Home > Mobile > Follow-along Campaigns

---

"I'm gonna make it work" - low $xxx in 3 months (23)

---

Hi everyone,

My name is James from Australia, I've been an STM reader for quite a few months. I launched mobile pop campaigns here and there but due to lack of dedication (9-5 job + family), they did not go anywhere. I therefore start this Follow-Along thread to commit myself into giving this a really really serious effort.

I'm quite used to picking up offers, ripping up landers, launching campaigns, I had no problem getting conversions but still struggling with optimization, let alone scaling.
I will actively update my follow-along thread, hopefully you can provide me with valuable inputs to guide me to profit.

This time I'm gonna make it work with all of your help.

My target is to achieve high $xx or low $xxx/day in 3 months. (Please wake me up if you think I'm day dreaming )

My gear:
1. Voluum
2. Adplexity Mobile
3. Amazon AWS S3 + CDN

My budget: a few $1000s as suggested won't be a problem to me.

Vertical: App Install (Android boost + clean). My AM advised me to start with app install offers because they are easier to convert and don't cost user money like other pin submit/mobile subscription offers

My strategy:
- I will start with an English speaking GEO, 8 landers, 1 offer (high EPC in Affiliate network dashboard).
- I will narrow down to 3 landers and test the other 2 offers.
- I will then have the 3 landers translated and run traffic on other non-English speaking GEOs.
Please correct me if you think I'm making any mistakes. Even with choose offer verticals.

My actions:
- My AM recommended 4 app install offers. 3 of them are android boost/clean apps. The other is web browser. The offers are international with payout varies with GEOs.
- I focus on the 3 android boost/clean offers because I think they are very similar and can use the same landers.
- I ripped 16 relevant landing pages from Adplexity (English), chose 8 then cleaned and fixed them. I also made amendment to those landers in a way I thought would improve them (change picture, rewrite the words - Am I wasting my time?).
- I then had my AM approve those landers and launched a PopAds campaign in an English speaking GEO (payout $0.64)
- Target: android mobile phone, BOTH wifi + carrier (include all connection speed and ISPs)
- I adjust my bid so that $40 will be spent in a day (i.e. $5 spent on each lander).
- I will come back with campaign stats and keep you updated.

Again, it would be wonderful if you can correct me in case you think I'm making any mistakes, I'm all ears

Thank you everyone.

---

Looks like you've got a solid plan! For optimisation, if you haven't already, I'd highly recommend looking at the following threads regarding "statistical significance":
https://stmforum.com/forum/showthrea...nificance-quot
http://stmforum.com/forum/showthread...Banners-Part-2

The 2nd in particular has a link to a tool that works with Voluum data directly, making the optimisation / decision-making process a lot easier.

Regarding the landers, I don't think those changes are a waste, as long as they aren't drastic changes from the original. Once you narrow down your landers to the best performers, you can start using multivariate testing to further optimise, and can re-include the original images/text.

Good luck!

---

Thanks for starting a follow-along to document your journey!

This is a realistic goal if you're talking about a single campaign. However, pop camps usually don't stay profitable for last long - a few days/weeks maybe. So to be able to make consistent profits, it would be necessary to have a system where you're always testing to find new profits to replace old profits from dying camps.


EPC/CR stats provided by aff networks are often inaccurate, or may not actually be averages, but results achieved by one top affiliate. So, I always take them with a grain of salt.

It's good to start your testing with recommended offers (like what you're doing), but once you have a good lander, testing other offers like you're planning to, would be great.

Why not just cut down to the last lander first, and then test the other 2 offers?

I can't pick faults with your process though - so maybe let's just proceed and see what happens.

Here's the recommended method for cutting landers:

https://stmforum.com/forum/showthrea...Banners-Part-1


Sounds like a good plan! Eager to see stats when you have them!



Amy

---

Hi Amy,

Thank you for your swift response, very appreciate it. I will get back to you tonight when I have time.

Campaign 1 details: (please correct me if any option is not appropriate)
- Quality: top 60%
- Freq: 1/24
- PrimeSpot only
- Referrer: Standard (don't know what it means)
- Adblock: non-adblock only
- Adv. type: Popup (NOT popunder)
- Max bid 0.0035
- Daily budget: 5 (I come home realizing my budget is depleted, traffic stopped for a few hours, should have allow a bigger daily budget).
- Categories: both adult and mainstream
- OS: Android only, all browsers.
- Device: smartphone only.
- Connection type: all, connection speed: all, ISP: all
- Summary: queue 22, estimate 12000 impression per day, i.e. max daily spent: 12 x 3.5 = $42 (~ $5 per landers per day! spot on).

Now some updates:

Campaign 1: 1st day update (mid-day)
- I come home realizing there are 5 conversions (total revenue $3.2), but Voluum does not show any revenue. Turn out I did not have my global URL postback setup with the affiliate network (I thought I did but I was wrong) Mistake learned!. I quickly ask AM have it fixed.
- Total spent: more than $9, DOESN'T MAKE SENSE as I set daily budget to $5. Don't know why. Maybe it's the timezone setting that I was not aware of. Is it possible to setup PopAds timezone to match with my local?
- In my case, Voluum miss tracking those 5 conversions. Is there anyway to know the conversions' stats (i.e. landers, website ID, categories...)?

Thank you. I'll provide more detailed update when it's the end of the first day.

---

Hi everyone,

Campaign 1: end of day 1 update
- Recall: 1 offer, 8 landers, payout $0.64, English GEO
- Spent $20
- Conversion: 10 (yay) --> revenue: $6.4 (not too bad considering no lander cut, no optimization yet).
- Issue: Voluum does not track the first 5 offers due to forgetting setting up postback URL.

Question 1: how do I update Voluum with the first 5 conversions? Is it possible? I looked at conversion report in my affiliate network and there are 'Conversion ID' and 'SubID2' (which is clickID). However, when I drilled down the campaign in Voluum, there's ImpressionID, which is different from clickID, unfortunately. Maybe I just need to move on and accept that my mistake lost me some precious data!

Question 2: I know how to update the cost in Voluum, i.e. select campaign, click 'Update Cost' then fill in the total spent for the period of time. However, is there a way to update the cost for individual websiteID. Is it worth doing so? (I saw someone else update the cost for individual target on Zeropark with FunnelFlux).

A few screenshot:

1. Report on PopAds by day:


Question 3: why PopAds does not track conversions? Is there a way to make it do?

2. Voluum Report: Landers (note that there are 5 conversions missed by Voluum, and it was my fault)


Question 4: PopAds records 6847 impression while Voluum only sees 5778. Is it normal?
Question 5: Should I dump lander 14 at this stage, 742 visits and only 2 clicks.
Question 6: 2 landers (10 and 16) have more than 100% CTR, I see other newbies reported this, maybe it's something to do with the <script> isn't it? I'm not fascinated by CTR by the way

3. Quality scores: to me there's not much to see here.


4. ISP/Carrier: don't know what to do with this data.


5. Mobile Carrier


Question 7: I think this campaign has both wifi and carrier traffic. How do I know which placement has high percentage of bot click? The 'other' in the screenshot is apparently wifi traffic with 90% of impressions. What should I do about it to optimize the campaign (in case I find the winning lander).

That's it for today.
I'm gonna send traffic for another day and will review and update tomorrow.

Thank you for reading such a long post with too-many questions.

Question 8: Oh, my bidding queue and estimated impressions drop compared with yesterday (yesterday 12k vs today 7k impression) although I haven't change anything. I guess it's normal isn't it?

Any advice would be greatly appreciated.

---

rolandb, thank you very much, I will definitely give these posts a good read.

Amy, I'm trying to see some profit first as I never did. Would be wonderful feeling wouldn't it? What sort of campaign would you think will make consistent profit? Native, facebook, mobile display?

Certainly, I would do the same

Of course I will, cutting down to the last lander it is

Thank you, I will give it a good read. and will keep you updated with the campaign.

Thank you for getting back swiftly.

---

Nice start to your follow along...

Here are some of my pointers. If you go with the utility app offer you will find that it's easier to get results but to do it right and retain profits you have to protect your landers/assets correctly.

What I suggest you do with the utility apps is to test wide... pick 2-3 traffic sources and test the utility apps in 5-10 geos per source, most apps accept multi-geo or they have similar apps for other geos. Results will greatly vary by GEO and source, then only focus on the ones giving you the best results early on.


Cheers!


Andrew

---

Thanks, Mr Payne for your advices

Can you give me some insights on how I should do this?

Great information. I haven't thought of that. Should I test wide once I found my winning lander or put all 8 landers in every campaign? Winning lander in GEO1 will likely be winner in GEO2 won't it? At least won't be too bad, I think?

---

Focus on getting profitable first, then come back to learn how to protect those profits - take things one step at a time.

Landers perform differently per geo and per source.. what I like to do is to grab 3 landers from AdPlexity per geo and test them on one source. If something works, then great, if not, then I move on to source #2 and test those geo's with the landers or if I've identify a good lander from the previous test, I will only use it for now.



Andrew

---

$3.20 revenue for $9 spent is looking hopeful, especially considering you're testing 8 landers!

For future reference: It would normally make more sense to target adult and mainstream in separate campaigns. The 2 traffic types are different in nature and will respond differently to your landers and offers. Also, traffic costs between adult and mainstream are different - so it would be better to bid separately for each.

Does the offer accept smartphones only? Or tablets as well?

The mistake with the postback url is common. Good to hear you got it fixed! You can still manually add the conversions to Voluum IF you've been passing the {clickid} token with your offer link. Please go to Voluum, go into the offer's settings, and see if the {clickid} token is in the offer link. If so, click on the little gear icon in Voluum (to go to settings), click on "conversion upload", and enter the clickids
and payouts - you can get the clickids from your aff network's stats (again, assuming you had appended the clickid token). If however you didn't include the clickid token in the offer link, then there's nothing you can do retrospectively - you'd just have to start your split-test over again.

As for spending twice your budget - yup it's probably a timezone discrepancy between popads and voluum so the budget got reset at midnight and started spending the next day's budget. Just make sure you have both set to the same:

PopAds - in campaign settings under "Time Targeting" section > "Timezone".

Voluum - little gear icon (settings) > Profile tab > "Default timezone".

Currently your focus should be on cutting landers down to a winner using the stats tool: https://stmforum.com/forum/showthrea...Banners-Part-1

So please check regularly and cut landers as statistical significance is reached.

Another thing to keep an eye on, would be your placement stats ([WEBISTEID] stats in voluum). If you're seeing any placement site in excessive loss, blacklist them. As for what would be considered to be excessive loss - since you're testing 8 landers, I'd say if a placement doesn't convert at 8x payout then that would probably be pretty hopeless (assuming all your landers are in good working order - function properly and are displayed correctly and load fast) - so $0.64 x 8 = ~ $5. Or you could cut at a lower amount, and retest them later on when you've identified the best lander. Would depend on how much traffic you're getting, how quickly you're wanting to get your testing done, and whether your priority is testing speed or conserving test budget.

Let us know when you've cut down to the last lander! Campaign's looking good!



Amy

---

Just saw that you've made additional updates - after I made that last post!

Very nice! This means you DID pass the clickID to your aff network! Please manually update the conversions as described in my previous post (don't forget to include the payout with the clickID as well).


I very very rarely update costs so don't know what the best method would be for handling this.

You can and should add the [BID] token in your popads traffic source settings in voluum, so that costs will be calculated automatically. However, there WILL be discrepancies between what you'll see in Voluum and the actual costs as recorded by PopAds. The difference can be up to two times as much!

But what you could do is figure out the percentage difference between the costs displayed in Voluum vs. the actual costs, and make a mental adjustment when browsing stats. For example, if you know that for your particular camp, the actual cost is around 20% higher, then when looking at costs in voluum, just mentally add the 20% to every cost you see.


AFAIK, popads does not support conversion tracking. There are traffic sources that do.


Yes - this is known as "clickloss". Up to 20-30% would be considered normal.


Use the stats tool recommended in the post I linked to above.


Most likely due to the backbutton script.


If you want to test placements for bots, see caurmen's thread:

https://stmforum.com/forum/showthrea...n-Any-Campaign

When targeting both carrier and wifi traffic, you'll get mostly wifi traffic anyway, so effectively you could just treat this camp as a wifi camp. What you could do later is start another camp to just target carrier traffic - you'll need to bid higher to get some volume though.


You probably got outbid - and yes it's normal for competition levels to fluctuate, as more/fewer competitors start/end their camps, change their bids, blacklist/whitelist placements - or otherwise change their campaign settings.

You definitely should test bids once you have a winning lander. For now, as long as the testing isn't going really slow due to there being too little traffic, I would suggest to avoid changing anything (including the bid). You're split-testing landers and it would be good for accuracy to test one variable at a time.




Amy

---

@vortex

I did, and it worked, yay !!!

For some reason I chose option "Do not track cost" , should have chosen AUTO instead shouldn't I. In the past I run PropellerAds traffic and there wasn't AUTO option. Maybe I thought PropellerAds and PopAds were the same.

Just figured out I got the tracking link wrong. Fixed the *.html file, re-uploaded to Amazon S3 however the old lander (with wrong tracking url) still showed up. Maybe because of the CDN. Is there a quick to have it refreshed? Thanks

Thank you, will do.

Thank you for the advice. I'll keep the bid the same for now indeed.

I'll keep that in mind.

Thanks Mr Payne, I will try your method.

---

Hi everyone,

Day 2 of Campaign 1 is a bit 'doomy'. There was 1 conversion early in the day and that's it. Another $20 spent with only 1 conversion. Compared to the same amount spent yesterday and 10 conversions. I did not change anything in the campaign.

I talked to my AM, he checked the offer and did not see anything strange.

Question 1: I don't know if this is related but I realized that in my affiliate network, the offer that I'm running turned to 3 identical offers (same id, same name) with different payouts. and the offer urls for each of them are slightly different.


The offer URL in Voluum is identical to the URL of the $0.72 payout offer. However, the payout for the GEO that I am sending traffic to is only $0.64.
Is there any chances that this is the reason why my traffic fails to convert?
What if I change the offer URL in Voluum to the URL with $0.64 payout?

I know it might sound confusing but I'm trying my best to describe what I'm thinking

Some screenshot:
1. PopAds report:


2. Category ID:

Question 2: Looking at the category I can see that ALL the conversion so far (11) come from Mainstream traffic. Should I stop Adult traffic as the purpose of this stage is too find a winning lander. And oops, the offer restriction say No Adult Traffic anyway.

3. Mobile Carrier:

Voxter was right, the majority of the traffic is WIFI, I should find a way to find websiteIDs with high percentage of bots and exclude them.

4. Landers:


Actions:
1. Add the bot-trap-lander (caurmen) to Campaign 1, direct all traffic to this bot-trap-lander (stop other landers for now), set a budget of $5, go to sleep, wake up tomorrow to check whether there is any website to blacklist.
2. Direct traffic to General only, no Adult. Leave the bid the same: $3.5.
2. Wait for STM advices on how to deal with dropping conversion.
3. My first thought was my bid was beaten by other affiliates, this left me with crappy placements that do not convert.
4. I think I'm gonna Increase my bid from $3.5 to $4.5? Maybe? Please correct me if I'm doing it wrong.

Thanks everyone, have a great Friday.

---

It must have refreshed by now. But for future reference, you can purge the contents of your CDN at any time.



Amy

---

Have no idea what the deal is here - would be best to check with your AM. Which link were you running yesterday that got you the 10 conversions?

If you're running the same link as yesterday, and your AM told you nothing changed in the backend, then surely it shouldn't be a problem?

The other 2 links would be worth split-testing against your current link as well, regardless of whether there's an issue with your current link or not.

Another thing to check: Check your stats from yesterday and note which placements were giving you those conversions, then look at your stats today to see if you're getting the same traffic volume from those placements today. If you're getting a lot less, then you've been outbid, in which case try increasing your bid to see if the conversions come back.

Or, you could have just had an especially bad day - sometimes if you keep running the camp, the conversions would come back all on their own.

Another thing to try, would be to "reboot" your camp, by pausing it for 1-2 days and resuming. A similar method would be to clone the camp. Then of course you could try both.

Absolutely!

Targeting JUST the best-converting traffic (e.g. a group of high-performing placements, a specific OS/carrier that's been converting the best for the offer) to test landers is a very smart way of conserving budget. You could even just use category 88 to do your testing if the slow speed is alright with you.

The only downside is that the result may be slightly skewed due to the fact that you're only using certain traffic segments (in this case one category), but I feel that the slight compromise in accuracy is worth it given the potential cost-savings.

Regarding adult traffic: Some affiliates will sneakily send adult traffic to offers that explicitly state that adult traffic isn't accepted. This will allow them to circumvent a lot of the competition (assuming the offer converts well on adult traffic - stuff like mobile gaming, antivirus, vid sub offers). But if you get caught by the advertiser, it's game over and they may refuse to pay you. In your particular case it's a no-brainer because adult traffic isn't converting well for you. In cases where it DOES, you'd have a dilemma on your hands, but I wouldn't try to swing you either way - it would be your call whether you'd want to take that risk or not.


There will always be more wifi traffic than carrier traffic.

It looks like the carrier traffic is converting better - it may be worth setting up a new camp to targeting just the carriers (or one camp per carrier even) and bidding higher to see how much traffic you can get. If the volume is OK, you may have some smaller green camps after cutting some placements.

Or, you could just hold onto that thought for now, and finish testing your landers first, or even wait until you've mass-tested offers using that winner lander.


Have you checked lander stats for statistical significance yet? It looks like some of them are ready to be cut! Here's the method:

https://stmforum.com/forum/showthrea...Banners-Part-1

Your plan sounds good - except I don't feel you need to increase the bid, unless you've compared placement stats like I suggested in the beginning of the post and are sure you've been outbid. You're not bidding low - should be sufficient for getting conversions. I'm not going to advise against it though, because doing so could get you more conversions to test your landers more quickly. It would just cost you more.




Amy

---

Hi everyone,

What happened on day 3:

1. I spent $5 on the bot-trap-lander.
2. Cut 5 landers using the lander cutting tool
3. Stop sending Adult traffic and direct my traffic to General ONLY. Max bid stay the same: $3.5 CPM
4. Spend $6.5 on the remaining 3 landers, got 6 conversions (today revenue: $3.84 - so it's back to normal, maybe yesterday was just an unlucky day).

1. Let's talk about the bot-trap-lander.
This is how I set it up on the same campaign - basically I add the bot-trap-lander to the other 8 and navigate 100% of traffic to it. As I understand this is effectively direct linking to the offer, which maybe a bit wasted as I can create a separate campaign, direct traffic to the bot-trap-lander and use the campaign1-url as the offer. (Oh, suddenly caurmen's tutorial makes sense when I'm typing this ) Please correct me if I'm doing it wrong

Cool, this is the result after $5 spent on the bot-trap-lander (visit descending)

Now if I sort CTR ascending with number of visit > 4:

Taking a look at this screenshot I'm not sure if there is any website ID worth cutting. There is no websiteID with CTR as low as 20% mentioned in caurmen's tutorial.
There are a few websiteID with a lot of visits with nearly 50% bot percentage (I'm assuming caurmen's method is correct). I don't know what to do with them. Probably need to see how they perform before making decision. Please give me some insight regarding these bot cutting things.

2.3.4. Lander performance today:

Lander 10 showing Green today but I'm not fazed by that.

Lander performance over 3 days: the lander cutting tool does not advise me to cut any of them so I wait for further data.


@vortex: thank you very much for valuable input, you've always been goddess to us.

Thank you for this idea, I would definitely try it sometimes in the future.

I think I would keep the campaign as it is and find the winning lander first.

I might be interesting looking the quality score:

Quality 5, 6, 7, 8 together account for 37% of the traffic but only 2 of 17 number of conversions (11.7%).
Maybe it's worth cutting those Quality Scores to conserve budget and/or allow me to bid higher? From $3.5 to $5.2 maybe? Please advise.

Looking at websiteID, I can see a few spent 3x-5x the payout without any conversion. I was tempted to exclude them but maybe I just leave them on for now


That's it for today.

I will have those good landers translated and/or modified and test other geos in the mean time. This 'find-the-winning-lander' process takes longer than I expected.
I need more actions.

---

All sounds good! Glad your conversions came back - it's common to have a bad day.


It's been quite a while since I've used caurmen's method to perform a bot test (because I've been running carrier traffic almost exclusively), but what you're saying makes sense!

Do feel free to ask that question again in caurmen's bot test thread to get an answer from the man himself.


Based on my limited bot-testing experience in the past, I found that cutting at 70-80%+ bots was optimal. Some of the placements that had as much as 40-60% bots were good-converters that were profitable. It may be dependent on the specific traffic source, geo, etc. as well. What you said about seeing how they perform first before making decisions sounds good. You can see which placements convert the best/worst and look at the corresponding bot percentages to see if you observe any patterns - then decide on a cut-off % based on that and apply it to your next bot test.

BTW - why visits > 4 only? I thought you had to get at least 60 impressions to be able to gauge bot percentages accurately?


Haha! Yet you have to admit, the green is always easy on the eyes.

One thing to potentially look out for: If the 3 remaining landers keep going head-to-head (i.e. continue to perform similarly), you may need to just pick a random winner instead of wasting money running to statistical significance. Caurmen talks about this in detail in his guide:

https://stmforum.com/forum/showthrea...ugh-Clicks-Yet

"3. If you've had a few hundred clicks worth of data for each landing page, all of them have probabilities of being best between 25% and 75%, and you're seeing high, overlapping graphs like the ones below, the remaining landers all perform about the same. Pick one to continue running, pause the others, and start a new split-test (see "What To Test Next"). "


Good idea!

Actually it's always a good idea to use the best-converting traffic for testing, because then you can conserve test budget. When you find your best offer+lander, you can always open up your targeting to retest some of the stuff you've cut out earlier - as some of that may be profitable then.

This is of course assuming the best-converting traffic has enough volume to allow you to get your testing done in a reasonable amount of time. For example it won't do to whitelist a couple of small placements and do your testing on those (would take a million years!)


Similar theme as last paragraph: If you want to conserve test budget, cut them for now. Could always retest them once you've identified your best offer+lander.


Sounds good! Have a good weekend!




Amy

---

Hi everyone,

On day 4, I narrow down my Quality Score from 5-10 to 9-10, then increase the bid from $3.5 to $5.5 (just wanted to see how it goes).
Results: Although I got 19 conversions from the remaining 3 landers. ROI drops from -20% to -60%. Traffic becomes more expensive then before.

To summarize what happens:

(On day 5, I stopped the traffic at only $3 spent.)

In the end I have 3 landers with very similar performance:

It's haunting me in choosing the winner. As vortex suggest, I might go forward with a random one.
Question: Lander 13 showing competitive performance with the other 2 although the click percentage is much less. I'm thinking maybe if I use the back-button script on Lander 13, there would be more conversions? Please correct me if my thinking is wrong.

Bigger question: Can you please suggest what to do with this GEO/offer?
What I'm thinking:
- On the same traffic source (PopAds): keep sending traffic using 1 lander and start cutting websiteID until my ROI > 0. Please advice on the bid price. Am I bidding too high at $5.5 ?
- On other traffic sources (Zeropark, PopCash maybe?): 3 winning landers, cut to the winner and optimize at the same time?

For other GEOs with same language, I'm gonna do the same thing: 3 lander per traffic source.
I will keep you update later.

---

I see you've run the landers through a Bayesian split-test checker - I did the same thing, and their performance is definitely far from statistically significant at this point.

Just pick one and start a new test. For all intents and purposes at this point they're effectively identical.

---

Day 6 update:

Actions:
- Campaign 1: cut 5 websites (>$3 spent without any conversions), leave the bid as it is ($3.5 - Mid Bid)
- Launch campaign 1-1: Low Bid $2.5, exclude 5 websites cut from Campaign 1
- Launch campaign 1-2: High Bid $4.5, same as Campaign 1 and Campaign 1-1, exclude the 5 websites.
What the 3 campaigns have in common: Freq 1/24, quality 9+, PopUp, Mainstream only, 3 landers (10, 13, 16)

- Launch Campaign 2-1 on Zeropark: WIFI, 3 winning landers (10, 13, 16), bid 0.0009, Mainstream
- Launch Campaign 2-2 on Zeropark: Carriers, 3 winning landers (10, 13, 16), bid 0.002, Mainstream

A quick look on how the 5 campaigns perform through the day (only 6 hours for the Zeropark campaign since they were approved late):


My verdict:
1. PopAds Campaigns:
- The Bid of the LowBid campaign is too low, as a result, there's virtually no traffic. Nothing to say here.
- The HighBid campaign has more conversions but lower ROI (-60%).
- The MidBid campaign performs the best in term of ROI.
- Lander 10 turns out to be the winner today (with PopAds), it even show green in the MidBid campaign --> I will definitely use it for later test.

HUGE Question: What do I test here? Please advise.

2. Zeropark Campaigns:
- Lander 16 seems to have best performance with this traffic source. It even show +56% ROI in the WiFi campaign without any optimization.
- Probably I will stop Campaign 2-2 (Carriers Only) and focus on Campaign 2-1 (WIFI), probably let it run for another day before deciding on Lander 16.

How did I perform bot detection on Zeropark WIFI?
- I direct 25% of the WIFI traffic to a lander called "Global - Zeropark Traffic Splicing".
- In the lander, I used a smart-link created in Afflow.
- I then look at the campaign in Afflow to spot targets with high percentage of bots.
- Cut those targets.
- My mistake: I had the token wrong: http://lalalaxxx.com/?utm_medium=6...6&utm_campaign={campaignid}&1={target}&2={source}&3={traffic_type}&4={visitor_type}&5={carrier}&cid={clickid} ({carrier} token seems to be okay)
- As a result, the campaign name in Afflow show unnamed_campaign, the keyword 1, 2, 3 and 4 showing nothing:

- The token should be like this: http://lalalaxxx.com/?utm_medium=6...6&utm_campaign={campaign.id}&1={var:target}&2={var:source}&3={var:trafficType}&4={var:visitorType}&5={carrier}&cid={clickid}
- After fixing the link, I can see another campaign with proper ID and its keyword 1 showing targets with bot%:

- I believe I can do the same with PopAds or other traffic sources (is it true? or is it only applicable with Zeropark?)
- Please fix me if you think any part of what I did was wrong.

End of day 6 actions:
- Campaign 1, 1-1 and 1-2: cut lander 13 and 16.
- Campaign 1-1: change the bid to 4.0.
- In summary: Bid Price for Campaign 1, 1-1 and 1-2 are $3.5, $4.0 and $4.5 respectively.
- Day parting these 3 campaigns: stop traffic from 0:00AM to 7:00AM.

- Campaign 2-1 (Zeropark WIFI) leave it as is. (Should I increase the bid)
- Campaign 2-2 (Zeropark Carriers) paused.

That's it for today.

My follow along is taking me a lot of time, probably I am putting too much details in it. Nevertheless, I'll make it shorter, more comprehensive once I become a master.

---

Good stuff!

You certainly could take one or more of the best-performing landers, produce variations of these, and split-test them against your original best lander.

It would make sense to add the back-button script to all landers by default anyways - otherwise, when the visitor clicks back, you'd lose them. It's one of those things that can't do damage to your lander.


What you ended up doing - testing different bids on popads - was a wise decision. And cutting placements was another.

Testing bids and cutting placements will also be the main optimization tasks on every traffic source you scale your winning offer+lander to.

You can also drill down to other stats such as OS, devices, browsers, hours, etc. If you notice anything doing quite negative ROI, cut it from your targeting to increase the overall campaign ROI.


You can continue to optimize the HighBid and MidBid camps.

You've cut down to a winning lander, so now drill down to placement stats for that lander, and start cutting anything that is in loss by 2x payout or more.


For sure you could do the same for other traffic sources.

The only thing you'll need to change would be the {var:[token-name]} variables - just look them up in the Voluum traffic source settings for the particular traffic source.


You can set up staggered bid camps to test bids just like what you're doing on popads, leave the promising bid(s) camps running and cut targets and sources.

Alternatively, just run a single camp and tweak the bids on the target level, by increasing the bid for well-performing targets and decreasing the bid for targets that are in loss but not by much (such that decreasing the bid and therefore the cost has a chance of turning the target green).


I don't think you're being overly detailed - the details are appreciated!



Amy

---

Hi everyone,

Just a quick update of the campaigns.

For the last 3 days I let the AU campaigns gathered more data until the budget was depleted.

I notice that for the 3 PopAds campaigns with same targeting and different bids, the campaign with highest bid (Campaign 1-2, bid $4.5) attract all the traffic of the other two. ONCE its budget is reached, the campaign with second highest bid (Campaign 1-1 bid $4.0) start to have a chance to receive traffic and convert. It's all make sense.

I realize the campaign with mid bid (Campaign 1-1 bid $4.0) has the highest ROI after 3 days. There were moments today the campaign turned GREEN for the first time ever which made me so pleased.

The campaign with lowest bid (Campaign 1-0, bid $3.5) virtually had no traffic for the last 3 days.

Anyway, I took Amy's advice, took Campaign 1-1 (highest ROI) further, cut placement with really low ROI (<-90%) or no conversion at all.
The things I cut:
- Brand
- Brand - Model
- Browser
- Browser versions.
- OS
- OS versions
- WebsiteID

It took me a lot of time actually because I have to combine data from 3 campaign to make a decision. But it's worth it. I am a lot more confident now in term of optimization.
I then have to increase the bid to $5 for more traffic (estimate 2000 impression per day).

Let's see if it will turn GREEN.

Thanks for reading.

James.

---

Some green is always encouraging to see! And learning is the main progress here.

Just want to point out a couple of things that may help in the future:

1)Next time, confirm there's enough traffic volume before starting a campaign. 2k impressions/day is way too low to be making more than a few bucks a day from.

2)I think it's terrific that you're learning so much about optimization! However, try to refrain from optimizing past the point of diminishing returns - where you're spending more time than it's worth just to squeeze a little bit of extra ROI from the campaign. Focus on the big segments (e.g. for OS: android+ios+maybe windows phones; only carriers that have major market share; etc.) with lots of traffic volume, and ignore the small segments (such as OS versions and browser versions). Exception is when you're targeting wifi on a large geo - in that case even the "small" segments can mean major money.

Instead of over-optimizing, use your time to do more spying, more research, set up more campaigns, test more offers and landers, scope out more traffic sources to scale your successful campaigns to...etc.

Looking forward to seeing how this camp will do!



Amy

---

Home > Mobile > Follow-along Campaigns