Home > The Newbie Zone > Questions and Answers

---

Possible optimizations (19)

---

Hello everyone,

I m trying to improve some campaigns i m currently running, and I m looking to find out all the possible optimizations linked to the infos i can collect about the guy reaching my lp. Here are the ones I found :

- Geo optimization (country, state, or even city)
- Device optimization (computer, tablet, phone, brands etc...)
- OS optimization : which is a subdivision of device optimization
- Day optimization (which day converts best)
- Time of the day optimizaton (which hour of the day converts best)

Do you guys have any other ideas of what kind of optimization I could use or informations I should collect from the surfer and which could help improve my conversions (informations that i don't need to request from him but that i can get through javascript/php) ?

thanks !

---

It depends on what/where you're running, what you use to track it and so on, but aside from the ones you listed right now I can only think of connection(IPs/IP Ranges, ISPs,connection types wifi/carrier etc) to be added to your list.

Aside from all of those things you listed, you can almost always improve your current, profitable landing pages even further if you haven't tried to do that already.

---

I ll put aside IP range but i ll keep :
- isp / carrier
I definitely went through that one, thanks !

As for lp optimisation Yes i m working that too

---

Hi

you can also consider if possible, the gender (female/male) of your traffic for the optimization and the different browsers they use. Also referral source/link (where from they came to your LP) I think can provide you an interesting parameter for you to better manage your campaigns.

And not related to the users, in my opinion split test on different combinations of text and creatives (LPs) might also help you optimize a campaign.

---

It only makes sense to follow and record those factors that your traffic source will let you optimize by. Its nice that you see iPad tablets do well for you, but if the traffic source wont let you block all the other devices, this info is pretty useless to you because you have to buy the rest of the traffic as well. Of course you can try to send the non-iPad traffic to other offers, but this will almost 100% not work as its too much factors to optimize, you might need different offer for every device etc ... this granular targeting never worked for me to be honest. You can try it of course, in that case you can try to collect all kinds of data like gender, interests, what they clicked on the LP, what type of site they come from ... the list could go on and on, but Im telling you, you are almost guaranteed to get lost in the data and the benefit will not be there IMO.

I prefer to keep the optimization as simple as possible and make up for the small loss in ROI by pushing volume, this approach worked the best for me.

So I stick to the most obvious like - banners/LPs/offers on my own side and spots/sites/cap/geo on the traffic cource side. If its a mobile campaign I also optimize carriers/OS/browsers/device ... and thats it pretty much I dont do day parting for example at all.

Its pretty much up to you to decide how much you want to get into this of course, but speaking about myself, whenever I got too granular, I just screwed up the campaign or lost the volume

---

Agree with matuloo.

Would just like to add:

It's good practice to run a burst of traffic first, then drill down into your data to see where the majority of your traffic comes from, and look at where the conversions are, and then decide on which traffic segments to focus on.

For example, if you're looking at OS stats, and you notice that most of the traffic is for android and ios - and then you drill down further into lander stats and notice that the best-converting landers for android are different than those for ios, then you may want to test landers separately for android and ios.

Or, if when you drill down on carrier stats, you see that most of your traffic is from carrier1, and your testing several offers that are converting at vastly different CR for different carriers, then it would be wise to cut offers based on carrier1 stats only (i.e. optimize offers for carrier1).



Amy

---

ok, there are several interesting answers here, but i ll give you more détails on what's i m currently doing and thinking so you can understand better my goald.

I run a campaign since about 6 months now. It has almost been profitable from the start (just some minor adjustements at the beginning). Once it got profitable, I never took the time to invest deeper in optimization because I have been a lazy ass. This is also the reason why I joined this forum, to find out some motivation

So, because I know you guys (and so do i) are addicted to numbers, here are a few (they won't be 100% precise though).
Over those 6 months :
Total budget spent : between 1xx.xxx and 2xx.000
ROI : between 15 and 20%

So you can imagine that I collected enough datas to find out real trends. I don't use any tracking software, but I m developping myself some stuff to improve my ROI, which is the reason why I came here to ask more informations about possible optimizations ! The reason why I don't use a tracking software is that for the one i tried, it didn't do immediately what i wanted, and i feel confident enough in my developping skills to get what i really want !

To reply to your answers :

- male/female is not an optimization I can consider because most traffic sources are not totally accurate on such a thing (except Facebook !) as you can never know who is phisycally behind the device. So I skip this.
- brower is something to be considered and i added it to my tracking !
- referral source/link is indeed something that should be considered but which i work aside, as I do for my LP layouts


I partially agree about what you say, one traffic source won't let you optimize, but the next you add might let you optimize, so if you want to have stats as accurate as possible, it s better to track it before, so that when you add a new source, you can optimize it from the start (or not if you prefer to have global stats for that specific source)

I agree that if the source don't let you optimize, you will get the traffic anyway, but then as you say, you can change your offer, or even better adapt a bit more your landing page. I ll give some examples below.

I m not interested in collected thousands of datas, i m only interested in 100% reliable data (or 99%), i should have maybe choosen a different title, but basically, I m looking for datas that i m 100% sure of :
ie :
genre : can't be sure who is really using the device
interests : would have to ask the visitor to fill a poll and he can give fake answers
while
device : this is almost 100% accurate
geo : ths is almost 100% accurate (with the good Tools)

So basically, I want to collect as many informations as I can and which I can trust at 100% !


Yes i can only agree to that.




As i said above, here are some samples (not 100% precise) from the data I previously collected :

Days stats (monday to sunday) :
My worst day in the week has a ROI between 5 and 10%
My best day in the week has a ROI between 20 and 30%

Hourly stats (from 0 to 23):
My worst hour in the day has a ROI > -30%
My best hour in the day has a ROI > 50%

Devices :
The ROI difference between the best device type and the worst device type is overs 50%

Again, those datas were collected over 6 months, I wouldn't bother if they were collected over a week and for 5$ a day budget, that would be non-sense... it s really long term here.


So as matuloo spotted, if my traffic source allows me to change my hour/day/device, then i can really improve my campaign.
And if it doesn't ? Well i can improve my landing. Simple example with hourly stats, I could setup a message saying "This offer is only valid until 11am !" (this is just a plain stupid example)

Another example, android/iOS :
everyone knows iOS users have more money, but they might be more confident in their devices (many people think iOS are far more secure). So if i sell antivirus and my iOS ROI is Under my android ROI, I could make a specific landing saying : "Damn new iOS virus, be aware hackers found out a way to fuck up your iOS" and having that tracker will allow me to know if that gets me a better ROI on that LP

Again, I m only interested in 99-100% accurate info, that I can get from the surfer without him telling me, so if anyone have other ideas, please share !

Here is the updated list :
- Geo optimization (country, state, or even city)
- Device optimization (computer, tablet, phone, brands etc...)
- OS optimization : which is a subdivision of device optimization
- Browser optimization
- Day optimization (which day - monday to sunday- converts best)
- Time of the day optimizaton (which hour of the day converts best)
- Network optimization type (wifi, 3g etc...)
- Carrier optimization (subdivision of network)


Rick

---

btw, i ll start optimizing my campaigns this month, and next months (i don't want to optimize everything at once to know what really works), and i ll try to keep you updated once a month on my global ROI


Rick

---

Well, Im not sure I would mess around with that campaign all that much since its profitable for such a long time without any further optimizations, you obviously hit something nice there so its always a risk to start messing with it. Of course I would need to know more details to really make the final decision, its hard to comment on it when I only know the general data.

The list of possible data to record with close to 100% accuracy is pretty much final tho - you can only track where the traffic comes from, what device/browser/os they use, what country is it, what carrier etc. The there is the obvious CVR patterns - hour/day/week... Not sure what kinda traffic it is, so maybe banners/lps too. And thats pretty much about it.

Before yuo start doing these tweaks, make sure you are not gonna screw something up, here are a few possible scenarios that you could run into :

1. day parting - This depends on the traffic source and their algo, but sometimes when you pause the campaign the algo is "reset" and has to re-run some tests in order to put you to some place in the ad chain. This can end up pretty nasty at times, especially with CPC networks - CPC algos usually throw a burst of traffic at the creatives and put you in the chain based on the average initial CTR. Sometimes you simply get worse CTR after a restart and that could mean a worse position than yuo had before. Let me emphasize that this isnt the case with all networks, but it can happen and it is something to keep in mind.

Some networks also favor buyers who are less picky and run 24/7, rewarding them with more traffic.

2. showing different LPs based on some specs - make sure the network doesnt see this as cheating. One LP can be ok, another one can be considered too agressive and they will reject your campaign.

3. setting time constraints on offers - some offers dont allow this as its not true, verify it before you use it.

4. redirecting carrier traffic elsewhere - carrier billing offers are usually quite aggressive and many networks rotate them - make sure the traffic source is ok with that.

5. generally sending traffic to a multitude of offers - is your funnel built to support this? I mean, is there some angle that you push? How well can unrelated offers convert with that angle? Run smaller tests first to see if rerouting traffic to other offers makes sense at all.

I could go on but Im sure you understand what Im trying to tell you.

Good luck!

---

yes i understand all your points, and that will be useful as soon as i develop it to other sources.

I deceided to go straight away yesterday with some of the optimizations we talked about, the results are pretty poor, as you expected and the reason is simple : If you pay more (per click or per impressions) to get more traffic, it doesn't mean that this traffic will convert better, and to keep the best ROI possible, i have to find the good balance. However i can't believe that without any statistics at the time and only approximations i found 6 months ago the most accurate pricing.

To give yesterday stats :
ROI was : -45%
compared to my 15-20%, it s not really good BUT :
- I increased way too much my bids (i knew that, but wanted to see the result... went up to +32% bid at some hours which increased the traffic a lot, but not their conversion !)
- We are in the second part of the month, and the offer i m running is usually doing better from 1-15 (this is probably some additionnal data i should collect, ROI depending on the period of the month (ex : 1-8 / 9-16 / 17-24 / 24: end of month)
- It allowed me find out that even on one singled day, some of the trends i predicted from the 6 months stats were approx. correct !

So today, I adjusted my bids and the max increase will be : 17% compared to my initial setting. And i ll let it run until monday as I can definitely handle the losses with those settings
My goal being to reach a ROI 15-20% but with spending more (and thus increase profit), and test that new settings beginning of next month, which should make an even better ROI !

thanks for your help !


Rick

---

Ah OK! So you're wanting to compile a list of parameters to DETECT and track in your tracker. Got you.

I've just compared your list with the data collected by Voluum, and you pretty much have everything listed.

Also, if we know what TYPE of traffic you're running, we may be able to offer suggestions on more parameters.


Amy

---

CPC right now and as soon i can confirm my trends i ll go for pops !

---

Higher bid should mean better quality traffic, but sometimes going to no.1 from no.2 or 3 is not what you want, apparently this happened in your case. The simple fact that you have a 6months old profitable campaign really could mean that you probably hit the best combo right of the bat This happens from time to time. I have one campaign running on 1 single banner that is giving me profits since april or may last year, I have tried about 200 different banners ever since, but I always return to the old one because none of the new ones can beat it for longer than a few days... go figure.

You should always increase bid in small increments, thats one of the rules that it makes sense to follow. When rasing traffic volumes, you usually wont see the same profitability. More often than not, the ROI will go down, but the extra volume should make up for it.

---

i still can't believe i would have hit the right bid just from random testing ! So I keep testing, and even if it s not totally related to the topic, here are some more infos :

Friday ROI : -35% (but runned with over half of the day with very high bids as previous day)
Saturday ROI : -22%
Things are getting better and I blame the ROI to not be better because of the time of the month.

There s also something I didn't take in considerations before and that I realised was pretty important when you have some volume : bookmarking guys and guys coming back by themselves. Hence the importance of the domain or the content of your landing page. If it s attractive enough, some guys will come back and you will acquire them later !

so I extracted some datas quickly for last month, just to see how much impact it might have.
January ROI with only direct acquisition : +-10%
January ROI including direct acquisition and guys coming back from previous months : > 22%

and i can't make 100% sure as it s cookie and IP tracked, but that's really something i have to take in consideration.

I updated my goal for this end of month to an average 5% ROI (as I only count new comers). If I can reach that, my bet is that I ll be making more money in the long run, cause I increase my daily budget by almost 70%, so in the long run, it will mean more guys coming back and acquiring them later !

Do you guys have some numbers on surfers that come back later on your LP ? Do you take that in consideration in your ROI caculation, or you get it as a "bonus" ?

thanks !

---

People are not coming back to LPs or bookmarking them, thats probably an error in your tracking - the direct clicks are probably just not tracked properly. Depending on what the LP is like, it might get indexed by google or other SEs and get a few clicks from them.

Well, maybe I shouldnt say this with 100% confidency, as I didnt see the LP. So lets put it this way - if its just some 1 page LP with no real value, then no, people will not bookmark it and return later on. If its some kind of fake blog or something that looks like a real site, then yes, some people might actually return to it. It is also possible that people keep the site open and forget about it, then discover the open page later on and convert - this happens with mobile traffic, especially when buying popunders.

I would also rerun your numbers through a calculator. You said that the site was running at 15-20% ROI, now you want to achieve 5%. Lets say you were spending $1000 per day and now you spend 70% more so thats $1700. In the first case, the profit would be $150-$200 per day, and with the higher spend it would be $85. I know you expect customers to return, but 5% still wont cut it.

Dont want to sound like a dick, but this starts to look like model example of "dont fix something that is not broken"

---

eh

As you say it was not broken, but I really think there is a possible optimization, and that will definitely cost me some money to find out.

I want to achieve 5% because end of month ROI is always worse, so if end of month ROI is 5% I m pretty sure it will be good beginning of next month.

Regarding the customers returning, you said it all, it depends on the LP, and I can't say it s a true LP it s more like multiple pages with some content which might interest customers.

Big loss yesterday again with a -45% ROI, but again, I can't stop now, I need to have at least 2 or 3 weeks of data to get to some conclusions, even more because :

- I noticed that current conversion ratios are really lower than usual (while they should be better with higher cpc)
- I expect to have one or two guys with higher bids than me to drop their cpc a bit within a few weeks (if i pay more, they pay more to stay ahead, if they don't then they will get behind me)

I probably sound a bit stubborned but I really think the datas I m collected at the moment will help me to get a better ROI within 2 or 3 weeks of time.
Still, this is a bit hard on nerves, when you know your campaign could be profitable, but that you are running losses to test out new things !

---

Nothing wrong with this, AM is about being stubborn from time to time, just dont overdo it. There is still that chance that you've hit the ideal setup 6 months ago already, so if all the attempts at optimizing this, turn out to be worse than before, dont be afraid to set everything back to what it was before.

---

Yeah i ll go back after 4 or 5 weeks of testing
In fact i m already trying to resist as much as i can from moving back to my old bids !

---

-19% yesterday

---

Home > The Newbie Zone > Questions and Answers