1 Geo, 1 Carrier, 1 TS, High XX/Day Profits - Part 2 (7)
1 Geo, 1 Carrier, 1 TS, High XX/Day Profits - Part 2
[Update Aug/2018: It has come to our attention that this traffic network may have closed down - please hold off on joining until further notice.]
(This is Part 2 of this case study. Part 1 can be found here.)
27-Aug-2016, Morning:
Compiled a list of suitable 1-click carrier-billing offers. Made an excel chart to note the carriers and OSs accepted by each.
Created offers in
Gold Tip #2: For this particular traffic source (i,e, wwwPromoter), I've had the best experience with bidding low first and increasing gradually to maximize profits, as opposed to testing staggered bids, where the higher-bid camps would take most/all the traffic from the lower-bid camps. Also, of all the high bids I've tested, I haven't seen a corresponding high increase in conversion rates. So starting with a low bid is recommended.
So I set my bid at 1.33 for all camps. (Aside: Many people will not bid at whole dollar amounts, to avoid competition from newbies; as more people started bidding 10 cents over the dollar, some people started bidding 20 cents over; so I like to bid at 33 cents just to be different.)
Set a daily limit of 10k impressions for each camp - I like to run a small bit of traffic first, just to see if anything will convert before deciding whether or not to run further.
Waited for traffic to collect some data.
27-Aug-2016, Afternoon:
The camps have collected some data:
As far as optimization goes, I was already using the best-converting lander, and was targeting carriers separately, so basically the remaining major traffic segments to look at would be OSs. So the goal would be to mass-test and cut offers, followed by OSs, and then placements. Will need to test bids and daypart as well.
So for now let's look at offer stats and OS stats for each carrier camp.
Hutch
The Hutch offers looked a bit bleak:
More data is needed to say anything conclusive. So I ran an additional 10k impressions to this camp. It was exhausted an hour later. I checked stats again:
I checked campaign stats on wwwPromoter, and saw that the total spend on the Hutch camp was $26.61, so approximately 1/4 of that went to each offer (~$6.65). Hutch Offers 1 and 4 definitely have some promise. Just for the sake of being complete, here are the OS stats for the best 2 Hutch offers:
BTW - this was how I checked campaign spend on wwwPromoter: Clicked on "Reports" tab > "Campaigns" to go to the "Campaign Report" screen. Selected the "Campaign" and "Advert", specified time range (I selected "Last 30 days" to see all stats), then clicked "Run Report" to generate stats at the bottom:
Airtel
Below are overall Airtel offer stats, and OS stats for each offer.
Both offers were in profit - although there wasn't enough conversions to be sure. However, there wasn't enough traffic volume to be worth my time. I would definitely test these offers on other sources and hopefully find more traffic elsewhere.
Etisalat
Below are overall Etisalat offer stats, and OS stats for each offer.
Again - both offers were in profit - although there wasn't enough conversions to be sure. And again, there just wasn't enough traffic volume to be worth my time. I would definitely test these offers on other sources and hopefully find more traffic elsewhere.
Mobitel
Only one of the offers made 1 conversion. Payouts were all in the range of $0.18-0.28. Total spend on camp was $13.30, which worked out to around $2.60/offer. Looked pretty hopeless so decided not to run further traffic.
Dialog
The Dialog offers looked extremely promising. Below are overall offer stats, and OS stats for each offer:
As you can see, 2 of the offers were green. Definitely needed to continue running this carrier!
It also looked like some offers may have reached statistical significance for cutting, so I tested on the split-test calculator at:
https://win-vector.shinyapps.io/CampaignPlanner_v3/
Offer 1 was in the lead, so I compared offers 2 and 3 to it, respectively.
Comparing Offers 1 and 2:
At a 96% "probability of being best" for Offer 1, Offer 2 was ready to be cut. Note: I've recently increased my cutting percentage to 95-100% from the original 90%, because the offer is such a crucial part of a campaign that I literally wanted 100% accuracy (or close to it). So I cut Offer 2.
Comparing Offers 1 and 3:
100% "probability of being best" had been reached - cutting Offer 3. Before I cut it though, I checked the OS stats to make sure that it wasn't performing exceptionally well for any particular OS. I would do this only for the major traffic segments, mainly the bigger OSs and carriers. A "loser" offer may turn out to be a winner if it converts very well for a major traffic segment - in that case all I had to do would be to start a new camp with the offer and that traffic segment.
This kind of "last check" will ensure that I don't throw the baby out with the bathwater.
Now - because the camp was in profit, I lifted the cap on daily impressions - simply by deleting the value to revert to the default value "unlimited":
Observations and Next Steps:
Mobitel looked hopeless, period.
Etisalat and Airtel offers all look promising, but there wasn't a lot of traffic on this traffic source for these carriers so would have to scale to other sources.
Hutch had the most traffic volume, and 2 of the offers looked like they had a chance, but it would likely require quite a bit of cutting placements to get green.
The most promising carrier was Dialog - 2 offers were already green from the get-go. I decided to focus on optimizing this carrier first, then maybe reinvest earnings into optimizing Hutch later.
Gold Tip #3: Mass-testing offers is much better than testing a few random offers. Not every test will reveal such promising offers! Mass-testing offers is a very good approach, because it will allow you to pick the best offers to scale. This approach is miles better than taking a random offer and trying to make it green by mass-cutting placements and other targeting - the latter is a lot more costly and takes a lot more time. Considering many pop camps don't perform well for very long, speed is especially important, so you can't afford to spend too much time or money on optimization - because the camp may die before you can recoup your investment.
Note: Because most of the traffic came from Android, IOS hasn't had a chance to see which offer would convert the best. It may be worth the trouble to set up a separate camp to retest all offers for IOS later (probably on another traffic source, because there's too little IOS traffic on this source to matter probably - although I wouldn't know for sure as I wasn't bidding very high).
Next step: So I had decided to focus on just the Dialog camp. I had cut down to the best offer. Most of the traffic was from Android so no need to cut OSs for now. The remaining optimization tasks would be to test bids, cut placements, and implement dayparting.
Please continue to Part 3 here!
Amy
OMG, where do you find the time to prepare such detailed threads? 
Thumbs up!
Wow such gold! Vortex strikes again!
awesome post! thanks for sharing it with us
You are awesome vortex. Thanks for the awesome share.
Thanks for this post Amy!
How did you choose which offers to run? Or did you simply run all the offers available to you in said GEO?
Thanks!
Originally Posted by q-west
You can do this for any major vertical. Rip lots of landers and test them as normal for a few geos, identify the 1-2 that convert the best consistently across most/all geos, and just use them to test other geos.
Amy
Home > General > Affiliate Marketing Forum