Calculating when to stop throwing money at advertising

I have an advert. How do I know if this advert will make a profit or not? When is the optimal time to stop spending money on it, if it seems likely to lose you money?

It goes like this: Advert > Landing page > sale

Currently, I am using arbitrary cut-off points. e.g. I have spent £20 on advertising, had 1,000 advert impressions and no sales. Or £20 spent, 2 clicks to the landing page and 1 sale.

I would like to have profit margin > ad cost, with 95% confidence. So I need to factor in ad spend, profits from sales, and ad impressions and/or clicks to landing page.

But I would like a scientific methodology around choosing the cut-off point, where I should stop the advert.

PPC | 👁 1300 | Posted 2018-03-28 | Share on Facebook | Twitter | Google+

| Modified: 2018-03-28 | Author:


JonL12345 2 years ago

At what stage do you decide if your ad works or not?

JonL12345 2 years ago

The point I am trying to make is that people are making arbitrary decisions on when to cut an ad loose. So, given that, it is also likely that money is being lost by a) spending too much money on an advert before realising it is not going to make a profit, or b) not spending enough money on the advert to see if it would be profitable, when in fact it would be profitable, leading to unbounded regret. In either scenario, you are losing money. Since in most cases the goal is to maximise profit, this leads to the inevitable conclusion: arbitrary decisions are contrary to your goal.

JonL12345 2 years ago

Conversion in this case is a sale. It is advert > landing page > sale

JonL12345 2 years ago

How did you decide on spending the $500, rather than $300 or $800?

JonL12345 2 years ago

When would you decide to stop the advert?

JonL12345 2 years ago

How did you arrive at 957 clicks? Forgive me if I am wrong, but wouldn't using statistics increase your probability of success? Otherwise, how do you make the decision to continue or not if you have no idea how confident you are in that decision?

odysPL 2 years ago

You should stop trhrowing money when your average CPA = average revenue per sale. ROAS = 100% I commited a post about that http://szymkowiak.online/what-is-roas

Cookedforthequeen 2 years ago

Call me a cretin but I think you may be overthinking this one a bit bud. Just run the damn ad for a bit to see if it gains any traction.

Mr24601 2 years ago

I've never seen paid advertising find a profitable strategy before spending at least $500.

Aero72 2 years ago

I have spent 20 on advertising, had 1,000 advert impressions and no sales.... with 95% confidence You don't. Say you need 1% conversion rate to get positive ROI with your given CPC, landing page, offer price, and all other factors. So out of 100 clicks you need one sale. For chi-square without Yates correction, you would need 957 clicks without a single sale to be 95% confident that your current setup won't ever be at least as good as 1%. With Yate's correction (which for sample size of 1 with a 2x2 grid is pretty much a must), you would need 3650 clicks without a sale to get the same 95% confidence level. Statistics doesn't work all that well when you talk about "one". So instead, you realize that life is too short and you are too broke, and perform some magic, and a few years later you have a successful business that allows you to spend more money at the same time on many different ads and keywords to get some actual statistical significance, and somehow it all works out. Or you go broke and never use PPC again. And somewhere along the way you start using "statistics" instead of statistics. That's when you eye the result, statistically significant or not, and act on it, because it looks like it kinda might actually be a winner. Which turns out not to be one. And the one after it. And the one after it. Until one works out.

PuttPutt7 2 years ago

You really need to let us know what qualifies as a conversion. This could be answered vastly different for different business types. (e.g. long sales process vs buying a product)

jiml1984 2 years ago

You're looking for statistical significance. Google statistical significance calculator and you'll find a bunch. You'll also find sample size calculators. And you can work back from conversation rates to clicks and impressions to work out your required budget... However, you won't like the numbers. Most people stay away from this because the numbers are so high to make it scientific that it makes your ad budget look tiny... Example from a sample size calculator: 1000 impressions to 2 clicks. If your 'conversion' is click (not sale) then you've got a 0.2% conversion rate. If you want a 20% detectable change in conversion rate then you need 270,000 impressions for 95% confidence. You're maybe better working on clicks to sales but you haven't made a sale. If your website converts a 3% then you'd need 13,000 clicks for it to be scientific... Not likely on most ad budgets.

redditforgotaboutme 2 years ago

The "science" is a/b testing. Start with a 1/4 of your budget. Test. If nothing works, start over. Once you find an ad that works, throw your whole budget at it. Repeat.

JonL12345 2 years ago

I am not sure what you mean by run the ad until it has enough data for CPA bidding? How do you define "enough" data?

verticae 2 years ago

Set your sale up as a conversion, run the ad until it has enough data for CPA bidding. Ad impressions are a virtually meaningless metric to use as a KPI on direct sales PPC advertising.

Post your Comments