89% of PPC professionals now use AI as part of their standard workflow. Only 7% use it to actually execute anything autonomously. That gap — 82 percentage points of practitioners who adopted the tools but not the workflow — is the defining split in performance marketing right now.
In this post:
- Why "using AI" and "automating execution" are not the same thing
- The two camps performance marketers are splitting into
- What keeps the 93% in manual mode
- Why the gap closes faster than most teams expect
The Survey Says: Adoption Is High, Agency Is Rare
The 2026 State of PPC survey covered 1,306 practitioners across 50+ countries. The headline: 89% use generative AI or automation tools as a standard part of their workflow. That number sounds like transformation.
Then you read the next one. Only 7% are consistently using agentic automation — AI that actually executes campaign tasks without manual hand-holding at each step.
That's the 7% problem. The other 93% are using AI to move faster inside the same workflow. They prompt ChatGPT for ad copy. They use AI tools to spin up variant headlines. They review AI-generated recommendations in platform dashboards. But when it's time to push the campaign live, adjust the budget, launch the A/B test, or rotate the creative — that's still manual.
The AI saves them time. It doesn't take the work.
Two Types of AI, Very Different Outcomes
The field has split into two camps, and they're not on the same trajectory.
| Faster Typewriter | AI Operator | |
|---|---|---|
| What AI does | Generates outputs for you to use | Executes tasks on approval |
| Where the human sits | In every step | At approval gates |
| Time saved (weekly) | ~5 hours avg | Most execution hours |
| Compounding effect | None — same workflow | Yes — account context builds |
| Example | AI writes copy → you upload it | AI writes, uploads, and launches |
The faster typewriter camp adopted AI to accelerate individual tasks. More copy variants in less time. Faster briefs. Quicker keyword lists. The throughput is real — AI saves around five hours per practitioner per week on average, per the State of PPC survey. That's meaningful.
But the work pattern is unchanged. A practitioner prompts the AI. The AI returns output. The practitioner reviews, edits, and executes. The loop still runs through a human at every step. AI increased the speed of individual tasks; it didn't change who does them.
The operator camp has moved to AI that runs the workflow, not just parts of it. These are agentic systems: tools that read live account data, propose a plan, and execute it on approval. The human role shifts from executor to approver.
The difference isn't philosophical. It's structural. One approach makes each manual step faster. The other removes most of the manual steps.
Why the 93% Haven't Moved
53% of PPC professionals say managing paid media is harder than two years ago. More platforms. More signals. More AI-driven features inside the platforms themselves to configure and monitor. The manual workload isn't shrinking — it's growing faster than individuals can absorb.
So why do 93% of practitioners still execute manually?
Three reasons come up consistently.
Control anxiety. Performance marketers who manage accounts at scale have spent years developing judgment. Handing execution to a system that operates without their direct sign-off on every step feels like giving up the thing they're measured on. The fear isn't irrational — it's a real accountability concern.
Tool-level thinking. Most AI tools in the market are task-level tools: they help with one thing (copy generation, bid suggestions, creative testing ideas). Practitioners have adopted them as productivity tools, not as operators. The mental model hasn't caught up to what's now technically possible.
Integration gap. A genuine agentic system needs access to your live ad account — not just a spreadsheet export. Most tools don't have that connection. The operator camp is smaller partly because the infrastructure to support it is newer.
The Gap Closes on One Side, Not Both
The gap between the two camps isn't stable.
The 7% running agentic workflows are compounding. Every campaign the system runs builds more account context. Every test closes faster because the setup happens when the insight appears, not when someone gets to it. Every budget reallocation fires on the signal, not on the next manual review cycle.
The 93% face the opposite dynamic. As platforms push more AI-native features — automated bidding, Advantage+ structures, AI-generated creative — the manual workflow gets more complex, not simpler. More to configure. More to monitor. More AI outputs to review and override.
The practitioners running AI-native workflows report fewer hours on execution tasks and more time on strategy. That's the return the 93% are leaving on the table — and it widens every month they stay in the faster typewriter camp.
What Execution-First AI Actually Looks Like
Not all AI calling itself an agent is operating agentically. A tool that suggests a budget change for you to implement is not the same as a tool that proposes the change and, on your approval, makes it.
Execution-first means the system is connected to your live account, reads current performance data, builds a plan in plain language, waits for your approval, and executes end to end — without you opening Ads Manager for each step. The approval keeps you accountable. The execution removes the bottleneck.
This is the workflow bulk is built for. bulk reads your Meta ad account — campaigns, creatives, performance history — and proposes exactly what it intends to do before touching anything. Once approved, it builds the ads, adjusts the budgets, or launches the test. The output is a live campaign, not a recommendation.
For teams running Meta Ads A/B testing at scale, the difference is concrete: tests that used to take days to configure and launch can close the same day the signal appears. That's what crossing from recommendation to execution looks like in practice.
The Honest Assessment
The 7% isn't a small group because agentic automation doesn't work. It's small because the category is new, the tooling is newer, and most practitioners haven't yet separated "I use AI" from "AI handles my execution."
Those are different things. One is a workflow accelerant. The other is a structural change to who does the work.
The 2026 State of PPC data captures the moment before that distinction becomes obvious to everyone. For the marketers already in the 7%, it's obvious now. For the 93%, the window to close the gap on their own terms — rather than on their competitors' — is still open.
bulk connects to your Meta ad account, proposes a plan, and executes it without the manual steps in between. Try bulk free →