The Podcast ROI Framework: One Number Your CFO Will Accept

Why podcast ROI arguments fail in the budget review

Picture this meeting. You have twelve episodes published, the download chart is up and to the right, and you lead with it. Then someone in finance asks what that number is actually worth. The honest answer is that you don’t know yet. So you reach for engagement rate, then a listener testimonial, and five minutes later you’re not defending a metric anymore. You’re defending whether the show should exist at all.

The number was not the problem. But using it to answer the question was.

Downloads answer “how many people listened.” The person across the table asked “what impact does the podcast have on the company.” Those are different questions, and a bigger download chart never closes the gap between them.

There is a second failure I typically see in these types of conversations. It’s the attempt to prove the podcast caused pipeline without data to prove it. Marketing builds a multi-touch model, assigns the podcast a slice of a closed deal, and presents a revenue figure. Finance pushes back on the assumption that the podcast plays a part, and your entire justification unravels if you can’t prove it. Now you have spent your credibility as well as your budget.

The framework below is going to help you prepare for these types of conversations. It’ll answer:

• What did it cost us to earn real attention from the accounts we are trying to sell to?
• How does that compare to every other way we could have bought that attention?

That is a claim you can confidently defend and prove to leadership.

The four inputs the model needs

You need four numbers. Luckily, three of them you probably already have.

1. Downloads

What is it: The total downloads for your reporting period, across every episode.

Where to find it: Your hosting analytics, using the same reporting period you’ll apply to every other input.

Downloads are the denominator, but not at all the whole story. They’re the size of the pool before you filter for quality, and on their own, they’re the exact number that got you into trouble back in that meeting.

Pull your unique listener count too, and use it whenever you’re reporting reach. Downloads count any play of 60 seconds or more as well as auto-downloads. Whereas unique listeners is tracking individual devices that have listened to an episode, and only counting them once. That’s much more indicative of true podcast reach (and yes, it’ll be a smaller number than downloads). Company identification runs on download data, though, which is why downloads are the input feeding the account count below.

The common mistake: Mixing time windows. If your download number covers every episode you’ve ever published but your cost number covers one quarter, the ratio you build is broken and just inaccurate. So pick your time period, apply it to every input, and label it. Most teams land on the quarter, because it aligns with other marketing reports and makes it so one big episode likely can’t swing the whole picture.

2. Companies identified

What is it: The slice of your audience you can actually identify. For a B2B show, that usually means company-level data resolved from your download numbers, so you’re looking at organizations rather than phones that quietly downloaded a file overnight and never pressed play.

Where to find it: In CoHost’s B2B Analytics dashboard, which turns your download data into named companies.

This is the input that turns podcasting into attributable sales metrics. It’s also the one most shows don’t have, which is why most shows can’t run this model. If you want to see how teams use those firmographic signals week to week, our guide to measuring podcast performance for B2B brands covers it.

What you get back at this stage:

• Which companies your listeners work for
• Listener job roles and seniority
• The industries tuning in

The common mistake: Reporting identified companies as one number instead of splitting the net-new from the familiar. A company the show surfaced is pipeline it sourced. A company already in your CRM is pipeline it’s influencing. Both are real, so tag each against your CRM and report them on separate lines instead of claiming them all as new.

3. Target accounts reached

What is it: The count of distinct identified companies that match your ICP. This is your denominator, and everything before it exists to produce this one figure.

Where to find it: The same B2B Analytics view, filtered against your target account list. If you sync to CRM through the Salesforce integration, the match happens against the account list sales is already working.

You’re counting companies here, not people. That’s on purpose, and it’s the thing that makes this ratio hold up: identification tells you which organizations are listening, so an organization is the honest unit to divide by.

This one step is what separates a number that survives scrutiny from one that doesn’t. It’s also the step most reports skip, because it makes your audience look smaller. But smaller, qualified, and true beats bigger and indefensible every time.

The common mistake: Defining your ICP parameters after you look at the data, not before. It’s very easy to widen what counts as a target account until the number improves, and trust me, everyone in the room can tell when it’s happened. So borrow the account list sales is already working, or the firmographic profile your demand gen team already targets. Then use that same one every quarter.

4. Fully loaded cost

What is it: The entire spend that goes into the show, including the costs teams love to forget:

• Production and editing
• Hosting and analytics
• Promotion and paid distribution
• Guest and/or host appearances
• The design and social work that ships with every episode

Where to find it: Your own ledger, plus an honest estimate of internal hours from whoever books the time.

The common mistake: Understating it. That may be tempting, we’ve all been there. But we also know it always backfires. The moment finance spots a line you left out, every other number in your model becomes suspect.

There’s an upside to being this thorough that you won’t see coming until you try it. A fully loaded cost number is one your finance counterpart can reconcile against their own books. And when the two line up, you stop arguing about the math and start talking about the channel. Which is the conversation you actually wanted.

Assembling the number

Okay, now the fun part of putting it all together. One ratio does the work here:

Cost per target account reached is your fully loaded cost divided by the number of distinct ICP accounts that listened. It’s priced in the same unit your demand gen budget is priced in: reaching a specific buying organization.

Report your average consumption rate beside it, not inside it. Consumption is measured per episode, across everyone who pressed play, so it tells you whether the content held up. Bring it to the meeting as a quality signal sitting next to the cost figure, and let the two numbers say different things.

Here’s the model with some illustrative inputs. These are made-up figures for examples, not benchmarks.

Now for the step that actually makes the number mean something. On its own, five hundred dollars is neither good nor bad. It becomes an argument the moment you set it beside what reaching that same account costs you everywhere else.

And you already have those figures. What does a qualified meeting from paid search cost you? What did your last field event cost per target-account conversation? What’s your blended cost to generate one qualified lead?

The goal is to get you an answer to the question, “what impact does the podcast have on our brand?”

Don’t reach for an industry benchmark here. Your own channel costs are the only comparison finance will (and should) accept. If the podcast is reaching decision-makers at named accounts for less than your other channels reach those same accounts, you’ve made your case. And you did it without ever claiming the podcast closed a deal.

What to do when the number looks bad

As a marketer, you know that sometimes the numbers don’t look good. Especially depending on the stage of your brand’s podcast. But rest assured, let’s cover what to do when this happens.

A high cost per target account reached usually has one of three causes, and they call for completely different responses. Work out which one you’ve got before the meeting.

Plenty of downloads, few target accounts

Your audience is real, and it’s the wrong audience. That’s a content and promotion problem, not a budget one.

The fix here is editorial before it’s financial. Look at which episodes pulled the accounts you wanted, which guests, which distribution channels. If a quarter of your downloads come from a segment you’ll never sell to, you either change the content or you accept that the show is doing a different job than the one you’re pricing.

Strong ICP match, small totals

The show is working and not enough people know it exists. This is a distribution problem, and it’s the cheapest of the three to fix.

Your content has already proved it reaches the right companies. What it hasn’t had is audience growth. That argues for promotion spend rather than an editorial rebuild.

Cost is the outlier

Reach and match are steady but the ratio moved, so the question is simply which line moved. A fully loaded cost breakdown answers that in about a minute: a one-off production spike, a paid push, a season of travel.

Naming the line yourself changes the meeting. A marketer who says “this quarter came in at five hundred dollars, here’s the line that moved, and here’s what I’m changing” is running a channel.

The two objections finance will raise

To be safe, expect both of the objections we cover below. The trick is to answer each one and have defensible terms and data to use if you’re pushed.

”You cannot attribute that”

Correct. And this model doesn’t try to. It measures reach into named accounts at a known cost, full stop. Attribution is a separate, weaker claim, and conceding that up front is exactly what buys you the rest of the conversation.

The difference matters in practice. You’re not saying the podcast produced the deal. You’re saying the buying committee at that account listened, you know what that cost, and it cost less than the alternatives. If the conversation heads into attribution mechanics anyway, our guide to podcast attribution covers what is and isn’t knowable.

There’s also a version of this objection that’s really about trust, not method. When someone says you can’t attribute it, sometimes what they mean is they suspect the number has been arranged to flatter the channel. Naming the limits of the model early, before anyone pushes, is what defuses that. And it costs you nothing, because the model was never making the claim they’re worried about in the first place.

”What happens if we stop?”

The fairer question of the two. If you hadn’t spent the money here, what would that same attention have cost you somewhere else?

That’s exactly what the channel comparison answers. Name the specific thing you’d have funded instead, and price it. The answer isn’t “nothing happens.” It’s “we’d have spent this on something else,” and that something else has a cost per reached account you can go look up.

Be specific about the alternative. “We could’ve run more ads” invites a debate you can’t win. “This quarter’s podcast budget is roughly one regional field event, and the last one got us ninety target-account conversations” gives the room something concrete to compare. Better still, it puts the weight of the comparison on numbers you both already trust.

And worth saying plainly: the podcast won’t win every comparison, and it shouldn’t be expected to. Different channels do different jobs at different stages, and a model that always concludes in favor of the thing being measured isn’t a model. It’s a sales pitch. If the podcast reaches fewer target accounts per dollar than your best channel but reaches accounts that channel can’t touch, say exactly that. It’s a stronger argument, and it happens to be true more often than not.

A reporting cadence that holds up

The model is quarterly. The inputs are monthly. Keeping those two separate is most of the discipline.

Monthly, report inputs only. Downloads, companies identified, target accounts reached, and cost. No ratio, no ROI claim. A single strong episode can distort a month badly, and if you publish a monthly ROI figure you’ll spend the next meeting explaining variance instead of talking about the channel.

Quarterly, run the model. The ratio, your consumption rate beside it, and the comparison against your other channels for the same quarter. This is the cadence that matches how an audience actually accumulates, and it matches how budget conversations get scheduled anyway.

Annually, show the trend. Cost per target account reached over four quarters says more than any single figure ever will. If it’s falling while reach grows, that’s the strongest version of the argument, and it’s one no download chart can make.

One practical note on your first quarter. You won’t have a trend, so don’t pretend to. Report the inputs, say plainly that the ratio needs another quarter before it means anything, and set that expectation early rather than being asked for it later. For benchmark-setting in that first period, setting benchmarks to measure podcast performance is a better starting point than this model.

What actually goes on the slide is shorter than people expect. One line for cost per target account reached, one line for the comparison channel, one line for the direction of travel since last quarter, and one line naming what you’re changing. Everything else in this article is the working behind those four lines, and it belongs in the appendix nobody opens unless they want to check you. Keeping the appendix ready and the slide short is the whole trick.

The reporting surface itself matters less than the consistency. Whether your inputs come out of podcast analytics dashboards, a spreadsheet, or a quarterly export, what earns trust is that the same four inputs arrive the same way every period, defined the same way, with the same ICP filter applied.

Where the framework breaks

Say this part out loud in the meeting. It’s what makes the rest credible.

It doesn’t prove causation. It never will. If your leadership needs proof that the podcast caused revenue, this model won’t deliver it, and neither will any other podcast measurement approach being sold to you.

It needs company-level data to run at all. Without identification there’s no denominator, and no substitute is worth reaching for. Reach and cost on their own are an efficiency note, not a strategic argument.

It counts accounts, not depth. One download from a target account and forty from that same account both register as one account reached. That keeps the number honest, but it does mean the ratio rewards breadth, so read your consumption rate alongside it.

It punishes small samples. A new show, or one publishing irregularly, will produce a ratio that swings on a single episode. Build the baseline first.

It assumes your audience is your buying audience. Plenty of good branded podcasts exist to serve customers, recruit talent, or support a community. Those shows are worth making. This just isn’t the model for them, and forcing it on them produces a number that makes a good show look like a failure.

For the broader picture of how ROI thinking fits alongside the other things a podcast returns, the complete breakdown of podcast ROI covers the wider set of outcomes. This piece is deliberately narrower: one number, built carefully, that survives contact with finance.