Maybe you run Google or Meta or Linkedin or Spotify ads.
If you do, you have 2 groups of customers. Right now, you’re probably focusing on the ‘prospect’, that future customer every business needs in order to grow. The human that will actually view your ad.
Here’s the truth about those ad platforms …
Your amazing ad creative, that landing page you spent $25k on, the video script you agonised about…
None of that matters if you can’t first convince the MACHINE that your ad is the one it should show to that prospect.
Because until the Machine is convinced that your ad is better for that prospect than your competitors’ ads, then that consumer will never see your ad & you’ll never reap the benefits of all that hard work & marketing genius that went into making it.
So how do you convince the Machine to show your ad & not your competitors’ ads?
You’ve probably guessed the answer: by feeding it better DATA than other advertisers.
I get it, you don’t want to hear more about ‘data’ & the need to be ‘data driven’ (or data informed).
Data is boring.
And yet. In a ‘black box’ world, where all advertisers rely on the Google/Meta/other algorithms more & more, the data that you feed the machine can (& should be) your competitive advantage in 2023… here’s how.
Step 1 – collect the right data
Obviously the Machine will track all the usual stuff: demographics (age, gender, location), device used & hundreds of other data points.
So you just need to focus on the outcomes you want – that are (hopefully) tied to your business objectives.
Let’s assume you’re one of the 99% of businesses that care (a lot) about how much leads & sales actually cost to acquire, so let’s start there, with those outcomes.
For an ecommerce business at a minimum track: actual sales, the revenue of those sales (probably excluding shipping, tax etc) and if the customer is new to your business or a repeat customer.
For a lead generation website, you’ve got many more options to choose from. What actions on your site are important to achieve those business goals? Do you just track successful form completions & phone calls, or do you also track the steps that lead to those outcomes (eg a PDF download, a video being watched 75% etc). Once you’ve decided on that, you’ll want to work out what each of those actions are worth to you as a business. Don’t worry about getting this perfect, it’ll change over time, just make sure you assign each action a different ‘score’ (a conversion value in google ads).
If the actual sale happens after a prospect leaves your site, then you’ll want to send that data back into the Machine as well. It’s known as ‘Offline Conversion Tracking’ & far too few accounts make use of this feature (because you’ll need a dev that understands why and how to do this).
Google has a ‘helper’ that will walk you through the steps here: https://ocihelper.withgoogle.com
Step 2 – differentiate your data by ‘scoring’ each lead
Step 1 is fairly simple & hopefully you’re already doing all of that – yay for you.
What’s going to make the difference in 2023 & help you convince the Machine to send more & more traffic to you, is how to ‘educate’ the machine on which of those various conversion ‘actions’ you want more of & which you want less of.
You do that by assigning a more accurate ‘score’ to those actions.
Let’s tackle lead gen first.
Uploading offline conversion data is great – but if it takes three months before you know if someone becomes a customer (& you know the value of that customer), then that data won’t change the automated Machine bidding much – if at all.
So what can you do?
Speed up the feedback loop.
Step 3 – get better data to the Machine faster than your competitors
If your competitors are doing all the right things (unlikely) & sending those offline conversions, then you can think about what additional data you could send – even if it’s not perfect – faster… ideally instantly!
How can you do this? By creating a ‘lead score’ for each of your leads.
If you can create a set of ‘scoring rules’, you can create that lead score instantly & send it back to the Machine every time that form gets completed.
If a human needs to be involved (usually the case for scoring phone calls) then aim to do this as soon as possible, certainly within a day & then upload that data back to the machine.
The faster you give the machine that guidance, the faster it’ll learn, and the faster it’ll show the right ads to the right prospects, and the bigger ‘black box’ advantage you’ll gain.
What about ecommerce?
For ecommerce, you might want to start thinking about:
What’s the expected refund rate of this purchase?
What’s a useful estimate of the ‘real’ value of that sale?
What’s the expected 90 day (or 1 year) value of this customer?
If your competitors are using Machine Learning to calculate the expected value of a sale & you’re not, then they will beat you in this game of digital marketing by 2024, so please, start now.
Create your own rules for this. Start simple. Get your sales team involved (they know what makes the difference between a ‘good’ & ‘bad’ sale). We suggest starting by looking at 1000 past purchases & which ones led to a full or partial refund. Put all of that information into a Google Sheet (without any personally identifiable information of course – just customer or order numbers).
Later, you can use a tool like SimpleML (a Google Sheets add on) to find patterns in your data. Ultimately you might train & deploy your own ML model (using SimpleML) for even better results. But don’t let the apparent complexity of all that stop you from starting. Today. With some simple rules.
Yes it’s confusing. But you don’t need to know ‘how’ to do all of the steps. Help is available.
But you must start – the winners in digital advertising (which powers so many small & medium businesses) will be the ones that feed the machine what they crave: more data, ever faster!
Do that & the machine will feed you more leads & sales.
And it’ll do so cheaper than your competitors can understand.
But the window to build that competitive advantage is closing.