Home > Paid Traffic Sources > Mobile

---

Need Advice for my targeting approaches (12)

---

HI,

I builded a rotator which rotate offers based on EPV. I have run at least 10+ campaigns, which consist at least 80-100 offers. But non of them are performing (mean non of them are making profit per day). Here are my approach.

Vertical: Mobile Billing Offers
Network: 5-8 networks
Offers: 100 Offer tested. (payout range 0.4 - 0.8)
TS: Mostly Popads& Zeropark (small amount tested at plugrush)
Traffic: MS & Adult
GEO: TH

I will have two type of campaign. Test campaign and potential campaign.
1. Test campaign
- At least 10 offers rotate base on EPV from time to time. Best EPV will gain more traffic weight.
- Each offer will have testing budget (2x or 3x payout).
- Cut placement if spend more than 2 USD.

2. Potential campaign
- Target single offer if that offer's ROI > - 50%
- budget 10-20.
- Cut placement if spend more than 1xpayout

For the first 5 campaign, it's direct linked. After that, I try to use landers. Lander are rip from Spytool. If MS offer, I use those general lander, while Adult offer, I will use porntube lander . For Test campaign, I target RON as I don't know which placement is good or bad. I even try premium or standard placement at ZP. I did try bid higher (something like 2-4 CPM). But the budget finish quite fast. If high bid traffic , the CR should higher since the payout is low. But the CR looks terrible too.

My Question: Anything step I need to improve ? because there must have something wrong with my approaches or targeting. Feel free to comment and give advice.

Best ,
Leong

---

Here are a few things I would consider taking a different approach, since in your current setup you're testing placements and offers as a bundle.

Run two pilot campaigns:
- One tested lander & offer combination for mainstream campaigns
- One tested lander & offer combination for adult campaigns

The reason why you need these campaigns is to clean out the major part of those shitty placements that no matter how many offers you rotate or bidding level your campaign has, the testing budget won't be enough. For these campaigns I believe it would be best to use landers as a must. Simple reasons:
- You get a better understanding of actual CTR per Bid level.
- Split test different angles, so you know what performs best for each traffic source.

Cutting Placements:
On ZeroPark you can easily understand what is your current competition level for a specific target (your current bid position).
On PopAds you can see current bid position of your campaign based on your targeting settings.

On both traffic sources if a publisher/target is sending a huge amount of visits to your campaign but your CTR is close to 0, you can kill it without waiting for the $2 spent.
If your lander has a conversion rate above 10% a CTR like 0.8% most probably will result profitable.

With this approach, after a few days you will end up with a list of publishers you can exclude right from the start of your Offer Test Camps, so you can save quite a lot of $$.
Also, if the Pilot campaigns result to be profitable, there's no harm on keeping them running.

Testing Offers
- Lower the number of landers & offers rotating in a single campaign. 2-3 tested landers and a max of 5 offers I believe is enough. Same for direct-linked camps.
- Divide offers based on their conversion flow and offer page. This might be good in case that some of your offers come with a pre-lander, where adding your own landing page will result being redundant.
- Divide each promising offer to a separate campaign. If you have a similar offer in an existing campaign which is performing decently, split-testing it against that existing offer would be good. You save on testing budget, as well as find out the winner between the two.

Concurrent Offers
Keep present that on some traffic sources, having several concurrent campaigns with the same targeting settings, will result in campaigns competing with each other. If you've partnered up with someone else, using more than one account will be better

I personally like to run tests in smaller lander/offers group so that each offer or lander has enough chances to express itself before cutting it down. Of course this one requires some good budget, however it can bring surprising results on the table.

Hope this helps!

---

HI Platinum,

Thanks for your great Advice. I going use tested lander and offer to filter the shitty placements before run any test campaign. What's your advice for testing budget of an offer ? 2x or 5x?

Thanks

Cheers,
Leong

---

I'm not Platinum, but statistically a 4x spend for offers works out pretty well if you've already cleaned placements and you're on desktop. It doesn't exclude many potentially profitable offers but also doesn't cost too much - a good balance.

---

Yep 4-5 times the offer’s payout would be enough, if you’ve already tested the traffic you’re sending to your funnel.

---

Thanks Courmen and Platinum.

---

My 2 cents:

1)Use the best placements for future testing. You've run a lot of TH carrier traffic on both traffic networks by now (by the sound of it), and for TH there's a lot of traffic, especially for AIS and TRUEMOVE. So just make a list of the most promising placements - they don't necessarily need to be profitable (because you were testing a variety of offers, not all of which converted well). Just pick placements have have converted, preferably with 2 conversions or more, and are not in extremely great loss.

This will save you money in future testing. Whether you're wanting to test offers or landers, test it on the best traffic first (as long as there's enough volume), then once you have a winning combination that is turning a profit, you can always retest other placements by "unblacklisting" them.

Just a reminder that whitelisting usually doesn't work too well for pop campaigns - so instead of targeting the best placements, blacklist the biggest ones of the placements you're wanting to exclude.


2)Focus on finding a good lander FIRST, and THEN mass-test offers. In the beginning, you have neither a proven offer nor a proven lander. A good way to find a good offer+lander, would be to throw some AM-recommended offers and 5-10 of the most-popular landers (i.e. highest-trafficked / most often-seen landers on Adplexity; or, for carrier-billing offers, you'd probably need to "invent" a few of your own) into a campaign, identify an offer that converts at least semi-OK (I would run the camp until one of the offers converts twice - the ROI should preferably not be too low or else it would take too much money to finish testing landers using that offer), then use that offer to progressively cut landers down to a winner. After that, you can use that lander to test more offers.

While in the process of cutting down to the best lander, you may very well lose money - the idea is to use an offer that converts at least OK, so that your lander-cutting will be cost too much. All that investment into identifying the best lander, will hopefully be recuperated later when you use that lander to test all other, similar offers you can find.


Start with these - they are bigger tweaks that will help save money and optimize your campaign more efficiently.



Amy

---

Hi Amy,

I always have this questions my mind when I create campaign or optimization.

1.Good placement VS Good Offer?
Does this placements ( I tested with others campaign with decent conversion,at least two) are consider good placement? Does it good enough for me to mass test others offers? OR on the contrary
Does this offers is suit for this " good placements" or supposedly it's a good offer but it's not suit for these list of good placements instead maybe RON or other placements is best for them .
Note: I choose same conversion flow and similar offer for testing.

2. Bid price
Initially I bid 1.00 CPM, the traffic is decent or not much enough. I increase the bid price to 1.5 or 2.0. But some placement was profitable at price 1.00 CPM but it become overprice or not profitable when I increased the campaign bid price. So, any advice for this?

Thanks


Best,
Leong

---

A placement is good when it has a decent conversion rate/roi according to the spent. If after the first two conversions or a considerable amount of spent the placement is still giving negative roi while others are performing well, it would be safe to cut it out. The vertical or offer flow may influence the conversion rate sometimes, but you have to test it.

When it comes to bids, while you are bidding below the placement's average, you may still get a small portion of it's traffic (some leftover of the unsold impression) but you need to increase bids in order to get the optimum from that one. To find the bidding sweet spot in case a publisher is becoming unprofitable when you increase your bid, running a clone campaign with different bid level may be helpful to squeeze them out.

Check out the following threads to get a better understanding on bids:
https://stmforum.com/forum/showthrea...9-Testing-Bids
https://stmforum.com/forum/showthrea...oach-is-better
https://stmforum.com/forum/showthrea...-The-Landscape
https://stmforum.com/forum/showthrea...ing-Techniques
https://stmforum.com/forum/showthrea...ic-Bidding-War
https://stmforum.com/forum/showthrea...Strategies-101

---

It is true that different placements will convert differently for different offers, but IMO it would involve WAY too much work to take that into consideration.

As far as pop traffic goes, I would just assume traffic from all placements to be "general intent" or "broad" traffic, and just leave it at that.

Meaning, I would test a bunch of offers, and pick the best offer based on the total number of conversions made, across all placements. If we start looking at which offer converts the best for each placement, it would drive us insane, and any extra returns will not justify the time/money spent on identifying the best offer per placement.

As for good placement vs. good offer: I would test a reasonable number of landers first, identify the best, use that to mass-test a fair amount of offers, then just see how many placements or traffic segments that best offer+lander can make profitable.

At the end of the day, we're not aiming to sort placements or offers into good/bad piles, we're wanting to see if we could get a campaign profitable after a reasonable amount of effort and time and budget. So, if after testing 1)the most popular landers found through spying, 2)a good amount of offers, 3)on a reasonably-high bid so that we know we're not getting junk traffic - and the camp STILL isn't looking too promising - then I would test something else. Either a new geo, or carrier, or vertical.


Basically, just increase your bid until the total profits per day starts to decrease - then revert to the previous bid.

If the total profits per day of the entire campaign no longer justifies the amount of time/effort you need to spend on maintaining the campaign, and you've already done a fair share of optimization - ditch the camp.

If the campaign performance takes a sudden and sharp dive, you can try these:

https://stmforum.com/forum/showthrea...l=1#post332874




Amy

---

Thanks Platinum. I will check the threads suggested.

---

Thanks for your great advice!

---

Home > Paid Traffic Sources > Mobile