut any bad performing OS types if you have not already, cut any placements that are worse than -15% ROI, adjust the bids (if possible) on placements that are -15% to 0% ROI or move them into another seperate campaign with a lower bid, cut any mobile carriers that are not performing well, etc.

Recap on What's Important

1. Finding an offer that converts is priority #1, the more offers you test, the better your odds of finding and offer that works great.
2. In my opinion, the next most important piece is actually the GEO. The more GEO's you are willing to test, the faster you will find a GEO that kicks butt.
3. The next most important piece is the Lander (although the GEO and Lander are almost tied in terms of importance). If you've found a solid offer, the Lander will likely push you into profits. And if it doesn't hit profit in GEO #1.. try GEO 2, 3, 4, 5, 6, 7, 8, 9 or 10. One of them will make profit.
4. In your Initial Test Phase and Actual Test Phase, review your stats carefully. If 75% or more of your conversions are coming from 1 OS or 1 Carrier.. you may want to consider changing your target settings to focus on just those segments for now.. and come back later to test the others. This is a good approach to reaching profit faster but you will lose traffic volume and in some GEO's cutting Android will eliminate 90% of your traffic, so be mindful of that.
5. The more willing you are to test many GEO's, get translations, etc. I would estimate that you increase your chances of finding a good campaign by 50%.
6. Commit to launching X number of campaigns.. such as 2 per day or 10 per week. Within 2-3 weeks you will have found several profitable campaigns and possibly 1 campaign that will achieve $xxx/day or more.

Note* - My interpretation of testing one campaign includes testing one Geo and one Vertical, no matter the actual number of campaigns I setup to test it. So, even if I split wifi and carrier traffic into seperate campaigns for GEO #1 and the Antivirus vertical, that is one campaign in my mind. I'm just setting up several actual campaigns to ensure it's thoroughly tested.

Note* - I've also started to incorporate the concept of testing multiple traffic sources for each campaign as well. Because Traffic Source A may get zero conversions but Traffic Source B gets 10 conversions. Initially I suggest focusing on one traffic source, then eventually start testing several sources per campaign as I'm starting to find more campaigns that way.

I hope that helps someone and glad I can contribute in some way after all of the support I've received. Stay tuned for a case study I am preparing to post soon.


Andrew

---

Awesome stuff mrpayne, it's a pleasure helping guys like you and actually watching them succeed. Great post, thanks for taking the time to write it!

---

This is gold! Thank you for this, very similar to my system as well.

---

Thank you very much for the golden post, well I was reading your follow along for two or three days before, and as a newbie I kept telling myself he is a master and he calls himself a newbie, I liked the way how you optimized your landing pages at first and then how you optimized your campaigns.

Now with this post you lifted some of obscurities.
I had a conversion with MrPayne in PM and he proposed to continue on the thread, so his advices will be helpful for other members.

I asked him about this setup and how much do I need to spend

Camapaign 1 : Malysia, three offers, three landers, three trafic sources
campaign 2: Indonesia, three offers, same three landers, three traffic sources
campaign 3: Vietnam, three offers, same three landers, three traffic sources

And the amount that I will spend will it be divided equally on the three traffic sources??!


So he proposed to stick with one traffic source at the begining, and to focus on offers rather than landers so he proposed to get 5 or plus offers with 2-3 landers.

I supposed that if the payout is 1$ how much do I need to spend on each campaign and this was his answer

I already shared in my post a basic formula to use for the test budget...

A general budget formula that I use for this phase is:


(Average Offer Payout) x (Number of Offers) x (Number of Landers) x (3-5 times) = Test Budget

If you have 3 offers that are all $1, the average payout is then $1.

$1 x 3 offers x 3 landers x multiplier of 3 = $27



Now my basic questions are:
should I run the campaigns for more or less than 24 hours?
Bids: high, avearge or low?

Thank you very much

---

The length of time is not as important as the time of day you run the campaign in your respective GEO. I recommend testing between the hours of 6pm to midnight in the timezone of your GEO. The purpose of that is so you get the best chances of getting conversions. That being said, I try to have it run for 3-6 hours.

Initially, I just do an average bid x 1.5 = starting bid

So if your traffic source says the average bid is $1 CPM, which is .001 bid per view, I would launch my campaign at .0015

Sometimes, traffic is coming in slow and I will increase the bid a little further.. to something like .002 in the example above.

Another method that I use is I adjust my initial bids based on the speed of the traffic I am getting. I will set my bid, turn the campaign on, and write down how many visits and the accumulated costs I get during a 60 second period (1 min). Multiply that by 60 (gives your hourly estimates) and multiply by 360 (gives you 6 hour estimates) to get an idea of how many impressions and the cost it would take to make it span over a 6 hour period. I will then adjust the bid, wait a few minutes and the write down the stats for that bid. Do this until I find a good bid to spread my budget across. This isn't always a good approach and you should probably get some experience with the first method until you know what to look for when using the 2nd method.

Assuming my campaign test budget is set to $50 and to illustrate what I mean by the 2nd method...

If I set the bid to .001 and my estimates are in:

1 min, 200 visits, .20 cents
1 hour, 12k visits, $12
6 hours, 72k visits, $72

but if I set my bid to .00085

1 min, 120 visits, .10 cents
1 hour, 7200 visits, $6.12
6 hours, 43k visits, $36.72

I would set my bid to the .00085 range because my budget would last around 6-9 hours. Again, I don't do this for all campaigns and I find that it this method works better for some traffic sources and certain geos. Otherwise, my default is to take the average bid x 1.5 or 2 = test bid

What you are looking for in the initial test phase is simply offers that convert, preferably more than 1 conversion.

So in your initial test of 5 offers and 2 landers.. your results may be:

Offer 1 - converted 1 time
Offer 2 - converted 0 times
Offer 3 - converted 0 times
Offer 4 - converted 2 times
Offer 5 - converted 3 times

Lander 1 - converted 2 times
Lander 2 - converted 4 times

Carrier 1 - converted 3 times
Carrier 2 - converted 3 times
Carrier 3 - converted 0 times

I would take offers 4 and 5, both landers and Carrier 1 & 2.. continue testing those with more offers and more landers.

The whole concept is to focus on testing quickly, find whats converting easily and exploit it. It's not the most accurate approach but helps get momentum in your campaigns.


Andrew

---

great post thank you very much... I just need a quick clarification though:

Let's say I pick 13 offers from Zimbabwe or whatever. 7 offers are for antivirus, with 2 landers, 4 offers are iphone 7 offers with 2 landers and 2 iphone 6s offers with 2 landers as well. My average payout is $0.73.
So my budget will be: .73*13*6*4=$227 give or take.. correct?

Which option would you recommend the most for a beginner?

---

In your example, I would break things down to several campaigns...

Campaign 1 - Zimbabwe Antivirus
Take the average payout of just the antivirus offers. Let's say the average is $1.25 and you have 2 landers

$1.25 x 7 offers x 2 landers X multiple of 3 = $52.50
$1.25 x 7 offers x 2 landers X multiple of 5 = $87.50

So your test budget should be between $50-90 dollars.

Campaign 2 - iPhone 6s Offers

Get more than 2 offers. Then repeat the formula above to calculate the test budget.

Campaign 3 - iPhone 7 Offers

Same as the iPhone 6s offers.. get more than 2 offers. Then repeat the formula above to calculate the test budget.


I'm not sure I understand what you mean by which option I recommend, what are you asking?

---

I was just asking which option in your first post is more suitable for a beginner (going for 2 geo and mass test offers in all verticals) or 1 vertical across 3/5 geo. But I guess that doesn't make much sense.. Do you have a preferred option?

Thanks a lot for this, really helpful

---

Either work fine. I would probably suggest Option 1 because you are focusing on one vertical at a time. Once you have your first success with a vertical, it's easier to replicate success within that vertical. Eventually you can start testing other verticals and make them work.

But... originally I was focused on the sweepstakes vertical and never got much progress, although I didn't test widely enough at that time. Eventually I started testing mobile apps and other offer types, that's when I had my break. But I've revisited sweepstakes and have had some ok results since then.

After my first success, I focused almost exclusively on that vertical for a while and replicated another campaign in that space before venturing back into the sweepstakes vertical again.

Now I've been testing several verticals and offer types I've previously avoided while on my pursuit to achieving stable revenue.

---

Here where I can't get it from two points:
- If I find two offers that are qualified for the actual test phase, should I test these offers with landing pages that I already cut or I need to rip some new ones, same applied with orher geos and careers ?
-If i test 5 offers with four geos and two landing pages and the minimum payout, I need to assign 30 $ x 4= 120$ - isn't too much for initial phase?

---

Initial Test Phase - Find offers that convert. Don't cut anything except offers that do not convert. Look at the data from this test and make a decision.
Actual Test Phase - Round 1 - Send more traffic to the converting offers and spend enough to take them to statistical significance before cutting that bad performing offers. Then you start testing more offers if you still have offers to test. Only cut poor performing offers when reaching statistical significance. Only cut landers if it is getting less than 0-25% of the overall conversions and just keep testing with the 1-2 landers that are converting. If all landers are converting about equal, don't cut any of them. Only look at the data collected during this test, not a blend of the data from the initial test and this one.
Actual Test Phase - Round 2 - Take your best 1-2 offers and test as many landers as you can rip. 10, 15, 20 however many you can find. I've tested as many as 30 landers for one campaign. Bring each lander to statistical significance, only looking at the data from the Round 2 testing. Make your decision.
Actual Test Phase - Round 3 - If your best offer and the best lander still are not performing well enough but are getting steady conversions, then look at your targeting.. if you haven't already.. and see if there is any OS type or Carriers that stand out. If not, then I recommended getting your best 1-2 landers translated and test the offer in a different GEOs. If the offer does not accept other GEOs, then you need to start all back over with a completely new vertical or test new GEOs with new offers, repeating this all over again.

I think the problem you are having is you think that the 'testing' part is just a one step process but it is not.

You have to test every major variable, one at a time. Find the best offers.. then test to find the best lander while using your best offer.. then test and find the best GEO using your best lander/offer (or you can test GEO's before doing the landers but you will need to translate or rip new landers for testing).

Testing = Collecting data and isolating the best performing metrics. And one step at a time you will get profitable.

Every campaign should only target 1 geo and 1 vertical. Do not mix them up.

So if Campaign 1 is in Brazil and you have 5 offers and 2 landers with an average payout of $1

$1 x 5 offers x 2 landers x 3 = $30
$1 x 5 offers x 2 landers x 5 = $50

Your Initial Test Budget needs to be between $30-50... this is solely to find an offer that converts.

And take that into consideration with your overall budget. If you have $1k per month in available budget and you minus Voluum, STM, domains, lander work, etc.. which is approximately $350/month.. leaving your $650 in advertising budget.. divided $650 / $50 (initial test campaigns) = 13 initial test campaigns.

You would be better off testing offers with a payout of .50 or less.. that way you have enough budget to do atleast 26 initial test campaigns, greatly increasing your odds of finding a profitable campaign and accounting for some of your losses being associated to you being a newbie.. you will make mistakes along the way that will cost you money = paid learning experience.

---

Another fantastic post from Andrew! *APPLAUSE*


Amy

---

Yea, another excellent post. Thank you so much Andrew!

---

This is alone is gold!

---

Such an awesome contribution @mrpayne . Thank you

Just wondering how many offers you have listed per campaign?

Would it be the same and landers etc, where you would distribute the traffic accordingly to go to those chosen offers? Or is it one offer per camp?

Cheers mate

---

I'm not sure I fully understand your questions but... I typically test as many offers as I can find and keep the best 2 but I usually just run the best offer by itself so I can profit more while the campaign is running. If the best offer gets taken down I will run the 2nd best offer or test more offers. Same for landers.

But... a wise man gave me slighty different advice in his reply to one of my questions, you should probably read his whole thread! (thanks matuloo) I haven't fully tested his advice but I plan to because all of matuloo's advice has propelled me forward, it's solid stuff!
http://stmforum.com/forum/showthread...l=1#post283834

Hope I answered your question.



Andrew

---

Hi Andrew, thank you for replying

I'm still learning how to be more specific when asking a question lol

What I meant to say or try to ask lol was

What would your average funnel look like within your tracker?

I noticed you had mentioned for option-1 to test a couple geo's, 2-3 landers and say 2 offers.

Just wondering in your tracker, would a singular campaign funnel look something like this

Camp 1

Traffic - lp1, lp2, lp3 - offer 1

Camp 2

Traffic - lp1, lp2, lp3 - offer 2

Or would you create some rules in your traffic, so that in a singular campaign, traffic will be going to both offer 1 and 2??

My apologies, I hope I made more sense this time lol, I think my main question would be regarding the funnel system through the tracker.

Regards james

---

@James - I understand your question now.

Option 1, I suggest focusing on 1 vertical, several offers, a few landers and do that for 2-3 geos.

So for example...

If I chose the Sweepstakes vertical..

In Australia.. Brazil.. Canada..

My campaigns would be like this..

I would gather as many sweeps offers for Australia as I can
I would rip 2-3 landers for Australia

Setup Voluum rules to.... Buy traffic > Send traffic and rotate between all the landers > rotate all the offers
After test, analyze results.

Do the above process for each GEO you are running a test in.




Andrew

---

Thanks @mrpayne for clearing that up

This process is aimed at mobile pops for time being??

Cheers

---

Yes, but the same concept is applicable to other types of traffic.

---

The key is to have a disciplined approach to testing and managing risk, and not just to panic and jump off the ship when the first campaigns look red.

That is how you kill the bull, and keep the bull from killing you.

---

How do you do that if your traffic source does not let you target by Carriers? Referring to Facebook here

---

You don't have to target by carrier, that was just an example.

You could be tageting by any major segment, age ranges could be an example.. 18-21, 22-25, 26-30.. and of those, only 22-25 is showing good results.

I'm not too familiar with Facebook so others may chime in.


Andrew

---

Question in regards to your formula...
(Average Offer Payout) x (Number of Offers) x (Number of Landers) x (3-5 times) = Test Budget

Shouldnt this also depend on traffic source? EG. I know Native Ads can be a bit pricey in comparison to other networks. Would you still recommend the same test budget?

---

Yes, because the formula is based on your average payout. If you are promoting the right offers for your type of traffic, this formula would serve as a fair estimation of what your initial test spend should be... but you are not limited to strictly this budget, the formula is a baseline to determine your minimum spend required for a adequate test.

From what I understand about native, you typically want to promote higher payout offers that are north of ~$20. Unless you're referring to mobile native in which case I'm not sure what the rule of thumb is for offer selection.



Andrew

---

Thanks a lot sharing this, really really helpful

---

I got the question in regards to the optimization process. So I understand that ideally the test budget should approximately based on the formula below.

However, in reality, you can't make any assumptions unless you have some conversions. So say you run your test budget and only get 1 conversion, do you continue running the campaign for another day? How do you determine when to cut it.

---

Your first goal is not to cut anything.. its to find converting offer(s).. so if your initial test only got 1 conversion, you either need to run more traffic to the test campaign or if you've ran more than enough traffic and its just not performing, it's time to move on to the next test.

1 conversion doesn't tell you much. The formula above is just used to see which offers are converting.. then you will have to spend more money to identify the best offer/lander.



Andrew

---

Excellent points and I have one quick question about the early testing phase. You said this:

So you're saying first test to find the best offers, then test to find the best landers.

But during the 1st step should I be using landers at all? I assume direct linking doesn't work so well most of the time. But how do you decide on a lander when you're only testing offers?

More specifically, let's say you test 5 offers each with the same lander. What if the lander sucks and drags down all the offer CVRs? Maybe one of those offers was good but your lander screwed with the results. Is it better to start testing offers via direct linking and then get into landers? Or which lander(s) should you use during the very 1st step when you're only testing offer quality/conversion rate?

---

You're missing the overall point.. your initial goal is to find an offer that converts. Then your next goal is to focus on optimizing one key variable at a time. Starting with the offers.

Step 1 - Test a handful of offers (let's say 5 of them) and grab 3 or so ripped landers. If there is one or more offers that are converting and showing a little potential after your initial test, then go to step 2.
Step 2 - Test even more offers for this vertical/geo and test it with whatever landers were giving you good results from step 1. Then run enough traffic to find 1-2 clear winning offers. Then step 3.
Step 3 - Test as many landers you can find along with your best offer(s).. run enough traffic to isolate 1-2 winning landers.
Continue from there...

You can form your own process, this is just to demonstrate how to test one major variable at a time, prioritizing the most important ones first.

Most offers will not be applicable to direct linking.



Andrew

---

I think I understand the main point, but I'm missing in step #1 where you say "grab 3 or so ripped landers".

Grab them from where? Adplexity? This assumes the person following this guide can study competition and rip their landers. My question is, let's say a newbie does not have access to any landers to rip. Should they build a lander to test? Or is it better to try direct linking to test offers first?

It seems like in step #1 we're testing two things at the same time:

1. testing the 5 offers
2. testing the 3 ripped LPs

So we don't know if the 5 offers are good, AND we don't know if those 3 LPs are good. The only way to know is test.

I agree with you about testing one variable at a time. But if you can't direct link offers then you need to test all offers with LPs. Ripping them is easy when you have tools. I suppose the tricky part is for newer affiliates who need to create their own landers and hope some of them work well enough to keep testing.

Excellent thread BTW you do have a good process here.

---

I advise that you get some competitive tools, like AdPlexity, and rip some landers. You will save yourself alot of time and money by doing so. It's easier to take landers that are being seen everywhere on the spy tools because they are likely to be working, the main step here is to determine which of these landers work best... after you find a good offer.

For step 1, you are testing multiple items which is why it's important to spend enough money and reach statistical significance for each variable... BUT YOUR GOAL is to find a converting offer. Without using atleast a few landers you are almost dead in the water right out of the gate.

Most offers will not convert without a lander. And while your initial goal isn't to narrow down to a lander, you do need some landers to start testing the overall campaign with. If you spend an appropriate amount on the test, each lander and offer will have had an about equal amount of traffic to each of them and give you a clear idea of what is working and what is not working.




Andrew

---

There are some minimum requirements to get started in AM, and good spy tools are definitely one of them.

You CAN learn without them, but it's a vast amount harder, and you'll waste a lot more in spend on data to reinvent the wheel than it would cost you to get six months' subscription to Adplexity or equivalent, learn from the competition, and then iterate from there.

---

Thanks caurmen & payne great info. I suppose I'm still studying it all, but even after reading the onboarding guides I never saw spy tools being a necessity. I'm more than willing to add one into the budget if you say they'll make a difference for beginners, I trust both of your judgements on this one.

And of course it makes following the tips in this thread a lot easier.

So much to learn about this industry and it's amazing to have such helpful people in STM who share their knowledge for free!

---

Great stuff @Mr Payne. Implementing this into an interactive checklist and mind map right now as type...

---

Hi Andrew,

I'd like to say that this thread is more than gold ! It totally unlocks my long-time-misunderstanding of how to test properly. I'm not saying I have success with this method but at least I gives me ideas to strategize my testing process before running out of budget.

At this moment I think I've passed the phase "finding a good offer" since it converts 6 times with -70%ROI. One lander (the winning lander) converts 4 times so I decide to put another 2 more landers to compete with that one for this offer.

In the mean time, I've set up 2 different campaigns to test 2 other offers in the same GEO with that winning lander. Since the 2 offers have different rewards for audience so I have to modify the winning lander and set them in 2 seperate campaigns.

My question is if I set my 3 campaigns have the same target in the traffic source (let's say PopAds), same bidding. Am I competing against myself ? And is it a good set up ?

And another question is when you say I should test deeper on the converting offer, for example I want to split test the Carrier of that combo best offer + best lander. What is the approriate bid should I set ? Is it equal to the Intial Test Phase or I should use the increase the average bid for the traffic 1.5-2 times like your method ?

Many thanks
Tien

---

Thank you for the wonderful post. Should we stick to just 1 traffic source ? Also what source do you recommend for the initial test phase?

---

1. Why not test it all under one campaign to see if the other two even have potential? In Voluum, you can setup a flow to split traffic to three segments.. Offer 1 > Lander 1 gets 33% of traffic.. Offer 2 > Lander 1 (modified) gets 33% of traffic and Offer 3 > Lander 1 (modified) gets 33% of the traffic. This way you aren't competing with yourself.

2. There is no exact method to testing bids or segments. If an offer has potential and is converting on both wifi and carrier traffic, I would split the campaigns up into one that targets wifi only and one that targets carrier only. Test various bid levels and find which level converts the best before further optimizing.



Andrew

---

You don't have to stick to just one source, I recommend starting with 2-3 sources in the beginning.

There are lots of good sources to use but PopAds, Zeropark, PropellerAds and PopCash are relatively good starting points.



Andrew

---

It is a very good stuff Mr Payne! Specially the flow chart what makes clear a lot of things for me.

One thing is not clear here. If I choice a vertical, lets say sweepstakes, there are a plenty of offer with different prize (iphone, tesco coupon, sasmung phone, etc).
So do we need grab landing page for all of these prize theme separately?

Thanks in advance

---

I recommend focusing on the offers by groups.. test all of the iPhone offer/landers, then test all of the tesco offer/landers, etc. until you find a winning combination or move on to the next geo.



Andrew

---

Hey Mr.Payne, great post! This has already opened up my mind and noticed that I was having a bad approach while testing and validating offers.

I've got just one question though: For each of the testing phases, do you build up additional campaigns or you continue testing on the same one for all the phases? What I mean is this:

1) let's say you started initial test phase with 6 offers an found out that 2 of them are converting more than once. So you ditch the other 4.

2) In the actual test phase (round 1), do you create another campaign with the 2 converting offers or just keep testing using the original campaign?

This probably looks like a lot of work, but in my mind the approach I'm suggesting is much easier on the eye, as you would look at data from each testing phase only.

Hope I was clear enough with the wording

Thanks!

---

I apologize if this question has been answered, but if step #1 is to identify the top performing offers wouldn't it make more sense to keep the LP exactly the same in each geo? Then once the best offers are found change up the LP and so on.

---

1. I only keep running traffic to what is converting. So long as you run an adequate amount of traffic in the initial test phase, this usually plays out well.

2. You can keep running it in the same campaign but only look at the data from that point forward.



Andrew

---

If you have a lander that is already known to be good, then yes, that is fine.

However, most of the time in a new geo/vertical you may not a proven lander to work with, you are shooting in the dark. By using multiple landers you give yourself a better chance that atleast one of them is decent enough to convert the offers. Then later you will more thoroughly test the landers.


Andrew

---

Can't we also split test offers on Voluum? With Voluum we can create a workflow that will send different offers on a percentage bases. And with Voluum you can create rules, for example "if WiFi send to Offer A" if 3G Send to offer B, etc. I believe you can split test multiple offers at once this way.....Is this correct?

---

Home > Mobile >