The following is my free translation of a speech by a cool dude named Tim Burd At the AWAsia 2018 conference.
It doesn't matter how much dough you pour dailies: 100$ or 20000$ - this method works on any budget and will reduce your costs. It also works at any stage of the drain: testing, optimization, scaling - whatever.
WARNING: if you have never poured on a manual bet and you are a beginner, do not use this method. It's dangerous and you can drain a lot of money very quickly if you don't do it right. Practice with manual betting before you start using this method.
So, 10 simple steps:
- Create a new campaign for Conversions. You can have a CBO, you can have a CBO, it doesn't matter;
- Create a MINIMUM of 5 adsets with different targeting: interests, lookalikes, broad to broad;
- Put a manual rate: "Limit Rate".
Translator's note: right now it would look something like this:
And whatever your normal price per lead is, put a bid on it half the size of her.
For example, you normally have CPL=40$, put 20$ in your bid settings.
4. Set "Delivery Type" to "Expedited Display".
5. Load 4-5 creos into each adsetDon't shove 30, but don't make only one. It's best to make different types of creos: picture, video, ring gallery, etc. The reason why you should make different creos is Facebook auction, which is roughly divided into parts like this:
More than half of the auction is "given" to videos. If, say, you only upload pictures, you will get only a small part of displays.
By shoving all kinds of creo into adset, you allow FB to give you the display regardless of what device the user is sitting from, and what he is used to click on.
If you pour in one thing, you won't get all of the possibly suitable audience, but only a segment of it.
6. Set a budget.
Your budget should be between 500$ to 5000$ per adset. DO NOT worry, you will not be spending that much. Our goal is to spend between 10-20% of that dailies budget. If you are using CBO, then set your campaign budget to 5 times what you actually want to spend.
7. Launch a campaign. Time for the stealth attack!
8. Watch the campaign closely and slowly raise the stakes.
In the beginning you should get no traffic at all, that's good! This means you're doing everything right. If you are getting traffic, your initial bid was too high.
Let's say we started with a bet of 20$, we raise it to 22$ - at 10%. Then we wait a few hours! If there is no traffic at all, or if there is traffic, but the shows are too slow, we raise it again to 24$. Then we wait again, and this is IMPORTANT!
When you change your bet, it takes 15-30 minutes before it is updated on all FB servers, if not longer! If you get nervous and raise your bet frequently or by a large amount, you can lose your dough VERY quickly!
So, let's say you've reached the point where traffic is coming in, but it's pretty slow. Let me remind you, our goal is to spend 10-20% of the set budget. Let's say we set budget = 500$, so we want to spend 100$ dailies. But if you look at the traffic and see that the spend is only 1 buck per hour, then obviously the bid needs to be raised!
Your goal is to get exactly as much traffic as you need, no more no less.
Well, that's a warning:
9. Set a couple of auto-rules
If you put no more than 500$ on adset, you can use the built-in FB autorules, but if you put 5000$, you need third-party tools like RevealBot. The thing is that FB auto-rule works every half an hour, while RevealBot works every 15 minutes. When the spread is large and it goes very fast - this can be critical.
Let's say it's 10 pm and some big rekl stops its advertising for the night. The auction is free and you will be flooded with traffic, and fb will spin off your entire budget for 2 hours, remaining until the end of the day!!! And the price per click will be exorbitant - 5-10 times more than usual!
So these auto rules are very important, they will protect you from such a situation.
So here's the first auto rule:
Turn off adsets without leads (Purchase is here, but we, as arbitrageurs, replace it with Lead) and the spread is larger than 60$. We initially want leads at 20$, right? Here we put the amount 3 times more, 60$, before the adset is turned off.
The second auto rule:
This rule is for turning off adsets that have leads, but the price for them is too high (Again, don't forget to change Purchase to Lead). Remember where we started? Your normal cost per lead=40$, so as long as it doesn't exceed that figure - lol gut.
The same 2 auto-rules in the RevealBot interface:
Here are a couple more auto-rules for those advanced in RevealBot:
- Increasing the bet by 10% every hour if the spread is less than 100$ is essentially automating our manual labor.
2. Almost the same, but we look at the last 2 hours instead of the day.
10. Optimize and profit!
Do everything as you normally would. Turn off those CREOs that do not bring leads or bring them at too high a price. And only then turn off the adsets. A lot of people turn off adsets right away, even though there may be some one in the creatives that ate all your money.
As soon as you have less than 5 working adsets, make new ones! Try new targeting, new lookalikes - anything.
And if you did everything right, you will reduce the price of a lead by at least 30%. Although we have had clients who managed to reduce the price by 50% using this technique.
That's all for now, with you as interpreter was Vygolov Daniel, ley in plus, gentlemen and send donations, I'll translate something else good for you.!
P.S. The full video in English can be viewed at right here.