[Tutorial] How to NOT Lose Money When Doing PPC

[tacopalypse]

i thought i'd write up this quick little math lesson for you guys.

if you apply it properly, it will save you hundreds if not thousands in the long run.
(more money for you, less money for that bastard google )

so the question we'll attempt to answer here is: "how do we know how many clicks to send into a keyword before we determine that it's unprofitable?"
i'm talking about an exact dollar amount where you can confidently say "this keyword sucks" and stop pouring any more money into it, and have the math to back it up.

so to start with, we'll need just 2 numbers:
1) the offer payout for each conversion
2) your bid for the keyword

we then calculate the required conversion rate for the offer to break-even. (that is, when your costs equal your income and your total net profit is $0)
cRate = bid / payout

if this number's unreasonably high (like over 25%), then stop right now, because there's almost no chance of turning a profit no matter what you do.
you'd have to either find a higher paying offer or lower your bid in order to put the required conversion rate into a more reasonable range, like under 10-15% to be safe.

so we start by assuming that this conversion rate is true.
then your chance of not making a conversion on each clicks is:
(1 - cRate)

and your chance of not making any conversions at all after a certain number of clicks is:
P = (1 - cRate) ^ clicks

P is the magic number here.
as the number of clicks goes up, the probability of making 0 conversions goes down, if our initial assumption about the keyword's conversion rate was true.

after P drops low enough, like say below 5%, then we can assume that our initial assumption about the conversion rate was incorrect, and therefore the keyword will not break-even. (meaning you'll only lose money if you continue to send clicks into it)

so in other words, P is the probability that your keyword will be profitable over the long-term.

so now we'll use this equation to calculate the number of clicks you need to send without getting a single conversion in order to drop P below 5%. (or whichever probabilty you're comfortable with).
note that P=5% is an arbitrary cut-off point, and increasing the value of P means that you'll be able to lose less money per unprofitable keyword, but your probability of falsely cancelling a profitable keyword would also go up. i'd recommend keeping P below 12% and above 4%.

reversing the equation above to calculate for clicks, we get:
clicks = log(P) / log(1 - cRate)

which can also be written as:
clicks = log(P) / log(1 - (bid / payout))

now just plug in the numbers and see how many clicks you'd need to send with 0 conversions in order to determine the keyword is a dud.

here's an example:
lets say you have an offer that pays out $30, and you bid $2 on a keyword.
then your required conversion rate to break even would be 2 / 30 = 6.6%.
6.6% is a reasonable conversion rate for targeted traffic, so everything looks good so far.

now you decide that you're willing to keep sending clicks until your keyword drops below a 5% chance of ever being profitable. so then we set P = 0.05
and then we plug in all the numbers:
clicks = log(0.05) / log(1 - 0.066) = 44

and that's it.
so for this specific bid & payout, if you sent 44 clicks without getting a single conversion, then you can be confident that the keyword only has a 5% chance of ever being profitable.
and so you stop dumping any more money into the keyword.


that's all.

enjoy your monies

EDIT: continued in part 2:
http://www.blackhatworld.com/blackh...scussion/353288-ppc-math-part-2-tutorial.html

 

[TimMe]

great post and thanks for saving me money in the future

 

[Luretryne]

Nice tip! Thanks. Good info when I will start with adwords.

 

[treska32]

Hi!

Very useful tip. I will present this in comming workshop about AdWords marketing. tnx

 

[Stzzzz]

thanks, nice tutorial mate

 

[jetton]

now you decide that you're willing to keep sending clicks until your keyword drops below a 5% chance of ever being profitable. so then we set P = 0.05
and then we plug in all the numbers:
clicks = log(0.05) / log(1 - 0.066) = 44

I guess I'm confused. I followed you up until this paragraph. When I plug in the numbers and do the math I do not get 44. I get:

clicks = log(0.05) / log(1 - 0.066)
Clicks = (0.05) / (.934) = .5353 (Rounded off and converted is 53)
Where is the number 44 coming from? Am I doing the math wrong? I'm not trying to be a smart-ass. I'm just getting into adwords and I think this is a great tool. Just want to make sure I understand correctly.

 

[tacopalypse]

http://en.wikipedia.org/wiki/Logarithm

you can calculate logarithms by opening the windows calculator and setting it to scientific mode.

log(a) / log(b) does not simplify to a / b

also, i specifically set up this example so that the base doesn't matter

 

[ajushi]

For PPC noobs like me, how many percent is a reasonable conversion traffic?
I'm planning on testing it out with my freebie optin page with affiliate+redirection links to the product I'm promoting. Would you recommend that?

TIA

 

[tacopalypse]

conversion rate is sort of complicated.. it depends on a ton of factors.

if you have no idea what you're doing, i'd shoot for around 1-3%.

of course you'll quickly realize that hardly any offer-keyword combinations would allow for such a low conversion rate.

which means that you'll actually have to learn some stuff about properly targeting your campaigns for maximum conversions before even starting ppc at all.

that's basically what the first equation in this tutorial shows you through numbers -
that if you have no idea what you're doing, you shouldn't even start ppc because the numbers you'll need to get to make a profit are literally impossible.
once you've done some research first and are fairly confident that you can achieve higher conversion rates (like around 10%), then you might actually have a chance at ppc.

 

[gin.of.da.game]

Nice post. Pretty informative.
Thanks, will check method this out soon for my new project.

 

[Goldie9]

nice tutorial,, bottom line is that PPC is a full time job

 

[8bitsource]

here's an example:
lets say you have an offer that pays out $30, and you bid $2 on a keyword.
then your required conversion rate to break even would be 2 / 30 = 6.6%.
6.6% is a reasonable conversion rate for targeted traffic, so everything looks good so far.

So by this example, spending $2 a click, after 15 clicks, you spent $30, the
amount you would make with 1 sale.

This means you better be making a sale every 15 clicks.

Is this right?

 

[tacopalypse]

yep. the conversion rate calculated here is the absolute minimum required to break-even.
meaning you're making exactly $0 every day.

if you want to turn a profit, you'll have to do even better than that.

however, i do recommend $0/day as the first goal for everyone venturing into ppc.
just simply not losing money is a pretty big step in the right direction

 

[jon99]

Thanks helps alot

 

[rassclat]

Useful. Thank you.

 

[rianova]

I am not yet using adwords before, but I would like use it in my new campaign later. Can someone help me how do it? Many thanks before.

 

[secretboy08]

can you pls advise in what range this should be:

"P = (1 - cRate) ^ clicks"
mine is saying that there is a probability of 0.36 that my campaign wouldn't breakeven. Is it too high? I put the clicks which are needed to breakeven.

 

[tacopalypse]

can you pls advise in what range this should be:

"P = (1 - cRate) ^ clicks"
mine is saying that there is a probability of 0.36 that my campaign wouldn't breakeven. Is it too high? I put the clicks which are needed to breakeven.

what are your actual numbers?

 

[secretboy08]

hope i didnt miss anything:
CR=0.67/29=0.023103
1-CR=0.976897
P = 0.976897 ^ 44 = 0.35755
clicks = log(0.1) / log0.976897 = 98.51

per click rate:$0.67 average
conversion:$29.

 

[tacopalypse]

hope i didnt miss anything:
CR=0.67/29=0.023103
1-CR=0.976897
P = 0.976897 ^ 44 = 0.35755
clicks = log(0.1) / log0.976897 = 98.51

per click rate:$0.67 average
conversion:$29.

that data says that after seeing 44 clicks with 0 conversions, your campaign has a 36% chance of breaking even (or doing better than break-even).

if you reach 98 clicks with 0 conversions, your campaign will have dropped to a 10% chance of breaking even.

in order to break even, you need to make 1 conversion every 43 clicks, on average.

as a side note, 36% is sort of a magic number with this equation. (1 - 1/x) ^ x converges to 0.36 as x gets larger. this means that if an event has a 1/x chance of occurring, then over x trials the probably of the event not occurring always approaches 36%.

there's a "rule of thumb" in ppc that if you spend 2-3x the offer payout without seeing any results then you should kill the offer, and this is the math behind it. regardless of what the payout and cpcs are, (1 - 1/x) ^ (2x) always converges to 13%, and (1 - 1/x) ^ (3x) always converges to 5%.

so spending 1x the offer payout with no results gives you a 36% chance of profitability.
spending 2x the amount gives you 13%.
and 3x gives you 5%.

intuitively, most people choose to kill the offer between 13% and 5% when they haven't seen any conversions yet.

if you don't want to do the math out every time, just memorize the breakpoints.