Hacker News Reader

Comment by thinkcomp | Hacker News Reader

If Facebook tells you that you got 1000 impressions, 100 clicks and 50 likes, how are you going to prove them wrong?

If the business' value of success is driven by impressions clicks and likes, they aren't doing it right.

It would be like a development team measuring success by lines of code written - it bears no relationship with the underlying success of the business.

The identification of success/failure should be attributable to revenue or profit. They can have an internal conversion factor from clicks->$$, but it should be there.

It's actually very hard and expensive to do end-to-end reporting, so it's not as simple as you either prove profitability or not. Think of how many SMB's there are out there that have to choose between hiring a new service provider to provide better coverage or spending money on an obscure-to-them tool like a CRM. I feel for them.

In addition, time is always an issue and is the ultimate variable. What worked two months ago didn't work last month, what do you do? Every company runs into this and there are factors that make it crazy complicated: payback period, LTV, AOV, Days to Close.

Ad platforms like Google Ads and Facebook Ads talk more about "reaching your target customer" with fluffy metrics since that's what they can easily measure and, IMO, it obscures the real end-to-end results you mention. When success is obscured, it allows them to absolve themselves of educating their customers.

The ad platforms build case studies of success and then simplify it down to impressions and reach so that companies feel like they have to do it. It's a profit machine.

If you've ever worked with FB ads, you know that you can directly attribute sales ($ hitting your account) to ad groups ($ leaving your account)

You bid to hit a certain CAC (customer acquisition cost) and to the extent any numbers are inflated, it would just quickly result in a lower bid price per inflated number

... Which isn't a causal estimate of ad effect at all, as most conversions would happen anyway regardless of whether or not the ad was seen. Incremental conversions unfortunately are usually 1 or 2 order of magnitude smaller than raw number of correlationally-attributed conversions. Which makes most incrementality-based ROI numbers of advertising campaigns be negative.

I don't follow. How would the conversion get measured if the user never saw the ad? They'd need to follow a link to a landing page that is associated with that particular campaign, and that registers the conversion after the user performs the desired action, right?

You can run "ghost ads": pretend that you did bid and won an ad auction, log this information, and measure what would have happened in terms of conversions for this counterfactual set of users. Both Google and FB have experimental products to do that, as well as a couple of third party companies. It's very difficult to set up properly, and the ROI numbers don't look good at all which is why big platforms are not super reluctant to offer it broadly, even though it would be a better and more honest way of measuring advertising effectiveness.

Ah, I think I understand. So, you'd measure whether this person made a purchase without ever having viewed the ad. Thanks for explaining.

I can see why platforms would be reluctant to offer this feature. If you're running ads on multiple platforms and only one offers this feature, it will make itself look uniquely bad. However, the other platforms probably also provide little extra value compared to running ads on a single platform.

You can look at your total sales before/after using Facebook ads, or even do A/B testing where you only show Facebook ads to people within a certain region and compare sales with other regions. It's muddy, but possible, and at a certain scale it becomes obvious.

I don't think it really works like that. Most people (and organizations) are likely fine operating under the assumption that because they saw some sales increase, and they invested in Facebook ads, the Facebook ads had something to do with that. And if sales go down, it could be a number of other factors.

Additionally, targeted ads on Facebook or Google are pretty much table stakes in the ad business now. Many (or most) firms probably think it would be silly not to use them.

Reply on news.ycombinator.com