How Spotify Advertising Works: Formats, Targeting, and What Audio Does That Display Can't

Streaming audio is not a niche add-on to a media plan anymore, it is a channel with its own inventory, its own targeting logic, and its own buying paths, and an agency that treats it like radio with a digital wrapper will misquote it and undersell what it actually does. Edison Research's 2026 Infinite Dial study found that 81 percent of Americans age 12 and up listened to online audio in the past month, roughly 233 million people, and that is the scale an agency is buying into the first time a client asks about audio.

Spotify is the largest single destination for that inventory, and its ad platform breaks into four formats rather than one generic audio unit. Per Spotify's own ad formats overview, audio ads are built to reach listeners "throughout their day, even in screenless moments when other platforms can't," which is the format's actual pitch: it runs while a phone is in a pocket or a car is on the highway, contexts where a video or display impression simply cannot land. Video ads add visual storytelling for listeners actively engaged in the app. Display ads run alongside whatever is playing, reinforcing the audio rather than competing with it. Carousel ads put multiple cards in the Now Playing view, each linking to its own destination.

Podcast ads run on a different mechanism than music ads

Spotify's podcast ad format page draws a distinction agencies need to understand before quoting: host-read ads are sold as custom sponsorships and require going through a Spotify sales representative, while dynamically inserted ads are the self-serve option, bought through Ads Manager with Spotify's free ad-creation tooling. That difference matters for a quote. A host-read placement is a negotiated, higher-touch buy with less programmatic flexibility. A dynamically inserted spot behaves like a normal digital placement: targetable, swappable, and reportable inside a standard campaign.

The format list runs deeper than the four headline types suggest. Spotify's own ad specs pages break audio ads into a standard spot plus an optional Canvas element, a short looping visual that displays on screen while the audio plays, and separates video into in-stream and in-feed placements, display into in-feed and leaderboard units, and adds a sponsorship tier covering playlist and homepage takeovers on top of the four self-serve formats. That level of granularity is worth knowing before a quote goes out, because a client picturing one generic "audio ad" is actually choosing among specs with different production requirements and different price points, and a takeover sponsorship in particular is a different conversation, and a different budget, than a standard self-serve buy.

Targeting runs on listening behavior, not just demographics

The targeting stack, per Spotify's audience targeting page, has four layers: first-party demographic data, behavioral targeting based on listening habits and genre preference, contextual targeting that defines the environment an ad serves into (a workout playlist, a commute-hour podcast), and custom audiences built from a client's existing customer lists or lookalikes modeled off them. The behavioral and contextual layers are the ones display cannot replicate. A display network can guess intent from browsing history. Spotify knows, directly, that a listener has a workout playlist queued at 6 a.m., a targeting signal display advertising has no access to.

Three ways to actually buy the inventory

• Ads Manager or Direct IO: self-serve, campaign objectives set around revenue, awareness, traffic, or promotion, with bid cap, pacing, and frequency controlled directly, per Spotify's get-started guide
• Spotify Ad Exchange (SAX): programmatic buying through a client's existing DSP, connecting to more than 40 demand-side platforms including The Trade Desk, Google Display & Video 360, and Amazon DSP, according to Spotify's own SAX overview
• Managed service through a Spotify sales representative: the path for host-read sponsorships and larger custom buys that do not fit a self-serve flow

The SAX path is the one agencies with an existing programmatic buy tend to underuse. If a client is already running display and OTT through a demand-side platform, routing Spotify inventory through the same exchange means one unified reporting view and frequency management across formats, instead of a separate silo for audio that never gets reconciled against the rest of the media plan.

The identity layer matters here too. SAX campaigns can run against Ramp ID, EUID, ID5, Connect ID, and PAIR, the cookieless identity solutions the wider programmatic industry has been consolidating around, so an agency does not have to solve identity resolution separately for Spotify while the rest of a client's programmatic stack has already moved past third-party cookies. For an agency already fluent in one of those identity graphs from running display or OTT, that is one less new system to learn before the audio line item can go live.

What audio does that a display line item cannot

The real pitch for audio in a media plan is not that it outperforms display on click-through, it usually does not have a comparable click surface at all. The pitch is reach into moments display structurally cannot occupy: driving, exercising, doing chores, any context where a screen is not in view but attention still is. Spotify's measurement tools include Brand Lift studies tracking ad recall, awareness, and purchase intent specifically because audio's value shows up in those upper-funnel metrics before it shows up in a last-click conversion report, and a client expecting audio to behave like a search or display retargeting line will be disappointed by a metric the format was never built to win.

Creative quality carries more of the outcome in audio than a media buyer used to display might expect. Spotify's Creative Lab attributes roughly half of a campaign's return to the quality of the creative itself rather than the buy mechanics around it, a sharper ratio than most display or paid social channels report, and a real argument for treating audio script and voice casting as a strategic decision instead of an afterthought bolted onto a media plan built for other channels.

Ad specs and delivery requirements agencies get wrong the first time

The format and targeting layers above are the strategic part of a Spotify quote. The technical delivery requirements are the part that stalls a launch when nobody checks them until the week the campaign is supposed to go live. Spotify's ad specs pages list separate requirements for each unit: audio spots have fixed duration options rather than a free-form length, video ads carry their own aspect ratio and file-size ceilings for in-stream versus in-feed placement, and the optional Canvas element attached to an audio spot has its own loop-length and resolution rules distinct from the video format's specs. Treating all of that as one generic "send us an audio file" request is how a launch date slips by a week while a client's creative team re-exports assets to match specs nobody sent them up front.

The practical fix is boring but effective: pull the current spec sheet before the creative brief goes out, not after the first file bounces back from ad review. A script written to a fixed spot length records cleaner than one trimmed after the fact, and a Canvas loop built to the actual resolution requirement does not need a second production pass. For an agency running its first few campaigns, building a short internal reference doc off the official specs page, one page, updated whenever Spotify revises a format, saves more launch-week friction than any targeting decision covered above.

A worked example: sequencing formats for a local services client

Take a plumbing or HVAC client running a full media plan already, search, local display, maybe some connected TV. Audio's role in that plan is not to replace any of it, it is to reach the exact moments those channels cannot: someone driving between job sites, a homeowner doing yard work while a pipe quietly fails inside. A first Spotify quote for a client like this in the home services space typically starts narrow rather than broad: contextual targeting against local commute-hour podcasts and drive-time playlists, paired with geographic targeting tight enough to match the client's actual service radius, not the metro area as a whole.

The second phase, once the contextual buy has a few weeks of delivery data behind it, is where behavioral and custom-audience targeting earn their place: layering in listeners whose genre and podcast habits match the client's existing customer base, and building a custom audience off the client's own service-call list if the CRM data is clean enough to upload. That sequencing, contextual first, behavioral and custom-audience second, keeps the first invoice defensible on its own terms instead of asking a client to trust a fully built-out targeting stack before the format has proven it can deliver reach in their market at all. Conduit's pricing for white label programs is structured around that same phased logic, so an agency is not quoting a client full complexity, and full cost, before the channel has shown it works for that specific account.

What to confirm before a first Spotify quote goes out

• Which objective the budget actually supports: awareness reach at a contextual level, or a custom-audience buy that needs a clean customer list uploaded first
• Whether the client's site can carry the Spotify Pixel or Conversions API, since a campaign sold on measurement that never gets installed produces a report with nothing in it
• Whether the client already runs programmatic spend through an existing DSP, in which case routing through Spotify Ad Exchange keeps reporting unified instead of adding a disconnected fourth dashboard
• Whether usable creative exists to repurpose from video or display, or whether script and voice casting need to be budgeted as real production time rather than an afterthought

For an agency quoting streaming audio for a client the first time, the format and targeting mechanics above are the easy part to learn. The harder part is building creative that works in an audio-only context and setting the right expectation for what the channel measures well versus what it does not. Conduit's white label Spotify advertising program runs the buying, creative, and reporting for agencies that want to offer the channel without building that operational knowledge in-house first.