Can AI Really Replace a Media Buyer's Strategic Instincts? A Balanced Look
15 min read
Loading...
Not entirely, and that's the honest answer. AI is brilliant at the heavy lifting: spotting patterns across hundreds of competitor ads, flagging what's new, what's stopped and what's scaling, and doing it every week without getting tired. But instincts built from budget pressure, brand context and market feel still make the final call. The best results come from treating AI as a sharp first-pass analyst and yourself as the decision-maker.
That's the short version. The longer version is more interesting, because the useful question isn't whether AI can replace a media buyer's strategic instincts. It's which parts of the job you should hand over and which parts you should hold on to. Let's get into it.
Picture a Monday morning. You've got a dozen tabs open: the Meta ad library, Google's Ads Transparency Center, a few TikTok searches, a couple of LinkedIn company pages. Your client wants to know what the competition is up to, and you've got forty minutes before your first call. You're scrolling, screenshotting, squinting at dates and trying to remember whether that carousel was there last week.
No wonder so many of us wonder whether software should just do this. And then, a few seconds later, we wonder what that means for our jobs.
Here's how I'd frame the tension. It isn't 'AI versus me'. It's 'what should I hand off, and what should I keep?'
To answer that, we need to be clear about what we mean by strategic instincts. I'd break them into four things:
A quick bit of transparency before we go on. We build a tool in this space, Rival Ads, so I have a stake in this conversation. That's exactly why I want to be upfront about where AI falls short as well as where it shines. If I oversold it, you'd spot it in a heartbeat.
Let's start with the good news, because there's plenty of it. AI does best when the job is high-volume, repetitive and needs doing consistently. That's true across advertising generally. Google's Smart Bidding uses auction-time signals such as device, location and time to tailor bids, which no human could do by hand across thousands of auctions. IAB's primer also points to high-volume monitoring, like flagging delivery changes or unusual performance, as a practical use that frees buyers to investigate causes and make decisions.
Competitor analysis has the same shape. Here's where AI earns its keep:
This is the workflow we built Rival Ads around. You give it a competitor's website, and you don't need to connect an ad account. It detects their presence across Meta, Google, TikTok and LinkedIn, fetches every active ad weekly (real creatives, exact copy and links) and produces the week-over-week diff. Then Claude interprets those changes the way a senior media buyer would, flagging what's new, what's stopped and what's continuing to scale.
In practice that's ad monitoring plus a first-pass analyst, delivered to your dashboard and inbox.
AI sees ads. It doesn't see spend, conversion rates or audience targeting. No ad spy tool does, however slick the interface. Longevity and creative volume are useful proxies, but they're still proxies. Hold onto that, because it shapes everything in the next sections.
There's a wider strategic boundary too. As the IAB's primer puts it, AI optimises towards the goal and signals it receives, but it doesn't independently decide whether that goal is commercially sound or whether a short-term change reflects a lasting opportunity. That remains the buyer's job.
It depends which kind of accuracy you mean. I find it helps to split it in two.
Factual accuracy is about what exists: which ads are live, what the copy says, when they started or stopped, where they link. AI-assisted tools can be genuinely dependable here, provided the data comes directly from platform sources such as the Meta Ad Library and Google's Ads Transparency Center rather than scraped guesswork. Source quality matters enormously. A confident summary built on shaky data is just a confident mistake.
Interpretive accuracy is about why. Why did they pause that ad? Why the sudden offer-led messaging? What's the strategy? Here AI offers a well-reasoned hypothesis, not proof. It's a probability, not a certainty. And to be fair, there's no standardised accuracy score for AI systems that analyse competitors' ads, so any claim that an AI can reliably pick out a rival's 'winning creative' needs validating against campaign-level evidence.
Public ad archives only show what the platforms choose to make available. They don't reliably reveal:
Coverage also varies by platform. Meta, Google, TikTok and LinkedIn each disclose different levels of detail, so your picture of a competitor will be sharper in some places than others. Anything an AI says about those hidden factors should be read as an estimate, not a fact.
Check AI claims against the actual creative and copy. That's why archiving real creatives and links matters so much. If an analysis says a competitor has shifted to urgency-led copy, you should be able to click through and see the ads yourself in about thirty seconds. Keep the evidence, separate what was directly observed from what was inferred, and have a human look at anything ambiguous, especially imagery or cultural references that software can misread.
Now the other side of the ledger. This is the stuff that never shows up in a competitor's ad feed.
Imagine a competitor suddenly launches a burst of new ads in a single week. On the dashboard it looks like a major push, and a summary might reasonably call it an aggressive scale-up. But anyone who knows that market recognises the pattern: it's a one-off clearance ahead of a product refresh. Responding with a big budget increase would have been a mistake. Nothing in the ad data says 'clearance'. Only market knowledge does.
That's the kind of call instinct makes, and it's why I don't think instinct is going anywhere.
Let's make this concrete. The scenarios below are illustrative examples, not real client data. They show how an AI read and a human read can agree, partly agree or diverge.
| Scenario | AI read | Human read | Which was more useful? |
|---|---|---|---|
| 1. 15 new Meta ads in a week, heavy discount angle | Aggressive promotional push; shift towards offer-led messaging | It's the week before payday and a known sales window, so likely tactical rather than a lasting change | Human. The AI spotted the shift, but the calendar context changed the response (partly agree) |
| 2. Long-running Google search ad stopped; LinkedIn thought-leadership series starts | Pivot towards B2B upper-funnel | You've heard they hired a new head of brand, which explains the move and suggests it'll stick | Both. AI flagged it early, the human explained why it matters (agree) |
| 3. Rival's TikTok creatives scale fast with a new hook format | This format is working; here's the pattern | Fits their younger audience but would feel off for your premium positioning, so borrow the hook structure, not the style | Both, in sequence. AI found the pattern, the human adapted it (partly agree) |
| 4. Same ads running unchanged for months | Stable, probably performing | Could be neglect. Check landing page and offer freshness before assuming it's a winner | Human. Longevity isn't always success (diverge) |
Notice the pattern. In none of these did the AI get it wrong, exactly. It described what was visible accurately. The gap appeared when the question shifted from 'what's happening?' to 'what should we do about it?'. That's where context, calendars, hearsay and brand sense take over.
My rule of thumb is trust, but verify. Treat AI notes like a briefing from a smart junior. They're fast, thorough and often spot things you'd miss, but they're worth a sanity check before they land in a client deck.
Trust can be higher for:
Trust should be lower for:
One habit pays off more than any other: keep a simple log of which AI reads you acted on and what happened. Over a few months you'll build a calibrated sense of when to lean on the analysis and when to push back. It's the same way you learned which colleagues' hunches were worth following.
So how do you put this together? Here's the routine I'd recommend.
Notice who leads each step. AI leads the first one or two, you lead the middle, and the loop closes with a joint review. It's a partnership, with the human holding the pen.
If you run an agency, this model has a bonus. Whitelabel reporting lets you add your own strategic layer on top of the AI analysis and present it as your own value-add. The AI handles the weekly legwork of creative tracking and diffs, and your team supplies the judgement clients actually pay for. That's a better use of senior time than screenshotting ad libraries.
The biggest shift is that the job moves from collecting information to interpreting it. Honestly, that's the more strategic half of the work anyway. Few of us got into marketing strategy because we love scrolling through ad archives.
The time you save can go into things that actually move results: creative testing, client conversations and deeper analysis. Junior buyers benefit too, since they can learn faster by comparing the AI's read with senior feedback, a bit like checking your working against the answer sheet.
On cost, a monitoring tool starting from $29/month is a modest outlay next to the hours of manual checking many teams spend each week. McKinsey has estimated that generative AI could lift marketing productivity by the equivalent of 5–15% of total marketing spend. That's a broad estimate rather than a promise for any one team, but it points the same way: the value is in freeing people for higher-order work.
So here's where I land. AI doesn't replace instinct, but it gives instinct better raw material. The buyers who'll do best are the ones who let software read everything and then bring the one thing it can't: a view on what it all means for their business.
It's very reliable at the factual layer, such as which ads are live, what the copy says and what's new or stopped week to week, provided the data comes straight from platform ad libraries. It's less certain on the 'why', because it can't see spend, conversion data or targeting. Treat interpretations as informed hypotheses.
Trust them as a strong first draft, not a final verdict. They're most dependable when summarising changes and patterns, and least dependable when predicting competitor intent. Check notes against the actual creatives and add your own commercial context before acting.
Use AI for monitoring and first-pass analysis, then apply your knowledge of your client, market and budget to decide what to do. A simple routine works well: review the weekly diff, read the AI analysis, add your context, then choose to ignore, watch, test or respond.
No. Ad libraries and transparency tools show what's running and for how long, but not budgets, ROAS or targeting. Longevity and creative volume are useful proxies, but they aren't spend data.
No. Because the ad data is publicly available, you only need the competitor's website. Rival Ads uses it to detect their presence on Meta, Google, TikTok and LinkedIn, then archives their active ads each week.