One of the toughest challenges in digital marketing is knowing when it’s time to scale spend, how much to scale, and which channels make sense to grow.

My approach to scaling digital marketing spend is fairly straightforward – it simply has to be earned. There are always going to be the core set of performance marketing channels that simply have to work for an ecommerce business to succeed – 1% lookalikes on facebook to drive prospects, Google Shopping at low bids on select products, and retargeting / retention marketing for example. If you aren’t getting positive returns on these channels, it’s time to go back to the drawing board and improve conversion rate – whether that be through adjustments to the classic 4 P’s, landing page experiences, checkout experience, or any other number of factors.

And how does one determine if you’re generating positive return from a channel? The best way to go about figuring that out is pulling a payback period by channel. There are going to be some inherent assumptions in this calculation related to the chosen attribution methodology, cross-channel mix, and several other factors, but I’ll ignore those issues to simplify things for now.

Payback period is fairly simple – first you need to pull the annual revenue generated from the typical customer acquired through a given channel. This can be done by matching the transaction IDs in your marketing analytics platform (GA for example) with your ecom data to pull the average customers average number of orders and AOV for that year. Once you have that, you compare it to the cost it takes to acquire the average customer through that same channel (we’ll ignore retention marketing costs for now as well, but they should be factored in). Once you have the annual LTV and the cost to acquire you can determine how long it typically takes for an order brought in through Facebook to be fully paid off – or the time it takes for the average user to generate more profit for a business than the cost it took to bring them in.

Each business is going to be different – for companies with enormous runways and huge funding, maybe you can get away with a five year payback period in the name of growth – but for company’s that are bootstrapped and/or focused on hitting strict profitability targets, that payback period is going to need to be much shorter.

Scaling Paid Social and Paid Search

Let’s start with scaling the core channels – social and search. Each of these channels will have areas that are no brainers – let’s say I run an organic fiber childrens’ clothing company for example: I’m going to have my core lookalike audiences, interest/demo based audiences surrounding organic products for people with kids, and keywords around organic children’s goods. There’s going to come a point in time where we need to scale out of these core audiences though.

Once we’re generating short payback periods in this channels, and profitably scaling in areas we know work, it’s time to get to work. If you’re having trouble scaling in your core data sets, once again, it’s time to go back to the drawing board and figure that out before forging into foreign territory. One of the biggest mistakes a business can make is trying to mask poor performance by throwing dollars at ‘tests’ without shoring up what’s broken – there’s no point in driving a leaky boat further out to sea, it’s better to fix it before hitting the gas.

Now that we’re running our core programs profitably, the key to scaling into new areas is to crawl before running – this means small tests into new tactics.

As mentioned, we’ve scaled out audiences in keywords specifically pertaining to ‘Organic’ kids clothing, where else can we go – what about general children’s clothing audiences? Fans of Mommy blogs? Non-organic kids’ clothing keywords? Can we compete in these places? This is the time for small tests with low dollar amounts in all these new areas to see which pockets of audiences and keywords actually work – then once (if) they succeed, they’re brought into the core program – if they fail at these low levels of spend, we simply turn them off and continue testing in new areas.

Another way to look at this is termed the Alpha/Beta approach. You have a core set of strategies that you know work and drive your business – these are always on and belong in your Alpha bucket; they’re driving the vast majority of your business. Your Beta bucket consists of new/test strategies – the many different audiences and keywords that you’re not sure work. Once something is generating consistent and statistically significant positive value for the business, it moves over to the Alpha set – any strategies that, over a stat sig level of investment, don’t work out, are simply eliminated from the Beta set and replaced with something new.

It’s up to you how much you want to invest in testing new strategies, and it really depends on the needs and pressures surrounding each individual business. Many companies use an 80-20 rule (80% to core strategies, 20% to testing) which is somewhat arbitrary but helpful as a starting point.

Scaling Beyond Paid Social and Paid Search

The decision making process for when it’s time to scale beyond the core marketing channels mirrors the process for deciding when to scale within them – it all comes down to earning the right to scale and grow with sound business metrics. Once, based on profitability and payback period, it’s been determined that scaling out of Paid Social and Search is the best course of action, careful testing should be the chosen method.

In performance marketing, it never makes sense to go all-in when you can simply dip your toes in the water with no repercussions. For example, if you think the New York Times is perfect for your brand – why spend tens of thousands buying ad space direct when you can test for a few hundred dollars via programmatic display? Similarly if you want to test influencer marketing, why not gift some product and spend a few hundred dollars on twenty micro-influencers rather than five figures on a single major influencer?

For most tactics outside of Paid Social and Paid Search, there are almost always tactics allowing one to test the waters in a way that is low risk and inexpensive – I’d always encourage careful testing for new channels at a low volume to get an idea of success, then these channels can all be scaled up when testing proves effective (or shuttered if they prove inefficient).

One of the biggest mistakes a brand can make is going all-in on an expensive and unproven ad medium on a hunch. Subway ads are a prime example – “I take the subway and I notice the ads, therefore everyone must!”. Unfortunately, subway ads are priced accordingly – the ad industry is a notoriously efficient market and it takes a keen eye to spot below-market opportunities to buy ad space. This is why Search and Social remain the most appealing options in a cluttered market – they offer relatively predictable pricing for a controlled reach that is scalable and well monitored.