For a long time, marketing was that weird, fluffy, kinda creative, kinda businessy job that everyone thought was somehow helpful, but nobody really measured to check whether that was true.
Then along came the internet, and with it, the wild world of paid online advertising across search engines and websites.
All of a sudden, rather than paying for a billboard outside and hoping for the best, you could pay extremely precise amounts for a certain customer action e.g. a view, a click, etc.
People went, “Oh shit, what if we optimise this?”, and along came the art & science of ‘performance marketing’ (disclaimer: nothing to do with sports, to my dismay).
It’s the nuts and bolts of digital marketing, but university today barely scratches the surface on how the heck it actually works.
To save you the wild Google Search goose chase, we had a cheeky chat with Bennie Liu, who leads performance marketing @ HealthMatch, a Series B healthtech startup that connects patients around the world to clinical trials.
Previously, he led growth functions at two VC-backed startups, so he’s got a strong quantitative marketing backbone to get y’all up to speed!
Here’s what he had to say about how performance marketing works 👇
So you’re meant to look after “growth”? Here’s how to think about ads.
G’day folks, so you may have heard of performance marketing or paid ads before, but you’re probably wondering: how the f*ck does it actually work?
To put a definition to the term, performance marketing is marketing where you only pay when a specific action occurs.
The two platforms you’d most likely know are Facebook and Google Ads, but there are a plethora of others including TikTok, Snapchat, native ads (Taboola & Outbrain), and more!
So why would anyone want to do it?
Companies including Square, ClassPass, Masterclass, Uber and Koala all grew large parts of customer base using performance marketing, especially in the early days.
Simply put, it’s one of the only channels out there that functions similar to a ‘tap’ - done correctly, you can turn it on/off growth without much ramp-up time. You can also scale it up and down with the amount of budget invested fairly linearly.
But isn’t paying for growth ‘bad’?
I don’t believe any channel is necessarily free. Organic channels require investment in the form of labour (content writers, specialists) and time (these channels take months to years to get going).
Performance marketing is one of the fastest channels in terms of time-to-value and one of the most scalable.
In fact, startups spend almost 40 cents of every VC dollar on Google, Facebook, and Amazon to acquire users. Safe to say - it’s a crucial channel for lots of startups!
Which sorts of companies is it best for?
I always start with the following questions:
- Product <> Channel Fit
- Volume and Cost
Product <> Channel Fit
Is there a match between the channel’s audience and your product?
Using Facebook as an example, if you’re a tractor company and you’re trying to sell five-figure tractors through Facebook ads, is that where people are most likely to buy?
On the other hand, if you’re a direct-to-consumer company or a B2C SaaS company, Facebook ads might be one of your best-performing channels.
Volume and Cost
The following diagram by Reforge is an excellent resource for thinking about it:
TL;DR: there’s a sweet spot of companies that benefit from paid, namely those who can afford the custom acquisition costs and have potential users/customers who would buy the product from the channel.
So ads could be a fit, how do you get started?
This article isn’t intended to be an A-Z of paid marketing. Instead, the focus is on a few concepts that I found to be core. Please note, the following tips are contextual to Facebook ads.
#1: Minimum Viable Tests (MVTs)
You’ve no doubt heard of MVP but a powerful adjacent concept is the MVT.
With an MVT, you’re aiming to answer the question: “What’s the smallest possible form of something I can test to get an early signal of whether this could work before doubling down?”
Working for a startup, you’ll likely be constrained in both budget and time when running ads. I’ll give a concrete example:
You’d like to test video ads on an account that has historically run image ads only.
a) Create a scrappy video on iMovie or Clipchamp
b) Hire a freelancer/agency to create high-quality videos?
Option A is almost always preferable. You’re looking to de-risk the concept before you invest further into it.
Besides, what resonates most likely isn’t what’s most polished.
#2: Silver bullets don’t exist
If you’ve looked online, you’ve no doubt come across someone recommending a certain account structure, type of landing page, or ad format that’s meant to work best.
This is typically not the case and instead, every account should be thought of as its own puzzle.
It needs to be solved on a case-by-case basis and what holds true for one ad account may not necessarily work for another.
Instead, you should work it out from first principles. Ask and solve questions such as:
- What creative would my audience be most familiar with?
- Does my landing page match my creative?
- Have I tested this before and what were the previous results?
#3: Useful mental models to keep in your back pocket:
Yep, that’s “Keep it simple, stupid”, straight out of your high school science classes. In the performance marketing world, there are infinite things to test, and clarity is the prevailing factor, like:
- How complex of a structure do you need?
- How many ads should I run?
- How many landing pages should I make?
How to answer them? Ask” What’s the smallest valid test I can run?”
With complexity, comes more setup time and more chances of something going wrong. Always start simple, make it work first, and then expand.
From Facebook’s side, they’ve been recommending an extremely simple account structure to best leverage their machine learning.
🗺 Explore, Then Exploit
You’ll want to test with high variance when you first start off an account.
In the early days, you don’t know what works, so you should structure your bets to cover wide variations.
Think of these as wide swings. Once you see signs of success, you double down and start creating smaller iterations to optimise.
Test things such as angles, creatives, formats, and account structure first and once you have winners - ask the question “why did this work?”
This becomes your next hypothesis. For example, if it’s due to the ad format, repeat the format with variations and A/B test against your old winning ad.
This forms the basis for your testing cycles. Remember, always test something specific and have measurements for what counts as a success.
Well, folks, that’s it for a quick intro to the world of paid ads. If you have any specific questions or just want to chat about growth/marketing - just reach out!