Home > The Newbie Zone > Questions and Answers

---

What Does An Campaign w/ Potential Look Like? (17)

---

I'm running POP/PUSH campaigns using PropellerAds and offers from Zeydoo, Mobidea, and ClickDealer.

When you're running a campaign for the first time (new offer, new geo, new lander, etc), what sort of things are you looking for to see whether it has any signs of life?

A progression of ROI within a few days? CTR on landing pages? Conversion rates?

And the opposite question: what indicators do you use to determine a campaign just isn't viable and should be stopped immediately?

---

Hi there!

The key is to have enough data. When you have had multiple conversions for all the LPs and offers you are running, you need to look for the best and worst segments the campaign has and cut the worst ones step by step. The number of leads depends on the type of offer you are running (SOI/DOI) and on the payout. For example, if this is SOI offer with 1.5$, you need to spend x10 (15$) of the payout to see if it will work.

Of course, the higher is CTR, the cheaper will be clicks, but it is not the main indicator for good campaign. You should look at your ROI. If it is around 0% or more, that's great, because it can show the positive dynamic in the future. If it is around -30% - it can have a chance, but you should make Black and White lists to turn your ROI in a profitable side. Whether it is below -50%, you need to turn it off or change something(creative, offer page, White/Black lists etc).

There can be a situation, when you could reach the 200% ROI, but don't make a conclusion in one day, because it can be incorrect statistics.

---

I'll approach this from the pop perspective - but the same reasoning can be extended to push as well.

It all comes down to how much room there is for optimization and whether that would be sufficient to get your campaign from its current profits+ROI to your desired profits+ROI.

To give two examples at opposite ends of the spectrum...


Example 1: Little room for optimization

The offer is direct-linked, it of the kind where it would be difficult to find a suitable lander.

Offer accepts traffic from one geo, one mobile carrier, Android OS only.

In this case, placements are pretty much the only main thing you can optimize. In that case, I would cut the offer if it doesn't make at least 5 conversions after spending 10x payout. Even at 5 conversions, it can take quite a bit of cutting to get from -50% ROI to 30% ROI (or whatever your target ROI is).

If there isn't a lot of traffic or placements in that geo to begin with, I would set the cut off higher than the -50% ROI above. In smaller geos, I wouldn't even optimize offers that aren't breaking even (or better) to start with.

(Of course, if you're planning on testing bids, then you may just "suck it up" and keep running anyways, just to cut the worst placements, then up the bid etc. etc.

Also: In some cases even when the overall campaign ROI is low, you can still pick up good placements to whitelist, but you'll need to find more than a few of those to get more than a trickle of traffic.

And even in these cases, testing more offers first to find a good one BEFORE testing bids or considering whitelisting, would result in less money lost and more profits made.)


Example 2: Ample room for optimization

The offer converts the best with landers are used, accepts traffic from one geo, both desktop and mobile, wifi and carrier, all browsers, all OSs.

The geo has a ton of traffic and placements.

In this case, you can potentially start with a lower ROI, cut all the way to green, and still have enough traffic left to make a good profit.

But even in this case, again - testing more offers first to find a good one before optimizing, would result in less money lost and more profits made.


The point is: The more room there is for optimization, the lower you can set the bar when it comes to which campaigns to optimize beyond the initial testing phase.

And, as you collect more data, that room for optimization will shrink. e.g. When you've cut landers down to a winner, cut more and more placements etc. But hopefully, your campaign performance will also increase to keep up. (If it doesn't, then you may want to stop that campaign.)

So @bluemarble to answer your question: I would say to only run a campaign for as long as it still looks QUITE HOPEFUL that your remaining room for optimization can still get you to green.


Something else to keep in mind - let me copy this from another post I made earlier today:

Hope that helps!




Amy

---

Does the 10x payout rule apply to the entire campaign or each zone ID?

---

I was under the impression that testing meant running a campaign and doing some rudimentary optimization. Thank you for clearing this up.

For your testing run, what type of ROI would indicate it's not worth pursuing? -50% or lower?

---

Different people define things differently. I personally consider "testing" as split-testing funnel elements, i.e. offer, landing page, ads (for push), and "optimization" as tweaking the targeting (blacklisting/whitelisting) and testing bids.

As for a cut-off ROI: Again, it depends on how much room for optimization the campaign has. In the case where the ONLY thing you can optimize are placements/zones, I absolutely wouldn't start lower than -50% ROI - and even THAT is ONLY if I'm getting good traffic volume from lots of placements. Even if we're only talking about placement optimization we can still evaluate how much room for optimization we have. For example if you don't have a lot of traffic volume in the first place, you'd only be able to afford minimum cutting, so your starting ROI has better be at least close to breaking even.

Another extreme example: If your geo only has a couple of big placements, with the rest being tiny placements you'll never collect enough data to decide whether to cut or keep, then you have practically no room for optimization, so the offer will need to be profitable from the very start.

Once you start mass-testing offers, you'll see what good offers (that are either profitable right off the bat, or can be with slight optimization) look like, vs. kind-sorta-hopeful offers (that would require quite some optimization), vs. duds. Most beginners (not meaning yourself specifically - just making a general statement of observation) would much rather spend all their money optimizing, rather than on testing. This is one of the biggest limiting factors to success on pop (and probably push). Experienced affiliates focus efforts on testing, and would only cherry-pick the best offers to optimize and scale. If you read through a few follow-alongs where members were successful in the end, you'll notice that in the vast majority of cases, the turning point was when they were able to make the shift from optimization to testing massively.


Hope that clarifies things a bit more!



Amy

---

Thank you for the guidelines. When testing, it helps to have limits so you can be systematic and take emotions out of it.

I know there are edge cases and not every rule is hard and fast. But it really helps to have something in place.

When you're mass testing offers, how do you approach it?

1. Pick a vertical (e.g. iPhone sweep).
2. Pick 5 offers?
3. Download 5 LPs (as widely different as possible)?
4. Pick a geo and test for 10X the payout for the entire geo?
5. See if any combo approaches -50% or higher?

(I know there are exceptions to everything but I'm just looking for a basic framework to approach testing.)

I really appreciate your input. It's been super helpful.

---

It is better to test every geo separately, because, for example, in that case if you have three geos for one offer, it will be too small budget for tests (about $3-4 per geo). Besides the budget reason you must work with different geo separately, because you need to differ your promoting approach on every geo for good CR.

If you try to promote the Smartlink, it is almost impossible to test all geos with x10 budgets, because you can lose a lot of money. You should decide how much can you spend on this, and what you want to figure out or achieve.

---

When I'm mass testing offers, I try to use as few LPs as possible, as that adds another element to the flow. In an ideal case, I would use just 1 PROVEN lander. The goal of this mass test is to find offers that show promise, and to identify those that are either dead or very poor. Once I have them, I add more landers and ads to see which one of them can be optimized into better numbers.

I know it's harder to do, when starting out, since you don't have proven landers just yet... but still try to limit the amount of LPs as much as possible.

This test won't be perfect of course, there is still a chance that some offers that don't perform with the lander you would use, could still work better with a different one... But you simple cannot test everything at once. You can always return to some of the failed offers later on, once you have at least somewhat proven funnel.

---

Thank you - glad I can be of service!

Your proposed method sounds sensible for the most part - I would suggest the following tweaks:

-5 offers: Being able to test multiple offers is always nice, but it depends on how many affiliate networks you're running with and how many good offers are available for the particular geo and offer type (e.g. sweeps prize). So even 1 offer will work, but of course several would be amazing!

-5 LPs: Yes definitely recommended, but can be fewer if you're REALLY confident that they include the very best landers being used out there. The idea is to include enough landers so that the winner of the lander split-test will be a decent converter, so that you can focus on just testing offers after that (when new offers become available).

If you don't include enough good LPs in the first round of lander testing, and the offers you test aren't giving good conversions, you'll be left asking "gee I wonder if it's because my landers suck or if the offers suck". You need to be confident enough in your landers so that when you don't get good results, you can assume that the offers suck.

So be sure to include landers that: Are popular ("received most traffic" on Adplexity), optimized for fast loading, functions as intended (i.e. gift boxes open and wheels spin), and include at least one lander for every popular lander style for that type of offer.

-10x payout for the entire geo: If lander testing is involved, 10x payout probably won't be enough. Checking stats after every 10x payout in spend though would be a good idea - to see if there are placements/zones that are zapping the budget that you can cut, and also check if there are landers you can cut from the split-test (that have reached statistical significance as a 'loser'). You can always run more traffic for as long as the campaign still looks hopeful.

That's basically it for a basic methodology. Feel free to continue this discussion!



Amy

---

Just thinking out loud.

Instead of looking for profitable offers, would it be better for me to find LPs that work convert, as my first course of action?

For example, I'd get 1-2 offers from my AM that works well for POP. (Running POP because I can eliminate the ad variable with PUSH.) Then I'd run 5-10 LPs and see which converts the best.

Right now, I feel like I'm taking shots in the dark and it'll help me if I come up with a testing strategy.



Also, how did you find your working LPs? Was it sheer brute forcing through multiple campaigns? Or did you have a structured way of approaching it?

---

Like matuloo said, when testing, it helps if you don't test too many variables.

So I should make a decision to either test landers or test offers, but not both.

Since I don't have any proven landers, it might be best to run with an offer that my AM recommends and test a bunch of landers instead.

Does that sound about right?

---

Think of it this way:

In order to test landers without wasting money, you need a proven offer.

In order to test offers without wasting money, you need a proven lander.

But in the beginning, you have NEITHER. (Even AM-recommended offers are NOT proven offers. If they were, we affiliates wouldn't still have to mass-test AM-recommended offers to find a winner.)


This is why in the 40-day tutorial I recommend this approach to get around that catch 22:

1)Throw landers and offers into a campaign.

2)Wait for one of the offers to make the first 2 conversions (as long as the conversion rate isn't too bad - if all offers have bad conversion rates, ditch all and start with a new batch of offers).

3)Pause all offers but that one, and use it to test landers to find a winner.

4)Use that winning lander to continuing testing offers that were paused, plus any new offers you can find.


Explanation:

Why pick the offer that has made 2 conversions with a not-too-bad conversion rate? Because that would be indication that it's a good converter - good enough to use for your lander-testing without breaking the bank.

Theoretically, you could use ANY offer to test landers. It's just that the better that offer converts, the less money you'll waste to find the winning lander.

However - note that the chosen offer may NOT be the winner in the batch of offers, because we didn't run to statistical significance. Thus the need to retest them once you find a good lander.

Let me know if that doesn't make sense.

If you want more accurate results, you could treat each offer+lander combo as a split-test candidate and cut as each candidate reaches statistical significance. This approach works as well. It's just that the additional accuracy will cost more budget, is all.



Amy

---

That's a great point and I really didn't understand it until this thread.

I think that this would be a great topic for a post in the newbie section. It really drives home why your 40-day tutorial is structured the way it is.




So I had a thought. You know how twinaxe releases campaign guides? In it, they release the working LP, offer, or sometimes both.

If I use his offers to "lock in" a proven offer, would that help me test LPs?

(Or vice versa with proven LPs and testing offers.)

---

I did talk about the 'catch 22' in the 40-day tutorial. But I can always explain it better! Making a mental note to do this - as I'm currently posting the new 2021 version of the tutorial.

As for using @twinaxe's offer or lander: Yes you can do that! Same goes for anyone else you trust that may throw a solid tip your way. There are ALWAYS shortcuts and exceptions to every methodology. Nothing's written in stone. Anything that makes sense to you - just try it.



Amy

---

Let me add my 2 cents as well

Even when you work with "proven" and recommended offers there is still no guarantee that it will work same good for everyone.

It can be good as a starting point but the huge downside is that then you always have to rely on someone else.

I strongly recommend to come up with an own system that you can use again and again.

What you could do is to first run offers and LPs from my campaign guides and when you receive good conversions you use that data to build BL and WL for testing further campaigns.

That can help alot to

1. Speed up testing new campaigns
2. Decrease costs for testing new campaigns

Use these lists then to test new campaigns, the more stats you receive the better the lists become and testing only gets easier.

About landing pages, I also prefer to keep it simple there.

For example when you run sweeps you basically need only 3 landers: Spinwheel, Giftbox and survey style.

When you have a good offer then at least one of these will work in 99% of your campaigns and can be used over and over again.

The difference in performance between the different styles can be huge but the difference in performance between several variations of the same style usually isn´t that big so that I wouldn´t test variations in the first test.

---

Based on my experience with running affiliate stuff, the OFFER is the most important part.

To put it simple, even a shitty LP can convert a GREAT offer at a decent rate, but a shitty offer won't convert, no matter how great your LPs are. So I'd recommend to stick with the strategy of finding a solid performing offers first.

When I was starting, I got kinda lucky that the first offer I tried to promote was a pretty good one. So I could focus on mass testing LPs and banners/ads. That's where I applied the bruteforce method... I tested 100s of ads and dozens of LPs

---

Home > The Newbie Zone > Questions and Answers