Thread: Day 13: Testing & Optimization - Cutting Placements

***********************************

EXPLANATION & ACTION


Intro:

This will be the first practical lesson on how to optimize your first campaigns.

Specifically, let's start by learning how to optimize what are known as "placements".

"Placements" are also commonly referred to as "publishers". PropellerAds calls them "zones". Other traffic platforms may have other names for them.

Basically, placements are the various websites that display our pop ads to their visitors. We pay money to the traffic network for the visitors, and the traffic network pays a portion of that to the owners of these websites.

The important thing to note is that, due to various reasons, some placements convert better than others.

Usually, when we're running a campaign in a new geo on a new traffic source, because we don't yet have data, we wouldn't know which placements convert well, and which ones convert badly.

In that case, we would include ALL placements in our targeting, run some traffic to collect some data, then gradually exclude placements that convert badly - a process known as "cutting placements" or "blacklisting placements". Essentially, it just means to exclude/pause bad placements at the traffic source.

In this lesson, I'll offer a method you can use to optimize campaigns by cutting placements. As you gain experience, you will develop your own method, but this one would be a good start.

The criteria used in the method below assumes that you're NEITHER using landing pages (i.e. that you're linking directly to offers, which is all we've been doing so far in this tutorial) NOR have other campaign variables that can be optimized besides placements (that is, your campaign only targets a single OS, and either wifi or a single carrier). The method below also doesn't explain how to test bids.

In future lessons I'll go into how to optimize other campaign variables, and how to test bids. For your first (or first few) direct-linked campaign(s), the method in this lesson will provide good practice on how to optimize placements.

However, even if your current campaign targets multiple OSs and carriers, you can still start with this placement-cutting method. When you get to those future lessons on optimization, you can come back to optimize your current campaign some more.



***********************************

Before going into how to cut placements, let's go over a couple of small things...



Where to Find Placement Stats

You can find placement stats in two places: The tracker, and at the traffic source.

The tracker will show at least the number of visits/impressions. Some traffic sources, including PropellerAds and PopAds, will also pass cost data to the tracker. However, the visits/impressions data and cost data in the tracker is rarely accurate.

Specifically, you'll find that the number of impressions/visits shown in the tracker tends to be less than what's shown on the traffic source. More importantly, the cost of traffic - how much was spent by each zone/placement - often appears lower in the tracker than what you were actually charged by the traffic source.

So, when your tracker stats are telling you that your campaign is finally profitable, don't open that champagne just yet - because if you were to look at the stats it may still be in loss.

There are ways to manually adjust the costs in the tracker - and I have recommended doing this in the past. But over time, I decided that it's rarely worth the trouble.

Instead, for cutting placements, I would recommend to view placement stats at the traffic platform, and cut accordingly.

If you're thinking at this point, "if tracker stats are inaccurate and I'm just going by traffic source stats anyway, why use a tracker at all?"

Answer: 1)Trackers allow us to split-test landers and offers way easier, and 2)trackers allow us to drill down to 3 levels of stats or more, which can allow us to make better campaign decisions (we'll go into this in later lessons).

Back to cutting placements - below are where you'll find placement stats on the two traffic sources, and how to cut a placement on each source.


Placement Stats on PropellerAds:

-On PropellerAds' Dashboard, click on "Statistics" in the left menu, then click on the name of your campaign to bring up campaign stats. Placement/zone data is displayed by default.





(Beside the "Zones" tab (circled in the screenshot) there's also a "Targeting" tab - feel free to play with that to see what kinds of stats are available there. We won't be looking at anything but placement/zone stats for now.)

-We'll learn how to cut/pause placements later. Whenever you want to cut/pause a placement/zone, simply click on the blue toggle to turn it grey. Another way to cut placements on PropellerAds would be to edit the campaign and specify them in the "zone limitation" section:




Placement Stats on PopAds:

-In PopAds, click "Campaigns" in the sidebar, find your campaign and click on the sandwich behind the campaign name, then click on "Reports".



-Specify the "Date Range" you want to view data for. Usually when cutting placements you'd want to see all data since the start of the campaign, in which case you can select "Lifetime". Also specify the timezone if applicable (if your date range is wide enough you shouldn't need to worry about the timezone).



-For the "Additional Filtering" tab - nothing to do here. Your campaign should already be selected.



-Click on the "Grouping and Ordering" tab. Under "Group data by:" only change "Website ID" to "Enabled" and make everything else "Disabled". For "Order data by:" select either "Website ID (Asc.)" or "Website ID (Desc.)".



-Skip the "Save and Schedule" section and click "Generate Report".



-Again, we'll learn how to cut/pause placements later. Whenever you want to cut/pause a placement, simply click on "Exclude". Another way to cut placements on PopAds would be to edit the campaign and specify them in the "Website Targeting" tab:





***********************************

How to Add More Campaign Budget


On PropellerAds:

Go into "Edit Campaign" and increase the "Total Campaign Budget" by the additional cost you wish to spend. e.g. Say your offer payout was $1.20 and you wanted to spend 10x payout = $12, and you've already spent $10, so you want to spend an additional $2.

Simply change the "Total Campaign Budget" from the original $10 to $12, and make sure the "Daily Campaign Budget" won't restrict your spend. For example, if you spent that first $10 earlier in the same day, and now you're wanting to spend an additional $2, you'd need to ALSO change the "Daily Campaign Budget" to $12 or more, or else the campaign won't send traffic until the next day (when your budget "resets").

Finally, click "Save and Start".






On PopAds:

This is how you can add more budget:




In the example above, there was $2.50 left for the campaign to spend if I clicked "Start". But I wanted to run $10 more, so I added $7.50 to the budget.

One more thing before you "start" the campaign: Be mindful of your daily budget and make sure it won't restrict your spend. For example, if you've spent $10 earlier today already, and you're wanting to spend an additional $10, then make sure your daily budget is at least $20.




Lastly, click "Start".





***********************************

With that prep work out of the way - let's talk placement-cutting strategy!



STEP 1: Set a Minimum Daily Profit Goal


You want each campaign to make at the VERY least, $5/day in profits, to even justify the effort required to keep it running.

At a minimum, you need to check stats daily to make sure the campaign is still profitable, and you'd probably need to cut a placement here and there. If the campaign is making peanuts, you'd be much better off spending your time on testing more offers/landers for the same or a new campaign, or scaling another profitable campaign.

When you're more experienced and have bigger campaigns running, you likely won't want to babysit the $5/day campaigns. But for a newbie, $5/day campaigns will give you the motivation you need to keep hustling.



***********************************

STEP 2: Run 5-10x Payout in Traffic as an Initial Test

By this point, you've already run some traffic to each of your PropellerAds and PopAds campaigns.

On PropellerAds, you've run $10 of traffic. If your offer payout is $1 or less, then you've already completed this step - by running at least 10x payout's worth of traffic. If your offer payout is more than $1, please resume the campaign to 10x payout in total.

On PopAds I've specifically asked you to run 10x payout in traffic, which was easier to do because their minimum budget is only $2.50. So again, you've already completed this step.

What we do next: Evaluate stats and decide whether the offer is worth running further or not.

My recommendation is this: If after the first 10x payout to the offer, you got fewer than 5 conversions, don't continue to promote the offer.

(Disclaimer: This recommendation is based on the assumption that placements are the ONLY variable you can optimize. If you have other variables that can be optimized, such as if you're targeting multiple OSs (e.g. IOS and Android) and/or both wifi and mobile carrier and/or multiple mobile carriers, then you MAY be able to lower your bar. It's all about how much room for optimization you have.)

You'll find that a lot of offers won't even pass this initial test. Don't worry - just test more offers!


What you do NOT want to do, is take a subpar offer and try to make the campaign profitable by cutting a lot of placements. Why? Two reasons:

1)Cutting placements requires running traffic to collect data, which can get expensive.

2)Worse, by the time you cut your way from negative to positive ROI, you may not have a lot of traffic left to give you profits.


So yeah - you may as well put that money towards testing more offers to find a real winner that you can scale.

So what if the campaign has made at least 5 conversions during the first 10x payout in traffic?

First of all, go through your placements stats (as described earlier in this lesson) and cut/pause any placement that is in LOSS by 2x payout or more.

Why 2x payout? Because based on experience it would be less likely for such a placement to "catch up" and all of a sudden turn into a profitable placement.

Then, proceed to the next step.

(Note here that you could actually just run 5x payout to the offer first - if you don't make a single conversion, it would be HIGHLY likely for the 5 conversions to arrive during the next 5x payout in spend. This way you can potentially cut your test budget by half.)



***********************************

STEP 3: Run Another 10x Payout in Traffic to Verify Offer Performance

All the veterans have experienced this: Sometimes, a campaign would make a bunch of conversions on the first day, everything would look really promising - and then on the second day the performance would drop and things would stop looking peachy fast.

So, before we invest into cutting placements hardcore, it would be safer to make sure the same performance will continue.

Which is why I'm recommending to run another 10x payout to the offer.

Again, you can monitor stats and cut placements that are 2x payout in loss or more.

If after spending an additional 10x payout on traffic, you make at least 5 additional conversions, continue to the next step.

Otherwise, pause the campaign.



***********************************

STEP 4: Run Traffic Unthrottled For 30 Minutes to Identify Major Placements

If the performance keeps up, then we can consider investing in cutting more placements - because there's a decent chance that the campaign can become profitable.

But let's talk about one other thing first.


Major vs. Minor Placements:

One of the biggest problems of cutting placements on pop sources, is the fact that there are so, so many placements - anywhere from hundreds to thousands - and many of those placements are so small, that they will never give you enough traffic to be able to decide whether to cut them or keep them.

To see what I'm talking about, go open up the placements stats of one of your campaigns, sort by descending cost (by clicking on the "Cost" column heading):





Then scroll down through the placements (may need to click to go to subsequent pages) - and you'll soon see placements that have only sent a few impressions, with tiny spend.

On many pop platforms, there are tons of these tiny placements. These little buggers elude the newbie into believing that if they would only cut enough placements, they could optimize a -90% ROI all the way to +10% ROI.

What's especially bad is that some of these tiny placements would even make a conversion here or there, and because the spend is so low, the ROI appears to be through the roof.



Always keep in mind that the first conversion made by a placement should be considered a "lottery" conversion, i.e. a lucky conversion that may never be followed by another one. In other words, in my example above, I'd be crazy to expect that by spending another $0.003, I'd make another $10.43 in revenue.

So how should we treat these tiny placements? Before I go into that, allow me to define a couple of terms.


Major Placements: These are placements that, at the current bid, are sending enough traffic to "matter". Later, when you have more experience, you can define yourself how much traffic is enough to "matter". For now, let's just define a major placement as one that can spend at least 1x payout every 2 days.

Minor Placements: Placements that each spends less than 1x payout every 2 days.


What's the significance of Major vs. Minor placements?

Minor placements will take too long to collect enough data for us to decide whether to cut it or not. Therefore, do not try to optimize minor placements. Rather, look at ALL minor placements as a COLLECTIVE ENTITY, with an average daily profit (or loss) that we CANNOT CHANGE.

In other words, our goal here is to cut enough Major placements so that Major + Minor placements together will result in a profitable campaign. We'll go into how to do that in a bit.

But first, let's look at HOW we can identify which placements are Major and which ones are Minor. Again, we defined the Major placement as one that can spend at least 1x payout every 2 days.


How to Identify Major Placements:

Here's what I would suggest to do:

1)Edit campaign setting to run traffic unthrottled:




2)Set the daily budget and campaign budget to a sufficiently-high amount so that we're not restraining traffic volume, for an hour. On PropellerAds and PopAds, $100 should be enough if you're bidding $1 or less. (We've covered how to do this at the start of this lesson.)

IMPORTANT: Set an alarm on your phone so you'll remember to pause your campaign after an hour! Otherwise all $100 would be spent on a campaign that may not even end up profitable. Another thing you can do is use the scheduling feature on PropellerAds/PopAds to specify a specific hour you want to run the traffic, for example starting on the next hour.





ALSO IMPORTANT: Try to avoid running traffic between 12am & 8am local time (of the country you're targeting), as these hours typically have less traffic.

3)When the hour is up, pause the campaign, sort placement costs by descending order, and export placements stats as xlsx or csv file:




4)Multiply the cost of each placement by 24 to estimate the daily spend of that placement. I'm assuming you know how to do this. Simply start a new column, type in the multiplication formula, and copy down through all the rows of data.

If the daily spend is more than 0.5x payout (= 1x payout every 2 days), then it's a Major Placement. Else it's a Minor Placement. So - if placement costs are sorted by descending order, there would be a "cut-off" Major Placement, below which would all be Minor Placements.

5)At the end of that hour, if you want to revert to throttled traffic an a lower Daily/Campaign budget to curb your spending, you can do so by editing campaign settings. Or, just continue to run unthrottled at the $100 daily budget (or even increase that budget further to run unrestrained throughout the day) to collect data quickly and optimize your campaign quickly. (If you choose the quick way, please remember to check stats often to cut placements, as well as make sure campaign performance is still promising enough to continue.)

Oh - and please retain this spreadsheet for future reference.


Don't Stop Checking Stats!

During and after this one hour, you should continue to monitor stats to make sure the campaign is STILL promising enough to continue:

1)Cut placements that are in loss by 2x payout or more.

2)Make sure each 10x payout in campaign spend results in an additional 5 conversions or more. Otherwise, pause the campaign and test another offer.

If the campaign still fits these criteria, proceed to the next step.




***********************************

(CONTINUED IN POST BELOW...)



Amy

STEP 5: Continue to Cut Placements As Long As Campaign is Profitable or Promising

Next, continue to cut Major Placements - but only for as long as the campaign is either meeting our daily profits goal (defined in step 1 above), or at least looks like it can still reach this goal with further cutting.


How to Calculate "Minor Placements EDP"

Remember when I mentioned above, that we want to judge Minor Placements as a single entity, because it would take too long to optimize them individually?

Here is where I'd like to introduce a measurement called the "Minor Placements EDP" or the Estimated Daily Profits of all the Minor Placements collectively.

To calculate "Minor Placements EDP":

1)Export the placement stats spreadsheet - including stats gathered since the start of the campaign.

2)Sum up the number of Visits/Impressions for all the Minor Placements. Let's call this number "Total Impressions for Minor Placements".

3)Sum up the "Profit" for all the Minor Placements, no matter if they're profitable or in loss. Let's call this number "Total Profits for Minor Placements".

PopAds' placement stats includes the "Profit" column. PropellerAds' placement stats though don't have the placement profits calculated for you. So you'll need to either start a new column to calculate the revenue (= conversions x payout) and another new column to calculate the profits (= revenue - cost). Or, just export placement stats from the tracker and add up the profits there. Costs stats in trackers aren't the most accurate, but for EDP all we need is an estimate, so tracker stats would be good enough.

Note that the sum of "Profits" can be positive or negative.

4)Now, this Profits number is a result of all the traffic since the start of the campaign, which could have been a mix of throttled and unthrottled traffic, that could have been collected over a couple of hours or many hours. We need to calculate an Estimated Daily Profits (EDP) based on our current data, i.e. to find out how much profits/loss the Minor Placements would make collectively, if we ran traffic unrestricted (unthrottled with unrestrained budget).

Remember that spreadsheet from when we ran that one full hour of unthrottled traffic? Open it up, use an excel formula to sum up the number of visits/impressions across all Minor Placements, then multiply this number by 24 hours to get an "Estimated Daily Impressions for Minor Placements".

5)Finally, we calculate "Minor Placements EDP" as follows:

Minor Placements EDP = Total Profits for Minor Placements x Estimated Daily Impressions for Minor Placements / Total Impressions for Minor Placements

Note: Of the 4 variables in this equation, only the one we got from the one full hour of unthrottled traffic will be a constant that won't change. The other 3 variables will change as you gather more and more data (as the campaign continues to run).

I'll explain where to use this Minor Placement EDP in a bit.


How to Calculate "Major Placements EDP"

Similarly, we can calculate an EDP for Major Placements as well. With one major difference:

We can't optimize Minor Placements, so our profit estimates needs to include placements that are in loss as well as in profit.

Whereas for Major Placements, because we can cut unprofitable placements, we can choose to only include PROFITABLE placements in our profits estimate!

Here's how to calculate "Major Placements EDP":

1)Export the placement stats spreadsheet - including stats gathered since the start of the campaign.

2)Sum up the number of Visits/Impressions for all the Major Placements. Let's call this number "Total Impressions for Major Placements".

3)Sum up the "Profit" for all the PROFITABLE Major Placements only. Let's call this number "Total Profits for Major Placements".

4)Again, this Profits number is a result of all the traffic since the start of the campaign, which could have been a mix of throttled and unthrottled traffic, that could have been collected over a couple of hours or many hours. We need to calculate an Estimated Daily Profits (EDP) based on our current data, i.e. to find out how much profits the Major Placements can potentially make AFTER we cut the unprofitable placements, if we ran traffic unrestricted (unthrottled with unrestrained budget).

Again, open it up that spreadsheet from the one hour of unthrottled traffic, use an excel formula to sum up the number of visits/impressions across all Major Placements, then multiply this number by 24 hours to get an "Estimated Daily Impressions for Major Placements".

5)Finally, we calculate "Major Placements EDP" as follows:

Major Placements EDP = Total Profits for Major Placements x Estimated Daily Impressions for Major Placements / Total Impressions for Major Placements


How to Assess How Promising the Campaign Is

If campaign is at break-even or better, or at least close to breaking even, without too much optimization - the decision would be obvious: Try to cut further to optimize the campaign to green.

If campaign is in quite some loss, the decision is not so straight-forward. How do you know if the campaign is worth optimizing, i.e. that it would likely result in profits?

Unfortunately I can't give you a step-by-step for this, and not for lack of trying. But I will tell you what signs to look for below. Normally, you should only run a campaign for as long as it still has a good chance of meeting your Daily Profit goals.

As you run more and more campaigns, you'll develop a better sense on what hopeful vs. not-too-hopeful stats look like, and would know when to continue and when to stop.


How to Use the EDP Numbers

Finally - here's how we use the above EDP numbers:

Our goal here is to cut enough Major Placements, so that the overall campaign will meet our profits goal (of at least $5/day).

So, a positive sign to look for would be this:

Major Placements EDP + Minor Placements EDP > $5

It's very important to keep in mind that in order for your Major Placements EDP to reflect actual campaign performance in the long run (over days/weeks, sometimes months for pop campaigns), you'll need to collect enough conversions first. If you calculate Major Placements EDP too early on, when you haven't made enough conversions, this amount could just be a big fat zero.

So you may ask, "how many conversions would be enough to calculate an accurate Major Placements EDP?"

This is a difficult question to answer, as each campaign is different. But I'd say no less than 30 conversions. Preferably 50-100+.

I'm sorry for not being able to provide a straight-forward way to assess how promising a campaign is. I can only provide a list of POSITIVE and NEGATIVE signs to look for - and Major Placements EDP + Minor Placements EDP > $5 would be one of those positive signs.

Note that these are SIGNS and not CRITERIA. Just because you see a positive sign or two doesn't mean the campaign will for sure reach profits. Likewise, just because you see a negative sign or two doesn't mean the campaign will for sure fail. Keep this signs in mind and run some campaigns, and soon you'll have enough experience to spot winners from losers earlier.

Below, I'll list other positive and negative signs to look for.


Other Positive and Negative Signs to Look For When Assessing Placement Stats

The more positive signs you see, the more promising the campaign.

1)Obviously the more green Major Placements, the better.

2)The more conversions a green Major Placements has, the better. Placements with a single conversion could just be "lottery" conversions. Green placements with 2-3+ conversions are more likely to remain green in the long run.

3)The more traffic a green Major Placement has, the better. If most of your green Major Placements are smaller placements, and most of your biggest placements are red (say the top 30-50 placements), then that's not a good sign.

4)Look at all the Major Placements that have made conversions: If a vast majority of them are in loss, it's a sign the offer doesn't convert well enough for the traffic.

5)If, after cutting placements for a while, your traffic volume decreases to the point where it won't likely give you much profits, consider pausing the campaign.


Additional Tips

1)Towards the start of the campaign, you may wish to throttle the traffic (i.e. run "distributed") so that you'd have time to cut placements without wasting too much money (e.g. running placements past the 2x payout in loss). After a couple/few days though, the biggest and worst-converting placements will have been cut, so you may want to run unthrottled to speed up the optimization.

Of course, if you have time to check and cut placements multiple times a day, feel free to run at unthrottled from the start.

2)If you're running unthrottled, and get to the point where you're not finding more than a couple of placements to cut for the whole day, it means you probably have enough data to make a decision as to whether to pause the campaign or not. You don't need to wait for enough traffic from every last Major Placement before making that decision.

3)Observe trends. For example, if the top 30-50 biggest placements are almost all red (after running enough traffic), it would be unlikely for the next 30-50 placements to all of a sudden shown a lot of green. Yes there are always exceptions, but we're always talking about probability and likelihood. Always choose an approach that will make your more money or save you more money in the long run. If a method can save you money across 10 campaigns for every 1 campaign you give up on by mistake, then it's worth it.

4)If you're operating in a big competitive geo on a big traffic network, and getting a lot of traffic from a lot of placements, and you're wanting to curb your losses quickly, you may want to cut placements at 1x payout in loss or even 0.5x payout in loss. Yes you'd be cutting some good placements by mistake, but once (if) you reach profits, you could always unpause the bigger placements in batches to give them another chance.

5)When testing a new geo on a new traffic source for the first time, pick very low-payout offers! The lower the payout, the cheaper it is to cut placements. Of course, test different offers until you find one that converts well enough first! You can potentially save a LOT of money.

Think: To run 30 placements to 2x payout, an offer with $2 payout will require $120, where an offer with $0.20 payout will require only $12!

Run a low-payout offer to cut placements first, THEN run higher-payout offers.



************************************************** *******************

Think This Methodology is Overkill?

If you're thinking at this point, "that's a LOT of trouble and calculations!" I agree with you.

The method above is meant to provide newbies with some crutches. After running some campaigns, you'll most definitely want to take shortcuts.

For example:

-Instead of running unthrottled for an hour and going through all the calculations to identify Major Placements, you may just want to "eyeball" and decide to take the top 100 biggest placements (i.e. by traffic or cost).

-Instead of doing excel calculations every time you check stats, just calculate the Minor Placements EDP once and reuse that in future calculations. And for Major Placements EDP, just visually scan through the Major Placements and mentally add up the biggest green profits.


No matter how you want to cut placements, keep the following big things in mind and you won't go far wrong:

1)Be selective in choosing which campaigns to optimize. Optimization is EXPENSIVE so don't try to optimize every campaign! Set the bar high, test LOTS of offers until you have a campaign that is CLOSE to profitable or better and THEN optimize that. This is what veterans do. Don't be the noob that gets emotionally attached to every offer they run - treat them as one-nighters, not a long-term relationship.

2)Don't count on small placements. They'll take forever to optimize. Treat them as one collective entity, whether they're collectively in loss or in profit. Focus on cutting Major Placements.

3)When looking at stats, always ask yourself "how much more room do I have to optimize the campaign?" and "is that enough room to get me to green?" If the answer is "not likely" or worse, don't continue without changing something significant (such as another offer, another lander, or the campaign bid).



************************************************** *******************

Blacklisting vs. Whitelisting

So far we've only talked about blacklisting, which means optimizing a campaign by cutting placements (or other traffic segments like OSs, carriers, etc.) I prefer blacklisting over whitelisting, and here is why.

Many traffic sources prefer broad targeting - the more traffic segments (including placements) you target, the more their traffic assigning algorithm will "think" your campaign is, and the more traffic they'll send you.

If you're targeting too few segments, the algorithm can send you next to no traffic. Example: You have a blacklist campaign where you're targeting lots of placements. You see a couple of green placements making a consistent $5/day combined, so you create a whitelist campaign targeting only these two placements, expecting that campaign to give you $5 profits .

Chances are, that new campaign will only get a small trickle of traffic, far from the $5/day you expected.

Whitelist campaigns still have their place though. Specifically, one situation comes to mind where whitelisting would make a lot of sense: It's when your Minor Placements EDP is a large negative number, which would require a large positive Major Placements EDP to offset. In that case it would be easier to just start a whitelist campaign with the green Major Placements.

I would encourage you to do your own testing when it comes to whitelisting. Just make sure you have more than a few placements that have good traffic volume - the more the better.

Here's what you do to start a whitelist campaign on PropellerAds and PopAds, respectively:







************************************************** *******************

Reusing Blacklists

After optimizing a campaign in a specific geo on a specific traffic network, you'll have identified some bad placements (ones you've cut for reaching the 2x payout in loss criterion) and good placements (profitable placements with 3+ conversions).

Next time you run a campaign in the same geo to test new offers, you can and should take advantage of this knowledge to save money.

Simply clone the original campaign with all the cut/excluded/blacklisted placements:





...then update the destination url etc. etc.

And - if the new offer converts better than the original offer, you could even un-blacklist some of the placements you've blacklisted before. There ARE hopeless placements that just don't convert no matter how good the offer is, but most placements will only convert as well as your offer does, so if you have a better offer, give those blacklisted placements any chance to shine.

If you're finding after a while that your blacklist is getting so long that new campaigns aren't receiving enough traffic anymore, retest all the placements again to start a "fresh" blacklist! Remember - you can do this for cheap by using a low-payout offer.



************************************************** *******************

The above is only one part of the entire optimization methodology. There are 2 more parts:

-There are other variables that can be optimized as well. These will give us more room for optimization, so that even if we don't have 3 conversions or more in the initial 10x payout spend, we may still be able to continue running the campaign.

-And then there's bid testing, which will change how many placements we can get traffic from, which placements, quality of placements, and how much traffic from each traffic.

But I think we've learned enough about optimization for now! Let's learn how to use landers first, then we can come back to discuss optimization some more.

Thank you so much for bearing with the long lessons!




Amy

Originally Posted by bluecape

How would I go about finding minor and major placements for banners or push ads?

Major and minor placements is just a concept we can use to help us focus on the bigger placements that "matter" - placements that are big enough to be worth our attention and action (to blacklist/whitelist or tweak bids to potentially make profitable).

You can theoretically apply this concept to any traffic platform that has lots of publishing sites of various sizes, where a lot of the sites are simply too small to optimize individually, but collectively still comprise a large enough percentage of your traffic to affect your campaign's ROI.

As for how big a placement needs to be, in order to be considered a major placement: You can define it any way you like. If you set the criterion at a bigger size (e.g. major placement = one that can spend 1x payout in 6 hr.), it will give you less room to optimize the campaign, but you'd be able to optimize faster, compared to if you set the criterion at a small size (e.g. major placement = one that can spend 1x payout in 1 week).

So in short, you can set this cutoff yourself, depending on how much time you're willing to put into optimizing, and how much work (i.e. cutting) you're willing to do.

The "big picture view" of this exercise is summarized in the "Think This Methodology is Overkill?" section above.

Hope that helps!



Amy

Originally Posted by bluecape

Hi Amy, thanks for the response.

If I'm using a CPC model and using creatives, how do I tell which placements are big enough since my cost will only be based on clicks, not impressions?
Banner clicks doesn't tell me how many visitors the site can potentially have.

The questions to ask in this case would be the same, i.e. how much time are you willing to put into optimizing, and how much work (i.e. cutting) would you be willing to do?

And the criteria would be the same: The amount of spend for each placement, before you'd be ready to make a decision on whether to cut it or not.

So if your tracker can track the spend of every placement, then you'd handle this in exactly the same way as I've described in the tutorial.

**********

However, IF you DON'T know the spend per placement, and only have the CPC and # of clicks per placement, then you'd need to export the data and do some math.

Let's look at 2 different scenarios...


Scenario A: CPC is the same across all placements.

Let's say your criterion for a "major placement" is one that spends 1x payout every 2 days (or more).

And let's say your offer payout is $2, and that your CPC is $0.03.

That means a "major placement" would be one that registers at least 66 clicks every 2 days, or 33 clicks/day.


Scenario B: CPC is different for each placement.

Placement1 - CPC=0.011 - 300 clicks/day -----> spend = 3.30/day which is > 1x payout every 2 days; therefore is a major placement
Placement2 - CPC=0.026 - 50 clicks/day -----> spend = 1.30/day which is > 1x payout every 2 days; therefore is a major placement
Placement3 - CPC=0.034 - 10 clicks/day -----> spend = 0.34/day which is < 1x payout every 2 days; therefore is NOT a major placement


There's really no need to overthink this. Just take the concept and run with it. Basically the idea is to focus on optimizing placements that ARE big enough to matter (to you), and lump all the small placements together and treat it as a single entity that you won't be bothering to optimize (because your time is valuable).

I designed this methodology to prevent newbies from attempting to close a big gap in ROI by cutting a ton of small placements. Doing the calculations I suggested in the original posts will provide a good idea on whether you can cut your way to green. If the answer is no, then you won't need to waste your time.

Hope that's clear! Please feel free to ask additional questions.



Amy

Originally Posted by bluecape

Ok so what I'm understanding here is that you are saying that, for each placement, it is more about the cost per period than it is about the amount of impressions received.

Let me know if I'm understanding this right.

Because at first I interpreted a tiny placement to be a placement that always receives a low amount of traffic, regardless of the bid, that it would always be slow to spend. This is what caused my confusion.

Your understanding is 100% correct!

Because our ultimate goal is to make profits, and profits Are dependent on spend and revenue (which in turn are dependent on conversions, which in turn are dependent on CR), ALL our evaluation criteria will be defined using one or more of these stats.

The number of impressions is of secondary importance. I don't care if a network can send me high volumes of dirt-cheap traffic if that doesn't convert profitably for me. At the end of the day, all our campaign decisions need to be tied back to profits.

As for bid: Traffic volumes and performance of each placement can differ significantly at different bids, but the evaluation method and decision making described in the lesson above would remain the same no matter what the bid is.

To illustrate with an example: Let's say you have 2 campaigns that are identical, except one has a higher bid than the other. You may find that some of the placements that are considered "minor" in the lower-bid camp, may qualify as "major" in the higher-bid camp. Such placements would typically be the "higher-quality" or the more-sought-after placements, with average bid prices driven up by competition as everyone wants a piece.

In the lower-bid camp you may get so little traffic and spend from it, it wouldn't be worth waiting for enough spend to decide whether to cut it or keep it. But in the higher-bid camp it may be spending enough for it to be worth waiting for - either until it proves itself to be shit (in which case you'd cut it) or profitable (in which case you may wish to whitelist it along with other good placements in a separate whitelist campaign).

I'm really glad you asked about this - it gave me the opportunity to clear things up. A LOT of information is communicated in this tutorial - so it's all too natural to be confused and have questions.

If you have ANY further questions on anything else - you know what to do!



Amy