Hacker News Reader

Comment by soared | Hacker News Reader

A sample size of of 2,000 impressions is absolutely useless. Even the larger example is completely useless.

I don’t even considering analyzing a campaign with less than 50k impressions, and I won’t look at a variable if it’s under 10k. Realistically I want a few hundred thousand impressions before trying to make big claims about fraud.

View on HN

I hear you, but often times in B2B advertising, your target audiences are going to be small. How would you suggest assessing the success/failure of your ad campaigns if it's an order of magnitude less than your 50k threshold?

Let's not start with the assumption that we can reach c-suite at large companies in 25 dollars per lead.

Or that you can really reach them at all using social/digital ads (without missing your target on 99% of your other impressions).

Yeah, it was a wake-up call for me that b2b tech companies are first and foremost sales companies. Once place I worked the actual product was developed and maintained by a team of about 20 people that included dev, ops, and product managers but sales was about 200 people.

Just getting a meeting with a decision maker at a med-large company could take a whole team months. I don't envy their jobs.

So if your audience is actually that small, you should do account based marketing (ABM) which looks entirely different. Rather than spend money on LinkedIn ads trying to target “decision makers at IT companies” you spend your time actively pursuing individual companies.

I’ve never actually worked in a role like that, but have had clients move from abm to general b2b marketing.

Teach us: what's the mathematical calculation you use to decide when to start analyzing a campaign?

Reply on news.ycombinator.com