Setting up, managing and analyzing ad campaigns on Meta and Google Ads
Over the last ten years I have launched hundreds of ad campaigns for companies across the Baltics and the Nordics. And honestly, most campaigns that come to us "for an audit" are broken for the same reason: someone pressed the Boost button, poured money in and waited for a miracle. One online store spent two months on boosted posts before we mapped that 70% of the budget had gone to phantom conversions the pixel was double-counting.
This marketing guide is not a Wikipedia-style summary of "how Meta and Google work". It is what I would actually do if you had a 1,500 € monthly budget and had to report on it to your director next Friday.
Why the setup stage decides everything
90% of a campaign's success is decided before you publish the first ad. If the structure is broken at the start, later optimization is just an expensive synonym for damage control. What I always nail down first:
An audience that is not "women 25 to 55, your country". That is not an audience, it is a census. A real audience is "added to cart in the last 90 days and abandoned", "viewed a specific category 3+ times" or "the Klaviyo VIP segment that has not bought in 60 days".
Measurement before money. If the pixel, GA4 and the Conversions API are not set up correctly, you are not optimizing anything; you are throwing coins into the wind. Always check with Pixel Helper that one event does not fire twice.
A budget that lets the algorithm learn. Meta needs about 50 conversions a week per ad set to finish the learning phase. If your CPA is 20 € and you give it 5 € a day, be prepared to wait a month and accidentally sell a safety pin to your grandmother's friend.
Maison Mint expertise: Never start with Performance Max before you have at least 30 days of conversion data from a Search campaign. PMax is a black box: if you have no signal to feed it, your money goes into YouTube side placements and you never learn what worked.
Meta Ads: where most of the money disappears
Campaign objective: choose conversion, not "engagement"
If someone recommends an Engagement campaign, ask them when they last went to the bank. The engagement objective brings likes and comments that do not pay the rent. Choose Sales (Meta) or Conversions, set a concrete event (Purchase, Lead, not ViewContent) and let the algorithm work.
For most small businesses a "brand awareness" campaign is simply burning resources. If you do not have a 50k+ monthly budget, the brand gets built as a by-product of conversion campaigns, not separately.
Targeting: broad beats narrow, but with data
Meta's algorithm has become so good between 2023 and 2026 that targeting too narrowly does harm. What I do: start with broad targeting (age 25 to 55, the country, no interest targeting), add a pixel- and customer-list-based lookalike (1 to 3%), and let the machine run. Interest-based targeting in 2026 is about as effective as a fax machine.
What actually works: retargeting people in your city who visited a product page in the last 14 days but did not buy. In smaller EU markets this segment typically costs 3 to 7 € CPM and brings a 4 to 8x ROAS (our accounts, illustrative).
Creative: three versions or death
One ad design is a guarantee of failure. My rule: every campaign starts with at least three different creatives, a UGC-style video, a static image with the offer, and a carousel. Not "blue button vs green button"; that is a waste of time. Test concepts, not colors.
In smaller markets a human face in the ad performs roughly 30% better than a product photo, especially when the person speaks the local language (our accounts, illustrative). A stock-photo blonde on a white background? Skip.
Google Ads: simpler, but merciless
Keywords: start with the letters B and C, not A
A huge mistake is to start with brand keywords or the most popular generic term ("dentist" plus your city). Competition there is merciless and the CPC can be 6 to 15 €. My approach: start with long-tail keywords, 4+ words long with clear purchase intent ("dental care without insurance" plus the district).
Google Keyword Planner is a starting point, not the end point. You see real search volume in the Search Terms report after two weeks of actually running.
In smaller EU markets the average CPA in Google Search is 8 to 22 €, depending on niche (our accounts, illustrative). If you report a first-month CPA of 4 €, you are probably measuring wrong.
Ad copy and extensions
Responsive Search Ads are the only sensible choice in 2026; Google gives you 15 headline slots, fill them all. But do not write 15 synonyms for "quality". Write 15 different angles: price, speed, guarantee, returns, social proof, local context.
Extensions (sitelinks, callouts, structured snippets) often lift CTR by 20 to 40%. They are free. If you do not use them, you are handing money to a competitor.
Negative keywords: the most underrated weapon
Every Friday ritual: open the Search Terms report, sift out everything with "free", "DIY", "jobs", "price" (unless you are a price list) and add them as negatives. One B2B client saved about 600 € in two months just by adding 40 negative keywords.
Controversial opinion: Automated bidding ("Maximize Conversions" and the like) works better than manual CPC in 95% of cases, but only when you have clean conversion data. If measurement is broken, automation just optimizes the wrong goal faster. Setup beats strategy.
What you actually look at every week
Forget dashboards with 47 metrics. Every Monday morning I look at four things, and sometimes nothing else:
ROAS at ad set level (not the account-level average; that one is misleading).
CPA trend, the last 7 days vs the previous 7. If it rises 20%+, something is broken.
Frequency. If the same person sees your ad 8+ times, you are not "building a brand", you are annoying people.
The conversion path in GA4. Did the ad really bring the sale, or was it a brand search that the ad glued to itself (the last-click attribution folly)?
Everything else (CTR, CPM, impressions) I look at only when the top four are red and I have to diagnose why.
Tools worth opening
Over the years I have tried dozens of tools. These three are the ones I really use every day:
Meta Business Suite : this is where the pixel, the catalog and Ads Manager live. If you are not here, you are not on Meta.
Meta Ads Help Center : mainly for troubleshooting, when the pixel says one thing and Ads Manager another.
Google Ads Help : especially the documentation on Smart Bidding strategies and attribution models. Required reading if you are trying to understand why PMax behaves the way it does.
Tools do not matter if the hands are not right. I have seen a client pay 400 € a month for a "premium" ad platform and still use the Boost button. And I have seen a solo founder hit a 15x ROAS with Notepad and Ads Manager.
If you feel stuck, or want someone to look at your current campaign and say what is actually broken, get in touch. The first conversation is free and I will tell you honestly whether you need an agency or just 30 minutes to fix the pixel.
