New customers,
not the same ones.

Every channel brings some new customers and some old ones dressed up as new. Someone who bought from you last year and searches your store's name today shows up in Google Ads as a sale from a brand search. That sale isn't acquisition. It's a returning customer who took the paid link because it was first.

So the first job is to know which channels genuinely reach strangers. The table is how they usually behave.

Say a store spends $6,000 on ads in a month and the platform reports 200 purchases: $30 a sale. If 80 of those buyers had bought before, the cost of each new customer was $6,000 divided by 120, or $50. Same spend, very different story. And if the 80 would mostly have bought anyway, part of the $6,000 bought nothing at all.

That's why I'd judge acquisition on cost per new customer, with returning buyers reported separately. The Lab piece on why Google Ads look better than they are goes into the counting problem. Marketplaces bring a different version of it: plenty of new buyers, none of them yours to email. The marketplaces page covers that trade.

Google Ads has a customer acquisition goal with three settings. 'Bid higher for new customers' keeps bidding for returning ones at a lower priority. 'Only bid for new customers' is, in Google's words, recommended only if you have strict budgets for customer acquisition, and needs a separate campaign for existing customers. A third setting bids higher again for high-value new customers. They work across Search, Performance Max, Shopping and Demand Gen, with the high-value option limited to Performance Max and Search.

Google needs to know who your existing customers are. If you track purchases, it builds a list automatically from up to the last 540 days, and you can add your own customer list, which must have at least 100 active members. Uploading the list is better than relying on the automatic one, because it includes people who bought in store or by phone.

Then you choose a new customer value. Google says it's added to a customer's purchase conversion, so bidding chases it, and the Conversion value column includes it. Its suggested default comes from your average order value. A better starting point is what a new customer actually spends with you again within a year, which is what the calculator works out.

Meta splits people into existing customers and an engaged audience who have interacted but not bought, and its breakdowns can show spend and results for existing customers, engaged audience and new audience in sales campaigns. Set the segments up from your customer lists and website audiences, then check the breakdown monthly. If most of the spend lands on existing customers, it's retention spend, whatever the campaign is called. The social ads page has a calculator for how long a new customer takes to pay back.

New customers are worth more if you can talk to them again, which makes consent part of acquisition. The ACMA's guidance is that a purchase on its own isn't grounds for inferred consent to market to someone. Ask at checkout with an unticked box, keep the record, and action unsubscribes within 5 working days.

The tracking behind new customer goals uses personal information too. The OAIC says the business using a third-party tracking pixel is responsible for making sure it complies with the Privacy Act. It should be disclosed in the privacy policy, and a hashed email can still make someone identifiable. I'd treat customer list uploads the same way: say in your privacy policy that you share hashed details with ad platforms.

If creators bring you new customers, the ACCC's 2023 sweep flagged 81% of the 118 influencers it reviewed. The regulator holds brands responsible too when they don't take reasonable steps to stop sponsored posts misleading people. Label every paid or gifted post.