Build a Bottom-Up Revenue Forecast: Existing Customers First, Paid Last

How did you land on next year's revenue number? Be honest. If the answer is "we did roughly 40% last year so we put 40% on top again", you don't have a forecast. You have a wish with a percentage stuck to it, and you're about to hand your media buyer a budget built on hope.

I get why it happens. Top-down is fast. You pick a growth rate that feels ambitious-but-not-insane, multiply, and move on. The problem is that a top-down number tells you nothing about whether it's actually achievable, and it quietly assumes paid acquisition will paper over any gap. That's the most expensive and least reliable part of your business doing the heaviest lifting in your plan.

Here's my take: build the forecast the other way up. Start with the revenue you're most sure of and add paid last, as the layer that fills the gap rather than the layer that carries the load. By the time you're done, your revenue goal has turned itself into a month-by-month ad budget you can actually defend.

Think of it as a reverse P&L. Most planning starts with a revenue target and works down to costs. We're going to start with the most predictable revenue and build up to the target, then read off what paid has to deliver. Four layers, in order of how confident you can be in each.

Layer 1: revenue from customers you already have

Your most predictable revenue next year comes from people who've already bought from you. They're the foundation, and almost nobody starts here.

You need cohort LTV data for this, value grouped by the month a customer was first acquired. If you don't have it, this is the single most worthwhile thing to pull before you plan anything. Klaviyo and Shopify between them will get most brands close.

Here's the logic. Every cohort you acquire keeps spending in the months after. Say a cohort that first buys in January does ~$100k in that first month, then roughly $28k more across February to December as some of them come back. The cohort you acquire in February does the same shape one month later. Stack twelve months of cohorts and the repeat revenue overlaps into a base layer that's remarkably stable, because it's not subject to CPMs, auction prices, or whether your next ad pops.

For a brand doing, say, ~$4m this year with a decent repeat rate, that existing-customer base might carry ~$1.3m to ~$1.6m of next year's revenue before you spend a single dollar acquiring anyone new. That's your foundation. Pencil it in with confidence, because it's the number you're least likely to be wrong about.

Layer 2: revenue from audiences you own

The next layer up is what I'd call owned audiences: the people you can reach for free, as many times as you like, without paying the ad auction for the privilege.

That's your email list of non-customers, your SMS subscribers who haven't bought yet, your organic social following. You don't control CPMs here, but you do control how often you show up and what you say.

The maths is refreshingly simple. Take your list of, say, ~40k engaged email subscribers who haven't purchased. Lay out your campaign and flow strategy for the year, apply a sober conversion assumption and your AOV, and you get a revenue line. Do the same for SMS. Do the same for organic social, where you're forecasting off reach and a much lower conversion rate, but it's still real.

Be conservative here. This layer is more volatile than your existing customers, because list fatigue, deliverability, and your own consistency all wobble it. I'd rather under-call this and be pleasantly surprised than build a budget on heroic email assumptions.

Layer 3: revenue from organic search you already rank for

This one gets missed constantly, so I'm giving it its own layer: organic search is an owned audience too, as long as you already hold the rankings.

I don't mean your branded terms, those are really just existing demand finding you. I mean the category terms you've earned a real position on. Here's the way to size it.

Take a category keyword you rank well for. Say it gets ~10k searches a month and your position earns you roughly 15% of the clicks. That's ~1,500 visits a month. Apply your site's conversion rate on that traffic, say ~2%, and your AOV, and you've got a forecastable revenue line from a channel you're not paying per-click for. Stack your handful of genuinely ranking terms and watch the search trend on them, because volume moves seasonally, and you've got your third layer.

It won't be your biggest layer for most brands. But it's predictable, it's free at the margin, and leaving it out means you'll over-ask of paid to make up the difference.

Layer 4: paid acquisition, the gap-filler (not the hero)

Now, and only now, do we get to paid. Add up your first three layers and compare the total to your revenue goal. The shortfall is what paid acquisition has to deliver. That's the job paid is actually for: filling the gap between predictable revenue and ambition, not carrying the whole plan.

This is also the most volatile layer, so it's the one you build last and trust least. CPMs move, conversion rates move, your offer might land differently than you expect. None of that is fully in your control.

But sizing it is straightforward once you've framed it as a gap. Say your first three layers come to ~$2.6m and your goal is ~$5.5m. Paid has to bring in the missing ~$2.9m of revenue, and crucially, most of that should be new customers, because the repeat value those new customers throw off is what becomes Layer 1 in next year's forecast.

Turn it into spend with your CAC. If your new-customer revenue needs to be ~$2.9m, divide by your AOV to get your customer count, then multiply by what it actually costs you to acquire one. At a ~$60 AOV that's roughly 48,000 new orders, and at a ~$32 blended CAC that's about ~$1.55m of ad spend across the year. There's your annual paid budget, derived rather than guessed.

Then split it across the months. Don't divide by twelve. Weight it to your seasonality, your launch calendar, your BFCM, and the months where the first three layers are softest and paid has to work hardest. That month-by-month line is the thing you actually hand your media buyer, and it's defensible all the way down because every number underneath it was built, not hoped.

The point of a forecast is to find where you're wrong

Here's the bit that takes the pressure off the whole exercise. You will not get this exactly right. Nobody does. Anyone running an inventory business will tell you half the job is managing the places you were wrong and had too much, and the other half is the places you were wrong and had too little.

So don't build this to be right. Build it to be directionally accurate, and to show you where you're importantly wrong the moment reality starts coming in. A forecast is a trellis you grow the real numbers against. When actual revenue lands, you compare it layer by layer and you can see exactly where the miss is. Existing-customer revenue soft? That's a retention problem. Paid revenue short at the same CAC? Your creative or offer is underperforming, and you know it in week three instead of month six.

That diagnostic power only exists because you built it bottom-up. A top-down number can be wrong by a mile and never tell you why.

Build your scenarios before you commit

One more move before you lock anything: don't run a single forecast, run a few.

Pick the two or three inputs most likely to make you wrong. Usually it's CAC, your repeat rate, and conversion. Build a base case, a soft case where CAC drifts up ~20% and conversion eases off, and a strong case. The gap between those scenarios is your real planning insight, because it shows you how much of your goal depends on things you don't fully control.

And it tells you something about your cash. If your plan only works in the strong case, you're not planning, you're gambling. If it holds up in the soft case, you can commit inventory and spend with a much steadier hand.

Where to start this week

If you do one thing, pull your cohort LTV data and build Layer 1. Most founders have never seen their existing-customer base laid out as a forward revenue line, and it reframes the entire plan, because suddenly paid isn't the whole story, it's the top slice.

From there, add the layers up the stack and let the gap tell you your budget. The number you end with will be a lot less comfortable than a tidy growth percentage, and a lot more honest.

This is the exact build we run with clients when we put a forecast together and then back it with the media plan that delivers the paid layer. If you want to pressure-test your own version, reply and tell me which layer you're least sure of, I'm always happy to talk through where a forecast tends to break.