Media buying is the process of purchasing ad space and time across digital and offline channels to put a brand in front of the right audience, at the right moment, for the lowest possible cost. A media buyer does the work: they negotiate or bid for inventory, set and manage budgets, launch campaigns, and optimize them against performance goals like cost per acquisition (CPA) or return on ad spend (ROAS). Media buying covers everything from a TV spot or a billboard to a Meta ad, a TikTok placement, or a real-time programmatic bid on the open web. In 2026, most of it happens through software, and increasingly through AI that bids, allocates budget, and even builds the creative on its own.

Here's why it matters now. The buy is where money meets the market. You can have a brilliant strategy and great creative, but if the placement, timing, and price are wrong, the campaign loses money. Global ad spend is projected to pass $1.2 trillion in 2026, with roughly 76% of it digital. Almost all of that digital money is bought programmatically, through automated auctions rather than handshake deals. And the job of the person at the controls is changing fast, because the platforms now automate most of the levers a media buyer used to pull by hand.

This guide covers what media buying actually is, what a media buyer does day to day, the four ways media gets bought today (manual, programmatic, automated, and agentic), the major channels, the skills and metrics that matter, and how AI agents are rewriting the role. If you're deciding whether to hire in-house, use an agency, or hand the buy to software, the last section lays out the honest trade-offs.

What does "media buying" actually mean?

At its simplest, media buying is purchasing the space where your ads appear. HubSpot defines it as "the process of purchasing ad space and time on digital and offline platforms, such as websites, YouTube, radio, and TV" (HubSpot). The goal is efficiency: get maximum relevant exposure for minimum spend.

Two things get blurred constantly, so separate them. Media planning is the strategy. Who the audience is, what the campaign should achieve, how the budget splits across channels, which placements to chase. Media buying is the execution: actually acquiring those placements and running them. As HubSpot's framing puts it, planners "outline campaign goals focusing on overall strategy, while media buyers carry out those goals through deal negotiations and budget management." In a big agency these are separate roles. In a startup or an SMB, one person does both, often before lunch.

The buy isn't a one-time transaction. A good media buyer treats it as a loop. You launch, watch the numbers, kill what's losing, pour budget into what's winning, refresh the creative when it tires, then do it again. That feedback loop is the whole game. We've watched DTC brands win not because they had the best initial idea but because they iterated faster than the competition, testing 15 creatives a week instead of two and reallocating spend within hours instead of at the end of the month.

What does a media buyer do?

A media buyer is the operator who turns a media plan into live campaigns and then makes them perform. The core responsibilities, drawn from how the role is described across the industry (HubSpot, Lifesight):

  • Sourcing and negotiating inventory. In direct deals, that means talking to publishers about rates and placements. In programmatic, it means setting bid strategies and targeting in a demand-side platform.
  • Managing budgets. Deciding how much goes where, pacing spend so it doesn't burn out mid-flight, and reallocating as results come in.
  • Launching and trafficking campaigns. Setting up campaign structure, audiences, creative, and bidding inside ad platforms, then pushing it live.
  • Optimizing. Watching performance against targets and adjusting: pausing weak ad sets, scaling winners, swapping creative, refining audiences.
  • Reporting. Tying spend back to outcomes so the business knows what each dollar returned.

The day-to-day mix depends on the channel. A buyer running connected-TV and out-of-home spends more time on negotiation and insertion orders. A performance buyer on Meta and TikTok spends almost all their time inside ad managers, reading dashboards and making micro-decisions about budget and creative. Rex Gelb, HubSpot's senior director of paid advertising, captures the judgment the role demands: "Some ad placements might be good for one set of goals, but bad for another" (HubSpot). Picking the right placement for the right objective is exactly the kind of call that's hard to hand off to a machine.

Is media buying the same as performance marketing?

Not quite. Performance marketing is the broader discipline of driving measurable outcomes from paid channels (installs, leads, sales), and it covers strategy, creative, measurement, and attribution. Media buying is the purchasing-and-managing layer inside that. Every performance marketer does media buying; not every media buyer owns the full performance stack. At smaller companies the titles collapse into one person who does it all.

How does the media buying process work? A 5-step framework

The mechanics vary by channel, but the underlying workflow is consistent. Here's the framework we use when we set up a new account.

Step 1. Inherit the plan and the targets. Before you buy anything, you need the media plan: who the audience is, what the campaign objective is (awareness, traffic, conversions), the budget, the timeline, and the success metric. Lifesight lists this planning phase as the first thing that has to happen before purchasing begins (Lifesight). Buying without a clear target CPA or ROAS isn't buying. It's gambling.

Step 2. Select channels and inventory. Match the audience to where they actually are. A B2B SaaS buyer leans on Google Search and LinkedIn; a DTC supplement brand lives on Meta and TikTok. In direct buying, you'd list outlets and send RFPs to vendors (HubSpot). In programmatic, you configure a demand-side platform to bid on the right inventory.

Step 3. Set up and launch. Build the campaign structure, define audiences, load creative, choose a bidding strategy, and go live. Lifesight describes this as entering specifications ("campaign type, creative materials, budget, target audiences, bidding strategies") into the platform and launching (Lifesight). Structure decisions here, like whether to use campaign-budget optimization or ad-set budgets, shape everything downstream. (We dig into that trade-off in our guide on CBO vs ABO on Meta ads.)

Step 4. Monitor and optimize. Once spend is live, the buyer collects data, reads it against the target, and adjusts. Pause the ad sets that aren't converting, scale the ones that are, and refresh creative before it fatigues. Creative fatigue is the silent killer of most accounts. When frequency climbs and the same audience sees the same ad too often, performance decays even if nothing else changed (see what is creative fatigue).

Step 5. Report and reinvest. Close the loop. Tie spend to outcomes, document what worked, and feed the learnings back into the next round of buys. The buyers who compound results are the ones who turn every campaign into evidence for the next one.

Manual vs programmatic vs automated vs agentic media buying

Most guides flatten this into "direct vs programmatic." In 2026 there are really four distinct ways to buy media, and they sit on a spectrum from fully human to fully autonomous.

Approach Who makes the decisions How inventory is bought Best for Trade-off
Manual / direct Human, deal by deal Negotiated directly with publishers; insertion orders Premium placements, niche or local media, guaranteed inventory, sponsorships Slow, relationship-dependent, doesn't scale
Programmatic Human sets rules; software bids Automated real-time auctions via DSPs Reach and precision at scale across the open web Black-box-ish; quality and fraud need watching
Automated (platform autopilot) Platform algorithm, within one ecosystem Meta Advantage+, Google Performance Max (you give budget + objective) Hands-off scaling on a single platform You lose granular control; it's a single-platform black box
Agentic An AI agent reasons and acts toward a goal Cross-platform, via APIs, with minimal human input End-to-end execution: research, build, launch, iterate Early; only as good as your data and tracking

Manual / direct buying is the original form: a human negotiates ad space directly with a publisher, agrees on a rate, and signs an insertion order. HubSpot describes the steps as listing outlets, submitting RFPs, making decisions, sending insertion orders, delivering ads, monitoring, and negotiating "makegoods" if delivery falls short (HubSpot). It's best when you want a specific premium placement or you're building trust with a niche or local audience. It does not scale.

Programmatic media buying automates the buying and selling of inventory through software and real-time auctions. Lifesight describes it as automating purchases "through software using real-time bidding, private marketplaces, and algorithms for targeting" (Lifesight). This is now the dominant mode. Programmatic accounts for nearly 90% of digital display spend, and real-time bidding holds about 55% of programmatic buying. When people say "media buying" today, they usually mean programmatic.

Share of digital display spend bought programmatically
Programmatic     ██████████████████  ~90%
Everything else  ██░░░░░░░░░░░░░░░░  ~10%

Source: affinco. Bars scaled to 100%.

Automated platform buying is what Meta and Google sell as their flagship products. Meta Advantage+ automates targeting, placements, creative testing, and bidding once you supply a budget and an objective; Google Performance Max unifies Search, Display, YouTube, Gmail, and Maps inventory behind asset groups and conversion signals (TensorOps). Adoption is near-universal. Google's PMax adoption climbed from 60% of advertisers in 2024 to 71% in 2025 (Fluency), and a 2024 survey found nearly 60% of US ad buyers had used or planned to use products like Advantage+ Shopping and Performance Max (eMarketer). The catch: these are, in the words of one 2026 field guide, "powerful black-box optimizers" confined to a single ecosystem (TensorOps).

Google Performance Max adoption among advertisers
2024  ████████████░░░░░░░░  60%
2025  ██████████████░░░░░░  71%

Source: Fluency. Bars scaled to 100%.

Agentic media buying is the newest mode and the one that's genuinely different. An agentic system doesn't just execute a fixed rule or generate one asset on request. It perceives a situation, reasons through a multi-step plan, and takes actions toward a goal "with minimal human intervention" (TensorOps). We unpack this fully in what is agentic marketing, but the short version is that an agent can handle campaign structuring, bid and budget setting, creative generation and rotation, performance pausing, and reporting, all in pursuit of a target CPA rather than following a script. The part that matters: an agent you control can "reason across multiple platforms," which the single-platform autopilots cannot (TensorOps).

What channels do media buyers actually buy?

Channel selection is where most of the leverage lives. The major lanes in 2026:

Meta (Facebook + Instagram). Still the workhorse for DTC and app performance. Meta is projected to surpass Google in global ad revenue for the first time, with Advantage+ at a roughly $60 billion run rate (Syntermedia, citing industry data). Buyers here live and die by creative volume, because Meta's auction now rewards feeding it many variants and letting the algorithm sort them. Getting the formats right matters, so see Meta ad sizes for 2026.

TikTok. The growth engine for younger audiences and a proving ground for hook-driven creative. The bar for entry is a strong hook in the first second. We cover the metric that tracks it in what is hook rate, and the platform's spec requirements in TikTok ad sizes for 2026.

Google (Search, Display, YouTube, UAC). Search captures intent, YouTube and Display extend reach, and Universal App Campaigns drive installs. Performance Max stitches it all together. Google remains the largest single advertising channel even as Meta closes the gap.

Programmatic / the open web. Beyond the walled gardens, demand-side platforms like The Trade Desk and Amazon DSP buy display, video, audio, and connected-TV inventory across thousands of sites and apps in real time (HubSpot's platform list).

Traditional (TV, radio, print, out-of-home). Still real money, still mostly bought through negotiation and insertion orders, increasingly bought programmatically via connected-TV and digital out-of-home. Less relevant for most performance-led teams.

Across paid social, the formats and aspect ratios differ enough that getting them wrong wastes spend. Our guide on aspect ratios for paid social covers the specifics.

What skills and metrics does a media buyer need?

The skill set splits into two halves: human judgment and numerical fluency.

On the judgment side, negotiation, analytical thinking, business acumen, and attention to detail show up in nearly every job description (Jobicy). The newer additions are programmatic fluency and, increasingly, "AI in advertising" as a named skill (Jobicy). For context on pay, the average US media buyer salary sits somewhere between $68,000 and $96,000 depending on the source and seniority (Indeed, Glassdoor), with the role growing at roughly 15% annually (Jobicy).

On the numbers side, these are the metrics a buyer reads every day:

  • CPM (cost per thousand impressions): what you pay for reach.
  • CTR (click-through rate): how compelling the ad is to its audience.
  • CPC (cost per click): efficiency of getting people to the next step.
  • CPA / CAC (cost per acquisition / customer): the headline efficiency number for most performance buys.
  • ROAS (return on ad spend): revenue per dollar spent, the metric most brands optimize toward.
  • MER (marketing efficiency ratio): total revenue over total spend, the blended view that catches what platform-reported ROAS misses. We compare the two in MER vs ROAS.
  • Hook rate and thumbstop ratio: for video-first channels, how many people stop scrolling in the first seconds.

The buyers who win don't just read these. They know which one to optimize for given the objective, and they don't let a great platform-reported ROAS hide a bad blended MER.

How are AI agents changing media buying?

This is the real shift, and it's worth being precise about. There are three rungs on the ladder, per a 2026 agentic-advertising field guide (TensorOps):

  1. Rule-based automation executes fixed scripts: "if cost > X, lower bid." Useful, but dumb. It can't reason.
  2. Generative AI produces assets on request but is stateless and prompted. It writes the ad copy or generates the video, then waits for the next prompt.
  3. Agentic AI is goal-oriented. It perceives, plans across multiple steps, and acts toward an objective with minimal human input.

The platform autopilots (Advantage+, Performance Max) are sophisticated automation bordering on agentic, but locked inside one ecosystem. The newer wave is agents the buyer controls, operating cross-platform via APIs. The same field guide frames these agents honestly: they're best understood as "tireless, fast, auditable junior media buyers and AdOps analysts, not as a replacement for senior judgment" (TensorOps). And there's a hard caveat worth tattooing on every account: "an agent is only as good as the data, tracking, and structure beneath it." Broken conversion data doesn't just produce bad decisions, it amplifies them at machine speed (TensorOps).

What does an agent actually take off a buyer's plate? Campaign structuring, bid and budget setting, creative generation and rotation, negative-keyword mining, pausing underperformers, and reporting (TensorOps). That's a large chunk of the manual workload. The honest read from the buy side is that agentic AI is, for now, "more interesting than urgent," with accuracy and transparency still the top adoption barriers (TensorOps). The teams adopting fastest are the ones with clean tracking and a clear target, because the agent can only optimize toward a signal it can trust.

We cover the autonomous end of this in depth in the AI agent for autonomous media buying and the broader role of these systems in the complete guide to AI marketing agents.

Where Superscale fits

We build Superscale, an agentic platform that runs the media-buying loop end to end. You paste a link (a Shopify store, an App Store URL, or a website) and the Agent analyzes your product, your competitors, and high-performing ads in your niche, then produces ready-to-launch performance ads (video and static) in minutes. It connects to Meta, TikTok, and Google Ads (on the Advanced plan and above), reads the account, generates variants, iterates on winners, flags or pauses underperformers, and publishes back to the platform. It's free to start, with paid plans from $49/month. HubSpot's CMO Kipp Bodnar called it "the best autonomous AI marketing agent that we have seen so far."

To be straight about it: an agent like this is strongest when you're producing and testing a high volume of creative across Meta, TikTok, and Google. It's not a fit if your buying is mostly direct deals for premium TV or out-of-home, or if your conversion tracking is a mess. It also has a narrower channel set than the enterprise programmatic suites, with no open-web DSP layer for, say, connected-TV programmatic. Match the tool to the buy.

In-house vs agency vs AI: which model should you use?

There's no universal answer, but the trade-offs are clear.

In-house media buyer. You get a dedicated operator who learns your business deeply and is available all day. The downside is cost (a salary in the $68k to $96k range, plus tooling) and the single-point-of-failure risk if they leave. Best when media buying is core to your business and the volume justifies a full-time hire.

Agency. You get a team, breadth of channel expertise, and someone else's playbook from running dozens of accounts. The downsides are markup, slower turnaround, and the fact that you're one client among many. Best when you need senior strategy you can't hire for, or you're spinning up fast and don't want to build internal capability yet. We compare this path directly in AI agent vs performance marketing agency.

AI agent. You get speed, volume, and 24/7 execution at a fraction of the cost. The agent doesn't sleep and doesn't bill by the hour. The downsides are that it needs clean data to work, it's newer and less battle-tested than a senior human, and the best results still come from a human steering it. Best when you're producing and iterating creative at volume and want the manual workload off your plate. The honest framing, again: treat it as a fast junior buyer, not a replacement for senior judgment. We weigh this against building a team in AI agent vs in-house marketing team.

In practice, the strongest setups in 2026 are hybrids: a senior human owning strategy and judgment, with an AI agent doing the high-volume execution and a programmatic DSP handling open-web reach. The question isn't whether to use humans or machines. It's which decisions you keep and which you delegate.

How we evaluated this

The definitions, role descriptions, and process steps here are synthesized from the current top-ranking media-buying guides (HubSpot, Lifesight) and cross-checked against each other. Market-size, programmatic-share, and salary figures are pulled from named industry sources and linked inline; where sources disagree (for example, on the average media-buyer salary), we cite the range rather than pick a single number. The agentic-AI framework, meaning the rule-based / generative / agentic spectrum and the platform-autopilot caveats, comes from a 2026 field guide on agentic advertising (TensorOps) and is corroborated by adoption data from Fluency and eMarketer. Superscale product facts come only from our own published materials. The operator perspective is our own, from running ad accounts and building agentic buying software.

Frequently asked questions

What is media buying in simple terms?

Media buying is buying the space where ads run (a TV slot, a billboard, a Meta placement, a programmatic display impression) so a brand reaches its target audience at the best price. A media buyer negotiates or bids for that space, manages the budget, launches the campaign, and optimizes it to hit a performance goal.

What is the difference between media planning and media buying?

Media planning is the strategy: defining the audience, objectives, budget split, and channels. Media buying is the execution: actually acquiring the placements and running the campaigns. Planners design the blueprint; buyers build it. At smaller companies, one person usually does both (HubSpot).

What does a media buyer do day to day?

A media buyer sources or bids for ad inventory, manages and paces budgets, sets up and launches campaigns, then optimizes them: pausing underperformers, scaling winners, refreshing creative, and reporting results back to the business. On performance channels like Meta and TikTok, most of the day is spent reading dashboards and making budget and creative decisions.

What is programmatic media buying?

Programmatic media buying is the automated purchase of ad inventory through software and real-time auctions, rather than direct negotiation. It uses demand-side platforms, real-time bidding, and private marketplaces to target audiences at scale. It now accounts for nearly 90% of digital display spending (affinco).

How much do media buyers make?

The average US media buyer salary falls roughly between $68,000 and $96,000 a year depending on the source and seniority, with entry-level roles starting lower and senior buyers earning into six figures (Indeed, Glassdoor). The role is growing at around 15% a year (Jobicy).

Will AI replace media buyers?

Not outright, at least not yet. The current consensus is that agentic AI works best as a "tireless, fast, auditable junior media buyer," handling high-volume execution while a human keeps senior judgment and strategy (TensorOps). AI also only performs as well as the data and tracking underneath it, so the human role is shifting toward setting goals, ensuring clean measurement, and steering the agent rather than pulling every lever by hand.

What skills do you need to become a media buyer?

Negotiation, analytical thinking, attention to detail, and business acumen are the foundation, with programmatic fluency and AI-in-advertising literacy now listed as trending and emerging skills (Jobicy). You also need to be comfortable reading metrics like CPM, CTR, CPA, ROAS, and MER and knowing which to optimize for a given goal.

Is media buying the same as performance marketing?

No. Performance marketing is the broader discipline of driving measurable outcomes from paid channels, including strategy, creative, and measurement. Media buying is the purchasing-and-managing layer within it. Every performance marketer buys media, but media buying is only one part of performance marketing.

What channels can you buy media on?

Digital: Meta (Facebook and Instagram), TikTok, Google (Search, Display, YouTube, Universal App Campaigns), and the open programmatic web via DSPs like The Trade Desk and Amazon DSP. Traditional: TV, radio, print, and out-of-home, increasingly bought programmatically through connected-TV and digital out-of-home.

Sources