For example: my wife has written a sci-fi novel. You can buy an eBook or paperback on Amazon. We would have to spend a few thousand dollars a month to acquire a single, purchasing reader. In contrast, I can drop around $100 on a booth at a book fair and sell 50 copies in about 8 hours. And we get almost all of those people to sign up for the mailing list (and almost all of them remember their email address well enough and have good enough handwriting to ensure it's a valid address). I don't get any chance to reconnect with my paying customers with just Amazon.
No, I'm sure it doesn't scale as well, but we don't have anywhere near the money it would take to scale online advertising to even those low of sales numbers. But we have a large enough margin on the paperback (we're self publishing and print-on-demand has gotten pretty cheap) that we typically break even for the combined print/booth costs. Similarly, growing the mailing list for when we finish editing the next book (which will be any day now) has been a lot easier in-person.
From the article:
We use them to measure the efficacy of campaigns we run our way, using our metrics.
We’ll spend a million bucks on a literal f**k ton of banners (I mean, just billions
of the things, it’s crazy). And then we’ll do targeted brand sentiment and purchase-
intent surveys using our internal peeps, online along with companies like Nielsen and
Foresee, and offline with a bunch of (really quite awesome) companies you’ve never
heard of. Then we’ll see whether the banners moved the needle, and if they did (and
they often do), we’re happy.
The basic statistics to calculate the RoI on a banner ad spend isn't particularly difficult when you have a software dev/web dev available.
No matter how much click/impression fraud there is...you are trying to track conversions not clicks/impressions at that scale.
If I know I spend $500 and get 10 sales, it isn't a statistically valid sample but its pretty clear if I break even on that $500...I should keep doing it since its essentially free marketing. Similarly, if I'm willing to eat $N per customer acquired via marketing, the same is true as long as $50/customer is equal to or less than my margin per sale + $N. "Sale" in this context is tracking what exactly those 10 people bought over the lifetime of them being a customer, generally.
The only real thing the "already successful" people have is a bunch of smoke and mirrors that honestly doesn't have verifiable value that they use to bullshit clients. Conversion rate is literally the only statistic that is verifiable, cost effective to verify, and able to give you a concrete idea of whether or not the RoI of the campaign is worth it.
We have an on staff statistician and his primary job is just automating the conversion rate calculations, the return-per-ad-campaign, etc. We spend enough money its worth hiring a staff statistician to do that but honestly any programmer with a book on statistics would 90% as good.
If you were starting a project on your own, tomorrow, and the only money you had was the money in your own savings, what would you do? You said $500 gets you 10 sales. Where? How?
Yes, if your conversion rate doesn't make sense for you, don't do it.
I'm not going to go arguing how to make $500 into 10 sales. It depends on what you are selling, what the competition is bidding for ads, your lifetime value of a customer, etc.
But you asked how you can go to $500 -> 10 sales?
2500/30 = ~83.3 clicks/day
$.20 * 2500 = $500 :: 2500 clicks
2500 * .004 = 10 :: 10 sales @ .4% conversion rate
This is really not a strange/weird/unusual scenario for a combination of 20+ low competition CPC ads on Bing, Google, etc. Given small budgets, that is what you'd aim for as you don't have the money to compete with the high cost keyword bidders.
I'm confused why you think that is hard to achieve?
These numbers aren't particularly optimistic given most people aim/expect a ~1% conversion rate, so if it was .8%, you could double your CPC, etc.
You keep throwing theoreticals around, inventing numbers to match your narrative. "If horses were 10 feet tall, then we'd be able to ford this river." A 10 foot tall horse is called an elephant, and we can't afford elephants. How do we ford this river with the one horse we have? And your answer is, "well, if animals were 5 times larger than they usually are, then I don't understand how you don't understand."
I'm saying it doesn't scale linearly.
And yes, I've done this with $500/month as the spend. If you go through my comment history you'll notice I used to work entirely with small businesses as a contractor which I've mentioned at least a couple times I believe.
I wouldn't use up-to-date hypotheticals from my current job given its a scale almost no one on HN would operate at. :/
I'm not really sure what your conviction this isn't possible comes from?
Obviously, $50/month/sale is not going to work out for a book that sells for $10 but that is a different problem.
The vast majority of ads for luxury cars, for example, aren't going to elicit an immediate purchase, but still provide value for the automaker down the line.
Obviously the ads have to eventually lead to sales, but an immediate and trackable sale isn't always an option.
In some cases, that may be right, but in general branding begins the first time someone interacts with your product (or even your industry). If I notice your business name on Google SERPS, I am engaging with your brand.
I didn't really use NewRelic for 2 years after I first heard of it, but I did keep seeing them and speaking to them at events, reading blog posts, etc. Eventually, I signed up and paid for the product. What convinced me to do that was the branding work they did before I became a customer.
Even for freemium products, the business builds a brand to encourage people to invest the effort into signing up. Even government services (e.g. in the UK - the BBC, the NHS) do branding work.
Branding doesn't start at the point the customer enters their credit card.
Lovemarks is a good read on branding - http://www.amazon.co.uk/Lovemarks-Effect-Winning-Consumer-Re...
Also, you're not sending me your credit card number, you're sending it to Stripe, or Amazon, or PayPal. Regardless, I think you severely overestimating how cautious people are with their credit card info.
What the article is saying is that there's an upper limit on the amount of fraud that can be perpetrated without showing up in other metrics, and at current prices, brand marketers just don't care. Because brand marketers don't care about impressions; they care about P&L at the brand level. If a banner ad campaign drives top line brand revenue, that's all they need it to do.
Impression based brand advertising is a different model to click based lead advertising. The former places all the risk and work on the adveriser, the later pushes it towards the publisher. If you run a website that's hugely popular and don't want to waste your resources on other business's ad campaigns then selling by impression is the only sensible option.
If you're a mom and pop store looking to spend $500 on advertisements they should be made aware there's a good chance they will be defrauded unless they know what they are doing.