Home > Push Traffic >

---

FA SweepsStakes #3 (31)

---

Alright, I'm starting a Grocery Voucher sweeps campaign in Singapore. I'm using a Spin Wheel lander, 1 Click Lander, and another "Special Day" one click lander.













It's mobile only with a CPC of 0.048 (Recommended bid was 0.039) and the offer payout is $2.2 for each offer. I'm running Classic Push, no broker traffic, and high user activity level.

Daily budget of $10, no targeting, and day parting is set to all hours.

5 Creatives (All Message Icons)



Will keep you updated.

Traffic Source: Propeller ads
Landers: 3
Geo: Singapore
Offers: 2
Vertical: Sweeps

---

Alright, no promising data so far.





As I stated earlier, the payout is $2.20 and I've already spent $14 so I would need at least 7 conversions just to break even.

But, given that there are no conversions so far, I'm sure this is unlikely to happen.

I'll let it run until it gets to $20 and see if anything changes.

By the way, should we do a follow-along for every campaign we run, or just campaigns that show some traction?

---

Yup, looks like a dud, zero conversions is always a bad sign. But again, check the data in a more detailed way... any large budget drainers, such as one (or a few) zone sending bulk of the traffic, large % of traffic where the language doesn't match the target GEO etc... it's quite common that a few bad zones/placements ruing the whole campaign, but I guess you already know that and you have checked it.

Depends on what kind of advice you want to get and how often you post. You can keep it all in one FA, which will allow us to check what you have already tried before, without having to browse different threads... but this can get confusing if you post a lot and the thread grows. On the other hand, starting a new thread for every campaign will lead to repetitive advice, because those who reply might not be aware of your other FA threads.

I would say it makes sense to start a new FA in case you make some significant changes, try a new source or vertical or simply want to start from scratch with a different approach. But as soon as you're just launching new camps with a similar concept in the same vertical, keep it in one FA.

---

Great start of a follow-along! I'll mainly be following-along - and let members that have more push experience give advice.

Not necessary to start a new follow-along for every campaign - just one per traffic type would be awesome!

Looking forward to seeing how you progress!


Amy

---

@anthonyh

My opinion, offer is a dud. First a few quick questions, then some math to round things off. Experience is teaching me to look at the math prior to a campaign even running.

First the questions:

1. Did you confirm clicks are making it to the CPA network? Reason I ask is I didn't see it in your screenshots. I see Propeller and CPVlabs right?
2. Do you follow your campaigns through every time before you launch them and make sure they work?

Alright, brace yourself for some math homework.

Offer pays $2.20, Geolocation bid is 0.048 or 0.05 just to round it off. Your LP CTR is 8% which to ME is low. Running most giftbox survey and spinwheel landers I was running SOI this month garnered 20% CTR on spin wheels AT LEAST. Survey and giftbox were lower but never 8%. Now Singapore could be different, never ran there. Let's assume offer converts at 15% when its a good offer for SOIs. I never had things convert above this mark in SOI sweeps others have better experience to share here I am sure.

Alright so...

24 clicks to offer...
15% CV rate = 3.6 CVs.
Let's just say 4 to be safe. That's $8.80 return based on your data if things went decent.

If the LP had a CTR of say 15% then you would have sent...42 clicks to the offer. With 15% CV that's 6.3 CVs which is $13.20 (or in your case break even). This is also prior to mentioning runaway zones like @matuloo mentioned, and splitting the traffic 3 ways on different LPs which diluted things even further.

Long story short I have fought this battle many many times already. It's an uphill one and it's a large bolder you are pushing. I am trying to show here, run math based on some averages and you can see before you even run a campaign the generalized odds it has to succeed. CAMPAIGNS CAN EXCEED EXPECTATIONS FOR SURE, but I found that tempering my expectations based on decent averages from previous tests let me know what I was up against.

Hope this helps you get some clarity. 0 results is painful for sure, but always stay hungry to succeed and you will trudge though it.

---

So this is something else that I never really do but will definitely start.

Here's the zone data for this particular campaign.



So say if a zone or two were eating up the budget, should you kill it immediately?

---

1. Ok, so the clicks definitely made it to the CPA network. Yep, Propeller and good old CPV Lab.

2. Yep, I take the Campaign URL and open up a private browser and make sure that the lander and Offer URL is setup correctly. (That's all I check)

Ok, let me make sure that I understand.

So, you're saying you create a mathematical scenario based off of averages from previous campaigns, and that'll give you a rough idea of whether the campaign has a chance to succeed or not?

---

Haven't run SG for a long while... but for me it looks like your creative CTR is very low (if you are actually bidding on hi only).

Your screenshot does not show over dominant zones. You have 3 zones in the 40tees and one in the 30tees. The other zones only have so few clicks because of low CTR.

In other words... it looks like your creatives are a bigger pain point than the zones you got served.

Do some spying or try some crazy stuff to get an idea how to get better ctrs. Once you reach a certain level you then can optimize to find creatives that convert.

Of course this only works when you have an offer that does actually convert frequently and stably.

Maybe go with some cheaper offers first so you can try more different things for the same budget.

---

@anthonyh

More or less. I am not asserting that's its going to be a 100% accurate method. I am saying though...That if traffic is 0.10 a click (like Poland was) and my offer pays $0.50 it must convert is 4 clicks to make any profit, and 5 to break even. Simple as that. My campaigns average CV % so far of around 8% when they are working (today is a weird day for some reason).

So taking that into consideration, in my history that means 1/11 will convert. That means a CV in PL would cost me 0.10 * 11 = $1.10 for a CV payout of $0.50. From my history I can pretty much tell that my odds of success here are lower chances are likely. Could I be proven wrong by higher CV amounts? Cheaper clicks? Better CV rates? Do different GEOs perform differently? YES YES YES AND YES.

This is only my experience, which is limited to my tests and this is what I found. With simple math like this I work to target traffic where CVs have to be way less often but I can afford it even at lower CV because my costs are low. Campaign has much higher odds to succeed. Like I said though, this is based on my because my payouts, tests, and overall experiences so far.

@larsometer is right. Creatives CTRs are very low. From my experience 1%+ is ok. 1.5% is pretty good especially in higher tier GEOs. This comes from a proper blend of clickbatey type and qualifying the user as well. Very tough to get right in my opinion. I struggled with this for about a month straight. My CTRs looked exactly like yours.

Now they are pretty consistently 1.5% in high tier GEO and 5%-10% in low tier. Took me a thousand dollars to get it right. Keep testing vastly different creatives and you will find ones you use over and over again.

---

Yes, in case there are some zones that really stand out and eat up a large % budget, without bringing conversions, those are likely bad from whatever reason, they spend your budget quickly and there are no $$$ left to actually let the other zones spend before the budget runs out.

Think about it like this: There are 100s if not 1000s of zones you can get clicks from, depending on the traffic source and type. And you simply don't want to buy just from let's say 10 large ones, because quite often, the smaller zones can deliver higher quality traffic. But if the large ones spend all the budget, the small ones won't even have a chance to spend.

But, there also also big zones that are great, so do not block zones just because they spend a lot! They must underperform, when compared to the others.

For example, when I run a campaign and I'm getting conversions, but there is some zone that spent I don't know, 20% or 30% of the budget and didn't convert... such a zone has to go.

Looking at the screen you shared, it doesn't look that bad, there are several larger zones, then it goes down slowly, looks pretty natural to me at the first look.

I would say that in your case, the problem might be with the creatives as the other guys have suggested.

---

Alright, I have some good news...I've finally got my creatives to a 1% CTR, so I'm improving in that area.




These are the stats from the campaign. It got quite a bit of conversions, but, at a payout of 0.25, I would need over 100 conversions just to break even.



I did notice that one of the landers got 6 conversions out of the 11 so I'm not sure if that's an adequate amount of data to say that it's a decent lander.



The zones look like the spend was spread out pretty well also.



All in all, this was just a small update.

I'm excited to be making some progress

---

What are your stop criterias?

Seems like you have a loss of ~60 USD so far on a 0.25 USD camp. ROI is very deep in the red.

Your cpc is around 0.04 USD. This would mean that every 6 clicks you would need a conversion to run at break even (and never recover your losses).

However your current LP CTR is ~50%. So you would actually need 12 clicks to get a conversion and 50% discount on your cpc to run at break even.

Of course in the beginning you kind need to lose money in order to learn something. But... it is not necessary to hit yourself with a sledge hammer in order to feel some pain.

Sooooo....

Why not try a way cheaper geo where you get loads of clicks for 0.005 USD cpc on hi quality on propeller?

You can try 10 times more offers with the same bugdet. So instead of sledge hammer you can torture yourself with a small needle instead. Will heal much faster

---

Congrats on the 1% CTR!

Just a small word of advice though - I've been burnt before when I got so fixated on improving the CTR that I neglected the CR. At the end of the day, it's profits we're after, not CTR.

A higher CTR doesn't always translate into better CR overall. For example if I promise the moon, that would get more people to click, but when they get to the offer and see that it's different from what I promised, they would bail.

The ROIs are quite low, even for the best-performing lander. Think: How much more room do I have for optimizing this campaign? And is the room for optimization enough to turn around the -85% ROI?

Example: If you've already tested a good number of ads and landers, and have drilled down into the data for various traffic segments and can't really find any big segments with significantly lower performance that you can cut to dramatically increase the ROI, then it may be time to move onto other offers/geos/verticals/traffic sources.



Amy




Sent from my iPhone using STM Forums

---

I actually made a mistake and forgot to kill the campaign the day before. I think I fell asleep or something.





...That's hilarious! I'll definitely take your advice and try out a different GEO.

---

Got it.

I'll keep launching and try to make some real progress.

---

You ARE making real progress! Keep testing and "failing fast" and you'll see what works and what doesn't. Learning how to increase ad CTR is no small feat either - please don't let that little "be careful" message from me discourage you! Doing testing to get to 1% CTR from where you were before is definitely progress.



Amy


Sent from my iPhone using STM Forums

---

1. Does it mean the result of a landing page with -85% ROI (after spending 10x payout) is worth optimizing further and start cutting placements after they spent 1-2 payout without conversions?
2. Let's say after spending 10X payout on a campaign, only one lander and only one offer shows -85%. Should I leave only that lander with the offer and start cutting placements?

---

Again, you need to look at your ROOM FOR OPTIMIZATION and see whether there's enough room to achieve positive ROI from your current ROI.

In general, with an -85% ROI, it would almost take a miracle to optimize to green. (Here we're only talking about pop, and possibly push.)

So the answer to both your questions is "probably not".

Pop and push only has so many "moving parts" you can optimize. Ad, landing page, and targeting (placements, OS, carriers, etc.)

(We're assuming the offer doesn't change.)

Landing page for push and pop - usually it's just ripping the most popular few formats, fixing them up, running to find a winner. Unless you get very innovative, not a whole lot of room for optimization beyond the first or second round of split-testing.

Ad - if you're running push there's more room for optimization there than for landers.

Targeting - The only way you can achieve big changes in ROI is if you can find big traffic segments that convert very significantly worse than the rest, that you can cut. Or, if you can find one or more big segments that convert well that you can whitelist.

So basically you'll just have to look at your data and do some math, to figure out whether you have enough room to optimize the campaign from current ROI to green.

If not, test new offers / geos / traffic networks.

I'm really glad you asked these questions, as it gives me the chance to once again give this advice: The most common mistake new affiliates make (and yes we're ALL been there) is try to optimize a campaign with very negative ROI by tweaking the small dials - like cutting a ton of small to medium placements. That will never work. It would be like trying to fill a bathtub using a bottle cap.

Hope that clears things up further! Please feel free to continue the discussion.



Amy






Sent from my iPhone using STM Forums

---

Thank you!
I have already tested 3 geos and around 15 different offers. Each time after the initial test (10x payout) I get -100% to -80% ROI. I always split test 2-3 landers and 2-3 offers and many (10-15) creatives (I run push traffic). So the question is: how many different campaigns do you usually test before you find that promising one which has -60 or -50 or better ROI after initial test? I am asking this because I dont really understand If I have tested less than I should to find a winner and if I should test even more or maybe keep trying to optimize those which give me -85% ROI. And, ofcourse, I understand that each affiliate's experience could be different and some of us can find a winner after 2-5 tests but some take more tests. So just asking your opinion based on general statistics from many affiliates. Knowing that sometimes we need to test 30-50 campaigns before we find a winner would make it clear to keep testing more and that testing 10 campaigns is not usually enough.

---

Thank you!
I have already tested 3 geos and around 15 different offers. Each time after the initial test (10x payout) I get -100% to -80% ROI. I always split test 2-3 landers and 2-3 offers and many (10-15) creatives (I run push traffic). So the question is: how many different campaigns do you usually test before you find that promising one which has -60 or -50 or better ROI after initial test? I am asking this because I dont really understand If I have tested less than I should to find a winner and if I should test even more or maybe keep trying to optimize those which give me -85% ROI. And, ofcourse, I understand that each affiliate's experience could be different and some of us can find a winner after 2-5 tests but some take more tests. So just asking your opinion based on general statistics from many affiliates. Knowing that sometimes we need to test 30-50 campaigns before we find a winner would make it clear to keep testing more and that testing 10 campaigns is not usually enough.

---

This means that you already found offers that convert. That's good.

Next step is to find offers that convert more often at a more reasonable loss.

So instead of aiming for numbers such as percentage of profitable offers, better aim for making progress. That's the only thing you can actually control.

It wouldn't help you if somebody who is doing sweeps for 5 years tells you that 1 out of 14 offers brings him profits.

You could only take this as a benchmark if you had the same / similar amount of experience.

So better compare yourself to the person you have been yesterday. And then ask yourself what you need to do to get a little bit closer to the ideal picture of tomorrow's you.

If you want numbers it might be better to look at budget required. Within spending 3-5k USD wisely you should fine a profitable offer for sure.

Wisely just means not to gamble and to avoid making same mistake more than 3 times (so no need to beat yourself up).

---

Thank you! Sure I understand that affiliates are getting better in choosing offers based on experience they get. And Im focusing on the progress. Just wanted to know how many offers (+-) professional/experienced affiliates test to find a winner? this could also be my aim for the next 1-2 years. So maybe you could share how many offers did you test when you started before you found a winner and how many offers do you need to test now (as you have experience) to find a winner?

---

I gave you a number (range)

Just keep on testing and trust in your journey.

---

I'm with @larsometer - it would be difficult to say something like "on average you'll need to test 30 offers to find one profitable offer" because everyone is different.

It depends on what traffic sources you're running on, which geos, which verticals, which affiliate networks. Competition levels are changing all the time. More/fewer good offers come and go on/from the market all the time. And of course your network/connections/relationships with networks can make a big difference, as can your experience level.

For now though, just know that it's not unusual to test 15 offers without finding a winner.

The name of the game with pop and push is to develop an efficient testing, optimization and scaling strategy - one that will allow you to do massive testing of offers as quickly and for as cheaply as possible to find promising campaigns, then optimized those quickly and scale.



Amy

---

I started a new camp a few minutes ago and it already has 15 conversions

NOTE: The data in the image doesn't reflect the proper amount for some reason, but my tracker and Aff network both confirm that I have 15 conversions.



Here you can see the actual data:



The Zones are spread out pretty good and nothing significant is standing out from one zone.



Here's the landers and creatives:






You can see that every single creative converted except one, I'm thinking I should kill it.

What do you guys think?

By the way, every time the data comes rolling in, I always feel like I don't know what I should be looking at first.

I need some sort of procedure that I can go through for every campaign.

I know to cut things that aren't performing well, but the issue is knowing when to cut them.

What I currently do is just see if one ad, lander, or placement is losing more than the others and I tend to want to stop it immediately.

Like the zones for example, though they are spread out, I tend to want to cut the ones that haven't converted, but I don't want to cut them prematurely.


Update:

It's still converting but seems to be slipping away. I would need about 50 conversions to come back from this and I don't know if it's possible or not.

It showed great signs early on so that's making me think that it has potential.



A few of the zones have more conversions than the others so I'm not sure if I should kill the rest and let them run and see if it comes back to life.

---

There's a possibility for this campaign to end up being profitable!

I'll write more when I get to my desktop later, but for now: Have you checked your lander stats for statistical significance? The "part 1" of my "how to cut banners and landers" post covers that in detail (please see my signature). You should start checking to see whether you could start cutting landers down to a winner.

Also: Check for placements that are in loss by 3x payout, and cut them for now. Can always resume later if applicable.


Amy


Sent from my iPhone using STM Forums

---

Another thing to check: In the tracker, when you drill down to the current-best landing (the one with the most conversions right now), are there a lot of profitable traffic segments, i.e. placements, OS, carriers vs. wifi?

Reasoning here: If there isn't enough profit even for the best lander, then you need to change something (test more landers/offers/etc.)

Optimizing a campaign can be like digging through a mountain of shit to find the pony that's buried in there somewhere. You got to make sure there's actually a pony in there first (i.e. enough profitable segments) before spending money hauling the shit away (i.e. cutting traffic segments such as poor-performing placements). If there isn't a pony in there in the first place, no amount of digging and hauling shit away will give you a pony.

However, if you're looking at the stats and thinking "hmm...I'm not sure if I should stop the campaign", then you may not have enough data just yet. So in that case maybe wait a little bit to collect more data before taking another look. Also, the more experience you have, the sooner you'll be able to tell. In the beginning almost every affiliate will overrun campaigns. It's OK as long as the campaign isn't completely hopeless and you're collecting stats and learning to analyze them.

Another thing you can do: Even if it doesn't look like you could make the current campaign profitable, if the ROI isn't hopeless (let's say -30% or better on a decent winner lander), you could try testing different bids. There's a lesson on testing bids in the 40-day tutorial.

Regarding not seeing conversions in your tracker: It can sometimes take the aff network a few minutes to post them to the tracker. Are they showing up now?



Amy


Sent from my iPhone using STM Forums

---

I'm still missing about 13 conversions from my Propeller stats.

The drill down column for the landers doesn't show me that data in CPV Lab...unless I'm looking in the wrong place.



I do see those options under the Placements, and here is all of the profitable ones.

---

It looks like there's just a delay in your aff network posting back to the tracker. The longer you run this campaign, the less of an impact this delay will have on your data and decisions.

Can you not drill down 2 levels deep? For example down to lander and then OS, so you can see OS stats for each specific lander?

And have you checked for statistical significance for the landers yet? You may already have a winner.



Amy


Sent from my iPhone using STM Forums

---

I figured it out. Here's the Placements, OS, and ISP/Carrier data with the winning lander only:




Yes, I used the A/B Split-test calculator for the creatives and the landers.

I cut one of the landers and now have the two best running.



I did this with the creatives as well:

---

Few random thoughts/ideas that you can work on. I haven't read your thread in detail, so I might be repeating something that has already been said:

- PropellerAds is hard. Very hard. I know quite a few people who make it work consistently, but I feel that if you're starting out you can branch out and try other traffic sources

- Different traffic sources have different push databases. Your offer could work really well on one traffic source and absolutely bomb on others. Sometimes its not your ads, the traffic just isn't a good match.

- You're competing with other guys. Just open up a spy tool and take a look at how others are promoting the same offer. Usually more aggressive approaches sell better. Find a way/loophole to remain compliant while being as aggressive as possible. (or just use another traffic source!)

- About creative testing: you need to test ideas that are vastly different from what you have currently. From your last post I see that you are making variations of a creative (same image, slightly different headline/description). While it is good to do this when you have a winning creative, you should look at testing completely different images, texts ( and try to even change the angle completely) and see which one works the best

- To add to the point above, you can watch some videos by Colin Dijs on youtube, its completely free (his native tutorials apply to Push very well). Theres many hours of content in it, but if you go through it, you'll find that he explains how traffic source algorithms work (according to him). This should give you some ideas on how to adjust your approach

---

Home > Push Traffic >