CAC payback by channel: what each source is worth, and how fast it pays back

A source that brings a lot of leads doesn't necessarily bring a lot of revenue. Funnelsight shows, for each source, the recurring revenue its customers generate and how many months that revenue takes to cover what the source cost, so budget discussions start from what a channel is worth rather than how loud it is.

How to calculate CAC payback period by channel

Revenue comes from the deals recorded in your CRM, whether monthly plans, multi-year contracts or renewals. The cost is what the source took to run: media, commissions, content production or the tools a channel runs on, not salaries or overhead. That keeps sources comparable with each other, and explains why the figures are much shorter than the company-wide acquisition payback benchmarks, which count the whole sales and marketing budget. The formula stays visible and adjustable, for example to apply gross margin, so finance can check every figure.

How to calculate marketing ROI by channel

Marketing ROI by channel tells you how much each channel brings back for every dollar you put in, so you can compare channels on the same scale before moving budget.

Say a channel cost $100,000 over a year, and the customers it brought signed $300,000 of yearly contracts. ROI = ($300,000 − $100,000) ÷ $100,000 = 200%. Put simply, each dollar spent came back as three: the one you spent, plus two.

Is a 200% ROI good?

It sounds huge, but channel ROIs often look this high. The calculation usually counts only what the channel itself costs, such as media or commissions, not the people who run it or the sales team who close the deals. It counts revenue, not profit, and part of every dollar goes to hosting, support and payment fees. And it stops at the first year. In the illustrative company on this page, channels range from 100% to nearly 600%. Counting every salary and tool in sales and marketing, the same company gets about $1.40 of new yearly revenue for each dollar spent: an ROI of about 40%.

So use channel ROI to compare your channels with each other, calculated the same way, rather than against a figure from another company. A channel below 0% doesn't cover its cost in the first year, which can still be fine if its customers stay for several years.

Questions it answers

Which sources pay back, and how fast?

Each source shows what it cost next to the new annual recurring revenue its customers bring, credited according to the attribution model your team chose.

Neither payback nor ROI shows how each cost behaves. Content costs what you decide to produce. Paid search prices move with competition and the season, so the same budget buys more or fewer clicks from one quarter to the next. A commission-based partner program costs nothing until a deal closes. Two sources with the same payback can carry very different risks.

Is a slow source a bad one?

Not necessarily. A multi-year contract signed with sales can take longer to cover its cost than a monthly plan, then keep paying for years. Whether 6 months is acceptable depends on how long those customers stay, which the same view shows by source.

The attribution model can also make a source look slower than it is. A LinkedIn ad can reach buyers before they search, and the deal then comes in through search or Direct. Under last touch, the ad gets little or no credit for that deal. Comparing last touch with first touch or U-shaped shows whether a channel opens paths that others close.

Timing matters too. Organic search looks inexpensive over 12 months partly because it also earns from content published in earlier years, while paid search brings customers the day its budget is switched on and stops when it's switched off. One is cheaper per customer, the other can be scaled at will. Funnelsight reads each source on its own timeline, so a channel isn't cut for being slower to show results. See how one team used this reading to cut its budget without emptying its pipeline.

Why doesn't the attributed total match the CRM?

The CRM counts every deal closed in the period. Attribution can only credit a deal that has at least one touch your rules count.

What falls outside isn't an error, and its size depends on your rules. Some deals have no marketing touch at all, such as deals opened by outbound prospecting. Others have touches the rules leave out: older than the window, or recorded on someone the rules don't tie to the deal. A team that reads touches at account level credits activity from anyone at the account; one that counts only the contacts named on the opportunity credits less, often by choice, to keep opportunity creation light for sales.

A mistagged campaign is a different problem: it doesn't shrink the total, it credits the wrong source, often Direct. The same happens when someone discovers you in an AI assistant's answer: nothing links the answer to the visit that follows, which is credited to branded search or Direct. Funnelsight shows both totals side by side and explains what sits in the gap, with the CRM as the reference for revenue. Why marketing and sales pipeline numbers don't match goes further into the two readings.

What is an attribution model, and which one should you use?

An attribution model is the rule that decides which marketing touches get credit for a deal. Before asking for a demo, a buyer may see an ad, read an article, join a webinar and search for your name. The model decides how the deal's revenue is split between them.

Single-touch models give all the credit to one touch: the first one (first touch) or the last one before the buyer converts (last touch). Multi-touch attribution shares the credit: equally across every touch (linear), or with more weight on the first and the last (U-shaped, also called position-based).

No model is the true one. Each answers a different question. First touch shows which channels find new buyers. Last touch shows which ones turn interest into a request. For a B2B sale with several touches over weeks or months, a multi-touch model such as U-shaped is a good place to start: it credits the channel that opened the path and the one that closed it, without ignoring the rest.

Then compare two models. If a channel ranks near the top under one and near the bottom under the other, decide what you expect from it before moving its budget. And keep the same model when you compare periods, or a change of model will look like a change in results.

How it works

Your team picks the attribution model and can adjust its weights, for example from 40-20-40 to 20-60-20, with two models side by side. Your team also sets what counts as a first touch and how far back touches are counted, often the length of your sales cycle.

Terms such as "sourced" or "influenced" vary from one company to another, so Funnelsight doesn't impose a definition: each reading your team names shows its rule. When a rule changes, the change is logged and past periods are recalculated or the break is marked, the principle behind a single source of truth.

What it does, and what it doesn't