Home > Paid Traffic Sources > Mobile

---

stop auto-bidding on mdsp. one sneaky trick to RON with 100% roi (11)

---

It seems that lot of people will use the auto bidder function on mDSPs when they RON as part of the placement discovery process.
i use to do that, and my RONs had -50% roi at best.

Then i decided to do a little coin math. let's say you have a 2usd payout offer. your creatives have a ctr of about 2%. and your lander should convert at about 2%. what cpm do you have to pay to make 100% roi?

1000 impressions x .02 ctr x .02 cv = 0.4 conversions every 1000 impressions. or 0.8 usd per 1000 impressions.

so paying 0.8cpm, you will break even and a 0.4cpm you will make 100% roi all else being equal.

why then would you ever allow your auto bidder to pay 1.50cpm for a placement? to break even at that price you would need a cv of just under 4% and a ctr of 2%. possible. but i wouldn't bet on it happening often.

my strategy for running the network on a mdsp is this:

get some bench mark stats, or at least make up something that seems reasonable. Figure out what cpm you can pay to get 100% roi.

bid that flat price and RON the exchange.

you will most likely get a pretty low win rate, and maybe not much traffic. but if your offer converts at all. you'll probably break even on the RON.

if you cut out all the shit placements that didn't convert, you'll have a profitable campaign.

Now you can use this data to crank things up.

how are your ctrs, how is your cv? how much more can you bid?

you can also use these low costs converting placements to test creatives and LPs on the cheap.

if you're using go2mobi you can set a specific bid for each placement. i always do this. i test different bids for each placement, high low.

once you have more data about how your camp performs in the wild you can start cranking up the bids, to increase your WR or find new placements. But you will be doing this from a position of strength, not just shooting in the dark hoping for the best.

knowing your numbers allows you to operate far more strategically. the challenge is to get the data as fast and as cheaply as possible.

i've used this exact strategy to do 100% roi on a RON with minimal optimization. making over 1k profit on the first day of the campaign.

i would love to hear other peoples strategies for profitable rons.

---

ps. this post is based on the assumption that you are only testing placements when you RON. not creatives or landers or offers. you should only test those on proven placements.

---

So basically you are direct linking with this strategy or that you don't pay any attention to your LP's? Even not ones withhorrible CTR's?

Interesting approach.

The issue and this is something that I hope someone can correct me is that when I first test a campaign I want it to have the best chance to succeed before I decide not to run it. And by making a low bid I'm always afraid I will get shitty quality and I won't know for sure If that was the reason it didn't work. What do you think?

Also, every mobile DSP acts differently. Did this method worked for you on all mobile DSP's?

---

Thanks for the post, planning on testing out a couple mdsp's today, will deff take a few hints and nuggets out of your strategy ! Any mdsp that you recommend? Thanks!

---

i haven't tried all dsp's. it works on go2mobi and avazu. i've probably run 50 campaigns with this method.

usually i test LPs with pop traffic. so by the time i get to display, i already got an lp offer combo sorted.

I never direct link.

when you say test a campaign, do u mean test an offer?

---

for me avazu and g2m perform very differently. in almost everyway. i don't really know why. generally i use avazu for volume scaling, and g2m for testing.

---

Thanks for the strategy, hlyghst. That's some food for thought.

Do you run on Zeropark with this method?

Edit: Okay, I saw in another thread that you don't spend much time with ZP, since there are some better sources.

---

What sort of verticals does this work in? can you do it with sweeps?

---

i've done this with sweeps. and apps.

i don't know if this would work with higher payout offers. never tried. but think it works well for sub 2usd payouts.
that way you can get data pretty quickly. cut placements fast.

---

Not only that, but you can actually "drill" into your top performing placements, and create reports based on carrier/make/model/OS version. You can get extremely deep into the data to maximize your gross profit. I've been realizing that many advertisers, who are very profitable, leave a lot on the table simply because they don't know that our platform offers this capability.

hlyghst is a smart man, though. Bid overrides are absolutely your best friend, especially when you start cross referencing placements with carriers and devices.

---

Thanks for this nugget. I am getting started with mDSP and am looking to achieve similar cost-efficiency in testing.

Are your banners more orientated to pre-selling the user or catching their attention? e.g. Win an iPhone vs System notification
Do you still use auto-bidder to get benchmark stats before setting a low bid?

And with regards to using pops to initially test landers:

How much did you spend on US pops to find a placement that you can isolate? I tried to run on historical placements on Popads and lost $80 in 1-2 hours. ROI was baaad.
Do you bid aggressively or start low?
Any preferred traffic source to achieve this? I have had some really good placements on Zeropark, but they also mysteriously disappear.

---

Home > Paid Traffic Sources > Mobile