Home > Paid Traffic Sources > Mobile

---

Bidding strategy for Pop traffic (22)

---

I seem to be guessing at bid amounts and think there must be a strategy which would give me better results with POP.

Currently I'm testing out Popads traffic and have experimented with bidding a highish amount and cutting website ID's which spend over 3x payout.

However, I've also tested bidding much less on RON and seem to be getting a little better result in terms of ROI.

How do I find the happy medium when starting or optomising a campaign? Any advice guys?

---

Have exactly the same problem as you. High bid always had bad roi, while going in with average bid is mostly ROI friendly. Such a dilemma everytime, haha.

Looking forward to figure out a strategy together!

---

Depending on the amount of volume and how quick you want it to be. Also, depending on the offer/ verticals you are currently running. I remember someone shared about the model structure of bidding on STM, you just need to find that hard enough.

But as for me, during testing I need large volume and quick data, I bid the highest and slowly cutting down low until you find your sweet spots. Good luck in finding yours! It may take some time, but slowly you'll get used of it. Also, bidding cost changes over time, so be sure to keep up with the bid if you aren't getting the volume or amount of data you want.

---

Yeah maybe we should have a chat on Skype sometime, sounds like you are at the same stage as me - seeing some green but still so many questions!

Felix - good comment thanks. So do you cut by placements first and then bring your bid down? or?

---

Actually both approaches have its own merits. It really depends on what you're trying to achieve in any point in time.

In the beginning stages of a campaign, ROI shouldn't be your main consideration (assuming the offer has shown enough promise for you to continue spending money to optimize further). If you bid high you can speed up your testing (as felix has pointed out), plus identify placements you can blacklist and whitelist more quickly. Also by bidding high, you will trigger quality placements that aren't accessible at lower bids.

However, bidding lower has its advantages as well. If you have a small budget and want to stretch every cent, then bidding low and testing slowly, while gradually uncovering some good placements to whitelist may be the cheaper way to go. Disadvantages would be you'd need to wait longer to cut anything, and as was mentioned before, there will be good placements you'll never discover at the low bid.

Another way to do this is to bid low first, weed out the worst placements, then bid higher to trigger the better placements.

Also, segregating placements into camps with staggered bids is a very good strategy. Bigeasy123 gives an example on how he does it in this post. Ultimately we want to find the optimal bid value for every placement, but be aware that even this optimal point will be changing constantly as competitors vary their bids, so it may be necessary to continuously analyse the most recent data and tweak bids from time to time.

At the end of the day everyone will develope their own bidding strategy, and this can differ for each traffic source. It's good to draft an initial strategy, execute it and refine it over time. I'm no expert at this and am still developing my own, but hopefully I've contributed an idea or two that will help you guys in your discussion.


Amy

---

Awesome. Thank you so much Amy for answering my threads this morning, I've just left a comment on the other one here - http://stmforum.com/forum/showthread...-cap-for-pops-

Off to read Bigeasy's post!

---

In the case of PopAds, there's another issue. PopAds charges you the 'win' amount. Means you been bidding 0.005 per pop, but maybe got charged only 0.0025.

Now what can happen obviously is that without YOU changing the bid, you'll be bidding double to what you been used to. I'd make sure to check the average prices in their reporting regularly.

Regarding a general strategy, pretty tough to give a recommendation here. Everybody has a different style, like vortex pointed out already, it's pro/contra for both. If I had to make a call, I'd prefer spending with lower bids than going in high unless you got a good idea of what you're doing.

Keep in mind with higher bids on some sources new inventory gets available. Means without changing the blacklist you have (assuming you created one already and blocked bad performing pubs) by raising the bid your overall numbers can still look worse than before, simply because a freshly appeared pub eats a part of your spendings and just doesn't convert.

---

Never thought of it that way, yes the bids are dynamically changing all the time.. I'm currently looking at Voluum to see my EPV and making sure I leave room for some margin - if I get the bid cheaper then its all good - if the bid goes up beyond what I'll make a profit on, I suppose I will stop getting that traffic until it returns to a lower amount.

Also very good point about upping the bid and seeing a worse ROI until those new bad placements that have been unlocked are cut too.

Appreciate your advice both of you!

---

Another quick question if anyone can answer...

What's your cut off for killing placements with impressions and 0 click throughs?

Currently, anything with 100 impressions without a click I'm cutting - any thoughts on this?

---

You can quickly pull up your placement stats in Voluum (or whichever tracker you're using) for that campaign, sort by impressions (or better yet, sort by conversions if you have lots of those), and just scan through the CTR column to see what the typical CTRs are. Then establish your rule of thumb such that you'll be maximizing your chances of catching placements that don't convert while keeping the possibility of cutting potentially profitable placements low.


Amy

---

Thanks Amy, makes sense although my question is more geared towards those placements that don't have a click yet so theres no actual CTR at that point.

At some point you just have to decide that the placement will be unprofitable even if it got a click - I'm guessing at 100 visits with zero clicks before I decide to cut it - I just think this will save money early on?

---

LOL! I actually DID read your question correctly.

Say from looking at your stats you notice that 80~90% of your placements have a CTR above 10%. In that case you know you can probably cut anything that has had 100+ impressions without a single click. This is just an example of course. Ideally you would want to look at placements that have gotten conversions. That would make your rule of thumb better than if you were to make a decision based on all placement data.

Another way: Check your average conversion rate for that offer on that traffic source. Do your math and work backwards to arrive at the MAXIMUM number of impressions that can occur without a single click, in order to break even. Then cut placements that go over that many impressions without a click. e.g. 1 in 10 visitors that arrive at the offer page is converting. Payout is $2. Average CPM is $1. Assuming you're direct-linking from banner to offer, you'll need at least 1 click every 200 impressions to break-even. So to be safer you set your rule of thumb to 300 impressions without a click as a threshold for blacklisting. Just know that if you choose to go this way, that you'll risk cutting potentially profitable placements because statistically speaking the data is not significant.

Ideally, placements should be cut using statistical significance. For example, using the calculator spreadsheet (see this post for details), by setting ROI to 0% (i.e. breakeven), confidence level to 50% (good enough if you have tons of placements and don't need to be too accurate), payout to $2 and CPM to $1, and then keep increasing the number of impressions until you trigger a "KILL" verdict - you can figure out how many impressions you need to cut a placement:



However, this method doesn't take CTR into account. You can save some money by using rules of thumb to cut placements a lot earlier.





Amy

---

Thanks for your help, sorry didn't mean to offend.

---

Why would I take offense?

I'm in the process of refining my own bidding strategy as well - and this discussion is actually helping me with that endeavor.

So yeah - I'm grateful you started this discussion!


Amy

---

Hi dazed1,

Vortex made a few good points there! At the end of the day, you might want to absorb all the knowledge and form your own bidding strategy. As for me, I will always go high bid and let it run for at least 24 hour, by then I should be able to identify which placements I would cut (or split test on your LP, headers, etc if you aren't satisfied wit the data generated.) It depends what you want to see and test. If you have a high CR% but low CTR%, you might want to look back at your LP, whatsoever. Same goes with vice versa.

But bottom line is, you have to understand your data and to know what you want and to test. For me personally, I may be wrong, but I don't quite emphasise on both CTR% and CR% although they are important during the initial testing to decide if the combo of your LP + Angle works or not. I will work on my CV% to make sure that every bit of penny make me profit and work backwards like how Vortex had suggested earlier on! Happy testing and launching!

---

Thanks Felix, yep going by CV% and EPV especially seems the best way to guage performance.

In terms of cutting placements which don't have clicks, I'm letting them run to 100 impressions (POP traffic and not banner) before cutting. I already know my typical CTR from performing placements and even the worst convertors tend to have 1/75 impression to click (CTR) rate.

---

Personally I feel this is taking it a bit too far. Watching campaigns for placements that have 100 impressions becomes a bit too much of micromanagement. Generally with pop it's more of a shoot shit at the wall, check what sticks approach in most cases. Losses are to be expected, but cutting after 100 impressions just because there's no click I personally feel is going to miss a lot of good placements.

Especially depending on the source, if it's popunder / tabunder / redirect etc., a lot of 'late clicks' are possible. Personally I'd use the time hunting for 100 impressions and no clicks somehow else in research or spying and focus afterwards after having some sample size for several placements and go through most of them in one batch.

---

Depends on what you want to see, cutting placements at about 100 impressions is too early to tell, you must at least let it run for a good 5000-7000 impression to even make the slightest decision or initial optimisation. My suggestions are to spy more on next few LPs that you want to test, angle/headline, ad content if you angle doesn't work, or even getting to know what is the next elements you want to add on your LPs. Keep testing!

Do not worry about spending a little bit more or to even care about your ROI% during initial testing, you're expected to have some losses definitely but remember to set your budget right. May I know what vertical you are running?

---

Thanks guys, I think I'm trying to get in the green much too quickly here and killing placements too early!

I'm running sweepstakes offers, using an interactive lander which gives the user a prize after a couple of trys. I'm still in the process of swapping out the end prize for various offers and have found one in particular giving a 0.0075 EPV - about 20% higher than my last best offer. I'm now working on the lander to see if I can improve it.

My bid is currently only 0.005 on POP traffic so running at a profit currently, but at the expense of volume and no doubt better placements above my bid amount.

It's not easy this lark is it?!

---

This thread is turning into a really great discussion!

I agree that 100 impressions may not be enough to make a decision. It would really depend on a lot of things - payout, average CR, CPM etc. If you're running a low payout offer with a low CR and average to high CPM, 100 impressions without a click could be enough to cut a placement. Like fjk87 though, I never catch anything that early because I make changes in batches as well. At the other end of the spectrum is when you have a high payout offer with really good margin, such that you only need 1 conversion every 10k impressions to turn a profit, and the offer's converting at 1%, then 100 impressions would definitely be too early to cut a placement.

I believe that a combination of statistics and historical stats for the particular camp, plus personal experience with the traffic source, should provide enough information for a person to formulate an initial strategy. At the end of the day, no method will be 100% accurate. Every time you decide to cut a placement you risk cutting something that could turn out to be profitable, but if you wait longer you risk spending more on something that could turn out to be a loser. IMHO, as long as your strategy allows you to be right more often than you're wrong, then it would already be better than not having one. As one gains experience the strategy will evolve too.

Keep those tips coming! I'm refining my own strategy based on all of your thoughts.

Amy

---

If you're already turning a profit and not getting enough volume, you should definitely test higher bids to see if you can trigger higher-quality placements.

Maybe even make a list of the current placements you're already getting traffic from and blacklist those in your new camp with the higher bid to make sure you get traffic from completely different placements.


Amy

---

Hi guys, just wanted to add I'm now using this calculator (recommended by Jenatalia on STM) which suggests an ideal bid - http://listspawn.com/Voluum/BidCalculator.php

Having a play around and it seems pretty good.

---

Home > Paid Traffic Sources > Mobile