The media buying workflow in 2026, step by step
The nine-step media buying workflow high-performing teams run weekly — from creative research to feedback — and what to automate at each step.
Ask five media buyers to describe their workflow and you’ll get five org charts and zero processes. But underneath every account that compounds, the same loop runs on a weekly cadence: research, insight, hypothesis, concept, creative, launch, run, read results, feedback. This post walks the loop step by step, with the honest note on what’s automatable at each stage in 2026.
If you need the fundamentals first: what is media buying and what is a media buyer.
The loop at a glance
| Step | Weekly output | Automatable in 2026? |
|---|---|---|
| 1. Research | Competitor + trend notes | Largely |
| 2. Insight | Why winners win | Partially |
| 3. Hypothesis | Testable statements | Partially |
| 4. Concept | Angles and formats | Largely |
| 5. Creative | Finished assets | Largely |
| 6. Launch | Live, named, tracked | Largely |
| 7. Run | Budget allocation | Mostly platform-native |
| 8. Read results | Winner/loser calls | Largely |
| 9. Feedback | Next week’s brief | The human job |
Steps 1–3: research to hypothesis
The research step is creative research more than audience research now — what competitors run, what’s been running longest, what the culture is rewarding this month. The free stack covers most of it (how to find competitor ads is our walkthrough), and ad intelligence software makes it continuous instead of quarterly.
Insight and hypothesis are where buying quality is actually determined. “Their longest-running ad is a street interview” becomes “third-party voices beat brand claims in this category,” becomes “we believe a street-interview format will lift CTR for us — measurable in a two-week test.” Write hypotheses down. The accounts that improve are the ones where week 30’s brief knows what week 1 learned.
Steps 4–6: concept to launch
This is the stretch that used to consume the whole week — briefing designers, waiting on edits, uploading, naming, QA. It’s also the stretch that agentic tools genuinely absorbed. Taxfix runs Superscale’s agent as an extra creative strategist, copywriter, video producer, editor, and designer in one: insights from Reddit, Meta, and TikTok become testable concepts, scripts, storyboards, and finished assets, which shipped 200+ ads across Meta, TikTok, and Google UAC at +45% CTR and −20% to −21% CPA. SumUp compressed idea-to-launch from weeks to days and shipped 20 Black Friday assets across 8 markets in one week.
Launch discipline is unglamorous and decisive: naming conventions that make step 8 readable, one variable per test, and a paper trail of what each ad is testing. A generator that publishes directly into the account (Superscale does, from its $99 tier) removes the download-upload gap where naming discipline usually dies.
Step 7: run
Platform algorithms won this step. Advantage+ and PMax allocate spend better than manual bid babysitting for most accounts; your leverage is the budget envelope, the offer, and campaign consolidation — not hourly bid nudges. Rules tools still help at scale; the field is in best media buying tools.
Steps 8–9: read and feed back
Read at two levels. Hook-level within 48 hours (thumbstop ratio tells you if the concept earned attention), conversion-level at the account view (blended CPA / MER — per-ad ROAS lies more as spend grows; benchmarks in ad benchmarks).
Feedback is the step machines don’t own: deciding what the results mean for next week’s bets. The practical artifact is a one-page brief — won, lost, untested, next. Teams that write it compound; teams that don’t rediscover their own findings quarterly.
Running the loop lean
A two-person team can run this weekly if steps 4–6 and 8 are automated. That’s the configuration behind most of the case numbers we cite — Lila runs 20 tests a week with a small founding team; Advercy runs five client brands solo, at −50% CPL. The tool layer for that configuration: best ad testing tools for the loop, ai media buying guide for the division of labor.
FAQ
How long does one cycle of the media buying workflow take? High-performing teams run it weekly. Monthly cycles were normal when production took weeks; they’re a competitive disadvantage now.
Which steps should stay human? Hypothesis selection, brand judgment on concepts, budget envelopes, and the feedback brief. Execution in between is automatable.
What’s the biggest workflow mistake? Skipping step 9. Results that don’t change next week’s brief are just expensive trivia.
Related reading
Letters from readers
-
Q·01 How is ad-stack funded?
We pay for every tool seat ourselves at the public plan tier, and the journal is reader-supported via the newsletter. No vendor pays for placement, and no review is sponsored.
-
Q·02 Why benchmark on the same brief instead of letting each tool play to its strengths?
Because the only fair variable in a head-to-head test is the tool. Letting each vendor pick their best demo brief is how the AI ad category got into its current marketing-led mess — every tool wins on its own showcase. Same brief means you can actually compare cost-to-published across the field.
-
Q·03 How often do you re-test tools that have shipped major updates?
Every quarter. Reviews carry a 'last tested' date in the byline. If a tool ships a meaningful capability change between quarterly cycles, we publish a field note rather than waiting — but the score on the main review only moves at the next full re-test.
-
Q·04 Can I send in a tool to be reviewed?
Yes — send a note via the contact link in the footer. We can't promise coverage of every submission, and being suggested has no bearing on the eventual verdict. Vendors who pay for seats themselves rather than offering us free credits are evaluated identically.
