AI seemingly hasn’t come for our jobs the way people feared. I don’t know of a single agency that has replaced its junior PPC specialists with it. What I do see is teams using AI on top of shaky foundations and producing unchecked work more quickly.
AI can write fifty ad drafts, summarize a month of performance, or run a Quality Score audit before you’ve had your morning coffee. What it doesn’t have is the context that never shows up in the account data. It doesn’t know the client’s margins are thin on one product line, that their leadership is nervous about a new channel, or that last month’s ROAS drop lines up with a site migration. It can suggest a dozen tests, but it can’t sit in the QBR, hear what the client is worried about and decide which test is worth the risk this quarter.
That judgment still has to come from people. So when I think about building a strong PPC team right now, my goal is simple: develop foundational skills strong enough that AI makes their jobs easier, without taking over the thinking.
Here’s what we hire for, how we train, and where AI fits once the foundations are in place.
The Skills We Hire and Train For
I’ve been asked what new skills PPC people need in the age of AI. My honest answer is that most of them aren’t new. The qualities that made someone great at this job five years ago still do. AI has made the gap between people who have them and people who don’t much easier to see.
Here are the three I care about most.
1. Curiosity
If you like learning, you’ll do well in this job. If you don’t, the job will keep getting harder.
Paid search is less procedural than it used to be, but the platforms are also more standardized. Every agency has access to the same automated bidding, the same Performance Max and the same AI assistants. The work that sets one account apart comes from a specialist who notices something in the data, wonders why, and digs for the answer without being told to.
When Google changed how its ROAS and CPA targets behave this fall, one of our analysts didn’t wait for a blog post to explain it. They pulled before-and-after data across our clients, compared affected campaigns with a control group, and worked out what the change meant for how we set targets. Nobody assigned that analysis, but it changed how our department manages targets.
2. Self-management, with visibility
Since we went fully remote, self-management has become essential. Success requires a few related skills working together.
People need to be independent and manage their own time and projects without someone checking in every hour. But working independently also means it’s easy to become invisible. A big part of our team’s success is that people manage themselves and also speak up. They ask for help early, request training when they need it, and tell their manager when something is off.
AI raises the stakes. Tools change too fast for managers to hand-hold everyone through each one. So the behavior I value most is when someone figures something out on their own and then shares it. One of our analysts built a Claude skill that audits low Quality Score keywords and shared it with the whole company for feedback. Within a day, other managers were using it on their own accounts.
One person learning a tool helps one account. One person sharing it helps everyone.
3. Data interpretation
AI can give you 40 KPIs in seconds. The skill is knowing which ones matter to each client, how they affect each other, and which ones are vanity metrics.
I want people who stay scientific. A good example is click-date vs. purchase-date ROAS. Google Ads credits revenue to the day of the click; GA4 credits it to the day of the purchase. Ahead of Black Friday, we compare click-date and purchase-date ROAS by day of the week to see which days’ clicks keep paying off. That tells us when to push spend and when to pull back, so we’re deciding from the full picture and not just what the platform shows that morning.
The same goes for metrics that work out of sight. Quality Score never shows up in a client report, but it affects every CPC. A strong specialist notices it, rewrites the ad copy and, when the problem is the landing page, has the conversation with the client about fixing it.
How We Build the Foundations
This year we spent most of our training time on two things AI can’t do for you: turning data into a point of view and saying it clearly to a client.
Reports that answer “so what?”
We started with an audit of our own client reports. Too many of them recited metrics: “CPC increased 44% and cost 25%, leading to a 13% decrease in clicks.” The data may be accurate, but the client can’t do anything with it.
So we set a simple standard: every insight has to say what happened, why it matters, and what we recommend doing next.
We ran an “insights mindset” workshop, built a library of strong examples and, for one month, had every report reviewed by a teammate or manager before it went out. We also reviewed real reports as a group, before and after, and picked them apart together. The PPC team called these sessions “roasts” and enjoyed them.
Communication you can measure
We took the same approach to client communication. We defined what “good” looks like, had everyone score themselves while managers scored their teams, and used the gaps to decide what to train.
Then we ran short workshops, mock client calls, and mock QBRs so people could practice before it counted. When we re-scored a few months later, preparation and energy had improved the most. The better your people communicate, the more an AI-drafted report or deck is worth, because someone can stand behind it in front of the client.
What we stay mindful of
The group roasts worked for PPC because of the team’s culture. They’re curious, they want to get better, and they aren’t afraid of being wrong in front of each other, even on a video call. Not every team is built that way.
The lesson for other leaders: match the feedback format to the team. Group critique builds a strong team faster, but only when the team already trusts each other. Start with private feedback and homework, and earn the trust for group sessions.
How We Hire for It
Our biggest hiring lesson from this year: we can fill skill gaps, but we can’t fix personality or culture fit. A great hire with no ecommerce experience can learn the ins and outs of shopping feeds in a few months. Someone who doesn’t care about improving or won’t speak up when they’re stuck rarely changes.
So here’s what we screen for now:
- Do they care about their work? Ask about a result they’re proud of and listen for whether they explain why it worked.
- Are they willing to learn new skills? We ask every candidate how they’re using AI today. We’re not looking for a specific tool. We’re looking for whether they’ve tried things on their own.
- Can they communicate on camera? In a remote agency, the interview is a preview of the client call. Do they present well over video? Do they have presence? Can they speak confidently about themselves and also stop when they’ve made their point?
- Can they interpret data under pressure? Our second-round interview includes a short, hands-on data exercise. We care less about the right answer than about how candidates think it through, and whether they ask for more context before making a recommendation.
- Have they juggled many moving parts? Experience managing lots of projects at once or working within a team is a big plus. Agency work is dynamic, and people need to be flexible.
For experienced/senior hires, agency experience is a must. We’ve found that people with only client-side experience often struggle with the pace of working on multiple accounts. However, we’d rather promote a strong specialist into a manager role and hire a new specialist for that spot than hire a manager from outside.
Where AI Fits on a Strong Team
Once the foundations are there, AI saves a lot of time. Getting a team to use it well took more structure than I expected.
Tools don’t get used just because you built them. Earlier this year, we built two internal AI reporting tools. When we surveyed the PPC team, few were using them as intended. Everyone else had gone back to their own personal chats, with mixed results. The cause was simple: nobody had been properly onboarded to them. We wrote short training guides, asked everyone to use each tool on at least one monthly report, and picked one as the team standard. With good client context loaded in, a first draft of a monthly report now takes 5 to 10 minutes and needs very little editing.
Test before you roll out. When we tried a popular open-source AI setup for PPC, our director found it needed developer-level configuration in the terminal. That’s fine for one curious manager. It wasn’t realistic for a junior analyst. We paused the rollout, had a small group work through it properly, and only then decided where it fit.
Give curious people a place to share. We have a Slack channel for AI experiments and a shared tracker where anyone can log a tool they’ve tried. Several of the tools our team uses every day started there. We find the best ideas come from the people closest to the accounts, not from leadership.
Keep a human in the loop, in writing. We built a QA checklist for any deliverable that AI helped produce. It covers accuracy, brand fit, strategic relevance, and policy compliance, and it requires a person to review the work before it reaches a client. The team builds the same habit into their own tools.
AI makes a strong specialist faster. It does nothing for a specialist who can’t tell a good recommendation from a bad one, so build that judgment first.
Ready to get more from your PPC campaigns? At Eight Oh Two, we combine experienced paid search specialists with smart AI tools to make better decisions and drive stronger results. Get in touch to see what our team can do for your business, or connect with us on LinkedIn for more insights on paid search and AI.
