AI tools can run a paid media account. A paid media consultant makes the decisions that sit above the account: what a customer is worth, who you sell to, which platform does which job, what to test next and what happens after the click. A retainer pays for those decisions. Mine starts at €2,000 a month. The execution is faster than it used to be, and it is the smaller part of the work.
This post is for B2B SaaS founders and marketing leaders deciding whether to hire for paid media, rely on a tool, or combine both.
What can't your ad reports tell you?
Every ad platform reports its own conversions. Google claims the customer, Meta claims the same customer, LinkedIn claims them too. Add the reports together and the total regularly credits more customers than you actually signed.
A tool connected to one account optimises against what that account reports. Tools that connect several platforms and your CRM can compare them side by side, and some will recommend budget moves. The comparison is only as good as the attribution underneath it, and attribution cannot show what happens to one platform when you change another. That takes a test, someone to read it, and a CAC target someone has decided is right. That question is the job, and it sits above any single campaign. The six decisions below answer it.
What is a customer worth, and what can you afford to pay for one?
Every paid media decision starts with unit economics: customer value, gross margin, how long a customer takes to pay back their acquisition cost, and how many stay. From those you get your allowable CAC, the most you can pay to win a customer and still run a healthy business.
That number sets the bar every platform has to clear. Whether a platform can clear it is a forecast, because nobody knows how leads from a platform will close until it has run. Click prices say little here. Cheap clicks with leads that never close are expensive, and dearer clicks with customers who stay can be the best deal on the plan.
A tool optimises on the data your account has produced so far, and a platform you have never run has produced none. Experience with similar accounts fills that gap. Having run companies that sell to the same buyers shows which platforms tend to produce leads that close, what a realistic cost per qualified lead looks like, and where the first tests should go. That is why I specialise in B2B SaaS, and within it in a few niches. The forecast goes into a test, and the test result replaces the estimate.
A tool optimises towards the target you give it. Setting that target is a business decision, and a wrong target gets pursued very efficiently. For a quick sizing, start with the CAC calculator.
Is your ICP right, and who are you really competing with?
Every AI tool starts from the context you give it: your ICP, your offer and your conversion data. The tool works with that input and has no way to question it. If the ICP is wrong, the tool gets very good at finding the wrong people, and the platform reports can look healthy while it does.
A consultant checks the input before it goes in. Lead quality data from your CRM shows which accounts close, and that feeds back into who you target and what you promise. Narrowing the ICP often means fewer leads and better ones. The consultant then shapes what the tool receives: the ICP, the offer angle and the conversion data it learns from. The tool still does the heavy lifting, and it works with the right input.
On one multi-product B2B SaaS account, the share of leads that qualified rose from about a third to just over half, on lower lead volume. The figures and their caveats are on the services page.
The competitive set is part of that context, and it is just as easy to get wrong. One client sells filmed testimonials with real customers, at a price above $3,000. Their landing page measured the company against other studios doing the same thing. Their keywords were also being bid on by tools that generate avatar testimonials for around $50. A searcher was choosing between a $50 option and a $3,000+ one, and the page only answered the second question.
No bid change fixes that. The offer angle had to speak to the real alternative. Defining the competitive set is a business decision that sits upstream of the ad account. It decides what the ads promise and what the page has to prove. A tool can test variations all day, and someone still has to tell it which alternative the buyer is weighing.
Which platform should serve which stage of the journey?
Any platform can serve any stage of the journey, and each is better at some. Google and Microsoft search capture intent, so they work best in the middle and bottom of the funnel. Meta, LinkedIn, Pinterest and Reddit are better at creating demand among people who are not searching yet. Those are tendencies, and your buyers may not be on a given platform, or may be there in the wrong frame of mind for your offer. A consultant advises which stage each platform should serve for your buyers, then judges it on a metric that fits that job. A demand-creation platform judged on last-click cost per lead looks like a failure and gets cut.
Category labels mislead here. Most marketers file Pinterest under e-commerce. On a B2C SaaS account I worked on, it worked. The category label said no. The audience and the journey said it was worth trying, and it paid off. The same question applies in B2B: which platforms reach your buyers at which stage, whatever the label says.
New platforms make the question harder. OpenAI Ads is the current example. ChatGPT began showing ads to US users in February 2026, self-serve buying opened in May, and it has since spread to dozens of countries, Europe included. Targeting follows the context of the conversation, so it sits somewhere between search and social: people researching and comparing options, with no keyword to bid on. Ads show to Free and Go users, so whether your buyers see them is a real question for a B2B SaaS audience. I run it for some clients, and it will not suit every account.
A new platform has no history in your account and little in your niche, so no tool can tell you whether it fits. A consultant decides whether it reaches your buyers, which stage it could serve, and what a small, time-boxed test should cost before any money goes in. More platforms will follow, and each will arrive with the case that it is the next big channel. Some will suit your buyers and most will not yet. Knowing which is the job.
How many platforms you can run depends on budget. At around €10,000 a month, one platform is usually the right call and two is the limit. Spread any thinner, and each platform has too little data to learn from or to test on.
With one or two platforms in play, the question gets sharper. Which one or two deserve the budget, and which stage each should serve, is a call you cannot afford to learn by trial and error.
What happens to one platform when you change another?
Platforms feed each other. A common setup has Meta creating demand and Google capturing it when people search. Pause the first and the second can fall.
A SaaS client's Meta account was suspended for about six weeks. Google activations fell over that period, with the brand campaign taking the biggest hit. That is the pattern you would expect when Meta creates demand that Google then captures, because brand search is where upstream demand lands. Read inside Google Ads alone, it looks like a Google problem. Across platforms, the cause sits somewhere else.
The same applies inside a single platform. On one Google Ads account, Demand Gen campaigns brought no leads of their own, so cutting them looked like free savings. After the cut, leads and MQLs fell across all the search campaigns in the account. Judged on its own conversions, Demand Gen looked like waste. A tool that judges each campaign on its own numbers has nothing to warn it until the drop shows up in the data. A consultant who has seen the pattern in similar accounts expects it, and tests the cut before making it.
Before you cut an upstream platform or campaign type, reduce its budget in stages and watch branded search and lead volume downstream. The measurement is imperfect, so treat the results as evidence and decide with that in mind.
What should you test first?
Not every test deserves running. A test plan is a ranked list of hypotheses, each tied to a number it should move. Rank them by expected effect on CAC and by how much data you have to read the result. Fix tracking quality first, then the offer and the landing page, then campaign structure, then bids and settings.
Budget limits what you can test. Small budgets cannot support many tests at once, and one change at a time keeps results readable. Tests also need time. A campaign judged after a few weeks on a tight budget teaches you very little. Patience and realistic targets matter as much as the hypothesis itself.
What happens after the click?
Click costs keep climbing. WordStream's 2026 benchmarks put the average cost per click across Google and Microsoft Ads at $5.42, up from $5.26 the year before and more than double the $2.32 recorded in 2016. Alphabet reported that paid clicks on Google Search rose 6% in 2025 while cost per click rose 7%. When the price of a visitor rises, the share of visitors who become leads decides your CAC. Raising conversion rate lowers CAC at the same budget, and it does so without competing in any auction.
Here is the arithmetic, with illustrative numbers. Take a $5.42 click. At a 4% visitor-to-lead rate, a lead costs about $135. At 6%, it costs about $90. The click price is the same and the auction is the same, and each lead costs a third less.
Sources: WordStream, 2026 Google Ads Benchmarks; Directive Consulting, Google Ads for B2B in 2026, citing Alphabet's 2025 results.
Post-click work covers landing pages built per campaign and per offer angle, a page that keeps the promise the ad made, and lead quality signals passed back to the ad platforms so they learn which leads sales wants. It is also where the offer and ICP decisions above show up on screen. A tool can optimise the ad. Someone has to decide what the page says.
When is a tool enough?
Sometimes. One platform, a modest budget, a single simple offer, a proven ICP and a conversion the platform can see end to end leave few decisions to make. That is where full automation will do best as the tools improve. If that describes your setup, try a tool first.
The decisions in this post appear as soon as one of these is true: more than one platform, a sales team that decides which leads count, an ICP you are still testing, or a CAC target that matters. That is the point where a person who owns the whole picture starts earning their fee. Come back when those questions start to apply to you.
What does a €2,000 retainer cover, and where does AI fit?
My ongoing retainers start at €2,000 a month. One fixed fee covers strategy and execution from the same person. The fee depends on scope, which we agree on the first call, and it stays fixed month to month. It is never a percentage of your spend. Terms are month-to-month with 30 days' notice, you own all accounts and data, and AI agents have read-only access.
The fee has to earn itself back in cost per qualified lead and cost per customer. On the first call we agree which numbers decide that.
I use PPC OS from PPC Hub as the foundation for the AI agents that handle audits, search term reviews, reporting and landing page audits. I also do competitor research and build landing pages around the offer angles, delivered as HTML that is ready to implement with minimal customisation, so your developer has less to do. AI makes all of it faster. The hours it frees go into the decisions in this post, which is where the budget gets won or lost.
Who is this not for?
Budgets below about €10,000 a month, or spread over more platforms than the budget can feed.
Teams where nobody owns tracking or the CRM, because the link between ad spend and lead quality cannot close.
Anyone who wants a single-platform specialist. A Google Ads specialist will recommend Google Ads. I recommend wherever customers cost least to win.
I keep three to five clients at a time, so fit matters on both sides.
Frequently asked questions
Can AI replace a paid media consultant? For simple setups, increasingly yes. For multi-platform B2B SaaS, the decisions about unit economics, ICP, offer, platform roles and testing stay with a person who sees the whole system.
How much does a paid media consultant cost? Mine start at €2,000 a month. The fee is fixed month to month, set by scope on the first call and never a percentage of spend. Prices elsewhere vary with scope.
Do you run the campaigns as well as plan them? Yes. Strategy and execution come from the same person.
Should I hire a single-platform specialist or a consultant across platforms? If budget needs to move between platforms, you need someone with no favourite. A specialist has a reason to recommend their own platform.
How long is the contract? Month-to-month, with 30 days' notice.
Bring your budget and your targets. I'll tell you where I'd put the money.
