Home > Paid Traffic Sources > POP / PPV / Redirect

---

Where to test creatives with "Staggered Bids" strategy on pop traffic (12)

---

I'm a little confused about where to test new creatives when creating staggered bid campaigns for pop traffic. From what i understand you have your main (original) campaign where you run a standard bid to uncover new sources to either add to your "profits campaign" or "low bid campaign".

But what is the best place to test new landers? Obviously you would first test landers in your main campaign since that is your only campaign, but once you uncover sources to add to either your profits or low bid campaigns, where do you test new landers?

Would you stop testing landers in the main campaign and instead test them in the profits campaign? Then when you beat your control, simply replace it with the current landers in the main and low bid campaigns?

Or would you also test new landers in your low bid campaign since those sources obviously weren't profitable using your original landers/angles so they may need a complete lander style/angle change?

Thanks!

---

There are so many ways you can approach this. I can only tell you what I usually do.

I would set up the staggered bid camps first to see which one seems to be giving me the best ROI. I don't use stats tools for this, and I don't wait for statistical significance. I just load up the best offer and lander I have at the moment (or if you're just starting out, load up some ripped landers + a couple of the best offers suggested by your AM/friends), run just enough traffic to see which bid seems to be best, pause the others, then use the remaining camp to test my offers and landers.

I won't even sort placements until after I've found a good offer+lander combo. This is because a placement's performance is closely related to how good your offer and lander are. By all means feel free to blacklist placements even while you're still testing - if you see placements that are spending considerable money and not converting. You can always re-test them later when you have a better offer and lander. But for the most part, you would want to leave yourself with enough traffic to test relatively quickly.

Also, instead of sorting placements into "higher bid" and "lower bid" camps, alternatively you could just reopen the other staggered bid camps from earlier (again, this is AFTER you have a decent offer+lander combo). Then you can cut unprofitable placements from each camp. This approach is less work than the "sorting" approach, but will take up more budget (because you'd be cutting out some of the same placements from multiple camps).

Trying to test offer and landers, at the same time trying to sort placements into different camps, would just be too many moving parts to keep track of. It also wouldn't make much sense to sort placements when you're not working with a very good offer or lander, because as you find better offers and landers via split-testing, your placements' ROIs will increase, therefore the respective optimal bids will change.

Note: It's easy to micro-manage when it comes to placements. Try not to spend too much time squeezing every last drop of profit from each camp via fiddling with the smaller placements. I usually just look at the ones that are responsible for big chunks of the overall traffic, optimize those, and only monitor occasionally. Instead, you could use that time to set up more camps to test more traffic sources/offers/landers.

Hope that answers your question somewhat! If not please let us know.


Amy

---

Amy, thanks for the such a detailed response, appreciate it! Sorry it's taken me so long to respond, i've been so busy setting up my traffic sources and affiliate networks this past week i totally forgot about this thread.

So to summarize:

First test landers and offers and limit placement cutting to big money drainers. Once you've found a winning lander/offer you can cut placements... but don't micro-manage, especially the smaller placements. Focus on the few placements that are brining in the most volume and use the time saved to test other campaigns.

Instead of starting with one campaign at a random bid then sorting placements into 'low big / high bid' campaigns, start with 3 campaigns at staggered bid's so you can see which bid is bringing you the most traffic for the best ROI. Pause the other campaigns temporarily and then test landers and offers (in your optimal bid campaign) to find a winning combo, only pausing big draining placements in the process.

After you've found a good converting offer and lander combo, then you can cut placements that don't bring you your desired ROI (30% but depends) and at the same time open up the previous paused straggered bid campaigns and simply replace your old creatives with your new winning ones. Am i right so far?

But from here, which campaign would you carry on split testing in? Your 'optimal bid' campaign and then replace the other campaigns creatives whenever you beat your control? Or test in one of the other campaigns so you can go full force profit on your optimal bid campaign, and whenever you beat your control in the other campaign simply test it out in your optimal bid campaign?

I've actually read a few posts of yours about this alternative method to the "BigEasy" but only bits and pieces... is there perhaps a thread where you've talked about it in more detail i could check out? It sounds cool, especially for traffic sources where you dont really know the top bidders bid. Thanks

---

Oh yeah, what would be your strategy for cutting placements in each campaign since you will essentially be cutting all the placements 3 or 4 times (depending on how many staggered campaigns you create). Would you just treat each campaign as separate campaign and cut independently? Or would you use one campaigns 'whitelist' as another campaigns 'blacklist'... if you know what im saying? lol

---

You're welcome! And keeping busy is always great!



This is how I do it yes. Is it the best way? I don't know. Is it a starting point? Most definitely!


There are a couple ways you can approach this:

a)Use one of your existing camps to continue split-testing like you've suggested.

b)Start a separate camp for testing, and set a low daily budget if you're happy with current profit levels and don't want additional testing to eat into profits too much.

c)Assign a lesser rotational weight to the new landers (using either an existing camp or a new camp), sending most of the traffic to your best lander+offer. However, I don't know if this will negatively affect the accuracy of the results, i.e. whether the results will still be statistically sound. For what it's worth, I've done it this way, but was never sure how valid the method was.

d)Perform split-testing only periodically, for example when your current lander/offer isn't converting as well as it used to.

I have the worst memory of anyone I know so I don't remember what I said in which post lol! Why don't we just start our own discussion in this thread? Just ask questions as they occur to you and I'll do my best to answer.


Amy

---

There are so many ways you could approach this. And frankly, I haven't given much thought as to which approach is best. There must be a "best approach" for each traffic source, and maybe when I have more experience I'll be able to answer this question better. But there are a couple of approaches I've used:

1)Run camps at staggered bids and cut placements independently like you described. This is the lazy way but will burn more money, because like you mentioned, there will be placements that will be cut multiple times. If you have more money than you do time, this may be a good approach.

2)Run one "test" camp, one or more "higher bid" camps and one or more "lower bid" camps, and sort placements from the "test" camp into higher/lower bid camps. This approach can potentially save you some money compared to the approach above, but will take more time to analyze stats and move placements around.


I kind of know what you're saying! Do you mean set up a lower bid camp first, trigger some of the lower-quality placements, then start a higher-bid camp with the placements from the 1st camp blacklisted, and so on? Is my guess even close lol?

I HAVE tested this in the past. The problem with this approach is that the same placement can perform differently at different bids depending on how a particular traffic source works. Typically, for the same placement, higher bidders' ads will be displayed to a visitor first, and lower bidders' ads will be displayed later on in the visitor's browsing session (i.e. some visitors will be subject to multiple ads/pops while they're browsing different pages on a site). Usually, the earlier the visitor sees an ad, the better the conversion rate. So by setting up a higher-bid camp, and blacklisting all the placements triggered by the lower-bid camp, we'd be killing off placements that may be profitable at the higher-bid camp but unprofitable in the lower-bid camp, if that makes sense.

And if my guess is wrong, please elaborate on your approach. This kind of discussion helps me organize my thoughts, and the insight I'm getting from you (and everyone else) also helps me to refine my own strategy. So thank-you!

Amy

---

LOL Amy you must be so tired of talking about starggered bids and whitelisting by now But i guess the more you talk about it the more you internalize it and also like you say refine your own process.

Ok the way I'm seeing this is:

a) Method number 2 of creating a 'test' campaign then moving placements to either a high bid or low bid campaign is the easiest method with a more defined strategy of cutting placements plus it's the least expensive. The problem though it determining where the optimal bid for the 'test' campaign should be. (Unless you already have a strategy for that?)

b) The method of creating staggered campaigns at multiple bids is a good way to determine the optimal bid but from there it gets messy with placements in multiple campaigns at the same time and determining a cutting strategy for them so you don't waste money cutting the same placements multiple times.

How about combining the two? So...

1. Create 3 or 4 campaigns at staggered bids and run just enough traffic to determine which bid gives you the most volume for the highest ROI

2. Then delete all other campaigns and use the remaining campaign as your 'test' campaign for strategy number 2 above, sorting placements into a lower or higher bid (whitelist) campaign.


LOL i thought of that too and also came to the same conclusions

What i was meaning though is doing it the opposite way. So taking the whitelist placements from your highest bid campaign and blacklisting them in your medium and low bid campaign. Then taking the whitelist from your medium bid campaign and blacklisting them in your low bid and high bid campaign. Then taking the whitelist from your low bid campaign and blacklisting those placements in your medium and high bid campaigns since if they're profitable at a lowest bid there's not reason to run at the highest bid.

I have no idea if that's even viable, probably so many holes in that process

---

Speaking of staggered bids, how would go about choosing what bids to come in at for the different levels? Say in ZeroPark for example, the 'average' bid says $0.01 ($10 cpm). Would you maybe come it up $7, $10 and $12?

Maybe for the highest bid, work out what is the max cpm you can afford to bid and break even with a good lander CTR and offer CVR and then come in just under that which leaves room for profit? So basically the max cpm you can afford to bid to still be profitable within your lander/offer/payout metrics?

---

This is what I was going to suggest! And what has worked for me and a strategy I'd like to stick too:

I guess with this part it starts to get...hectic pausing this and that between 3+ campaigns:

Would you just not keep them running if, for example, the placement is profitable at the lower bid and the test bid and the JUST pause the higher bid if that's not profitable? Does it matter if you're doubling up on placements at different bid levels? Not so sure it does. As long as it's profitable.

---

I like your analysis and approach! I don't remember if I've done the exact same thing before (I suspect I have), but there's every reason for you to go ahead and test it out, to see how well it works for you and go from there - to refine your strategy like you've mentioned.


Again, good thinking. However, here you're assuming that by bidding higher your conversion rate will stay the same, which is often not the case. In other words, bidding higher could actually increase your ROI. HOWEVER, just because I've pointed out a drawback, doesn't mean this strategy won't work! As we've both pointed out, every bidding strategy has its pros and cons. Basically what we're wanting to achieve with our bidding strategy are:

1)Identify the optimal bid for each placement (that will maximize profits), and
2)Do so at the lowest amount of testing cost, and
3)Do so with the least amount of time and effort.

So one bidding strategy may let you save on time but cost you more test budget, and another strategy may take up more of your time but give you better cost-savings, etc. etc. It would be awesome if we could come up with a method that could efficiently and effectively achieve all 3 goals, and maybe with more discussions like this one, and additional input from people with different insights and experiences, we will!


Amy

---

I'm afraid I don't have a standard strategy for this one! The way I choose staggered bid amounts is a combination of experience and guesswork. Here are a few considerations though:

How many bids to test: If I'm testing a bunch of different offers (not like split-testing a bunch of offers in the same camp - I'm referring to testing different types of offers, or offers in different geos), and/or I'm testing one or more offers that I have no idea how good they are, I'll just choose 3 bids - average, pretty high, and pretty low. On the other hand, if I'm testing an offer I know has promise (e.g. highly recommended by someone I trust; an international offer that is hot and has massive scaling potential) I would test 5 bids, or even more (very low, pretty low, average, pretty high, very high).

Bid info from traffic source: If the traffic source tells me what the max bid is for my particular targeting options, I'll bid close to it more or less. If the traffic source gives me a real-time estimate on the traffic levels I can expect based on my bid (e.g. popads), then I will set my lowest bid to a value that would still give me enough traffic to potentially worth my effort, but not lower.

And if the traffic source I'm working with doesn't tell you what the min/average/max bids are, I would get an estimate by looking at ad planners on other (similar) traffic sources and go from there.

Type of targeting and corresponding competition levels: Take the geo for example, if I'm targeting a competitive geo, I wouldn't go and bid 0.5cpm, whereas this bid would be a suitable lowest bid for Indonesia for example. Another example is OS - if I know the type of offer will typically convert better for windows phone traffic, I would set my bids higher than say for IOS or Android.


WARNING: Some traffic sources will charge you at your exact bid, instead of treating your bid as the maximum you'll ever be charged (which means you'll more often than not get charged lower or even much lower than your actual bid). For those sources you won't want to bid crazy high, and you'll want to keep the high-bid camps on paused until after they're approved, then activate them and watch your stats closely so you can stop traffic if it's not converting and/or it's coming too quickly. At any rate it's almost always (if not always) a good idea to set a low budget on the higher-bid camps just in case, unless and until the campaign is making steady profits.



Again, some very good and logical thinking here. The problem with this approach is that there are too many unknowns, and stuff like conversion rate will change with the bid amount so you can't treat it as a constant. It took me a long time to wrap my head around the fact that bidding higher doesn't always mean I'm sacrificing profit margins in exchange for more traffic volume. Often, bidding higher can mean a corresponding increase in traffic quality and therefore higher conversion rates. So in the end, only testing will tell you (brilliant I know!)


Amy

---

Thanks Amy, covered all possible scenarios! I'm building up pages and pages of notes on your posts, im calling it The Definitive Guide To Split Testing On Pop Traffic

---

Home > Paid Traffic Sources > POP / PPV / Redirect