A few years ago, winning at paid media meant segmenting well, adjusting bids and fighting for every keyword. Today Google and Meta automate almost all of that. The question is no longer how to control the machine, but what you feed it. And the answer has two parts: creatives and data.
The changes reshaping campaigns
Google Ads
- AI Max for Search campaigns. It expands the searches you show up for, adapts ad copy and picks the landing page. Google reports 7% more conversions on average at a similar CPA.
- Goodbye to Dynamic Search Ads. DSA campaigns are being migrated automatically to AI Max from September 2026, and Google plans to retire them completely in February 2027.
- A more transparent Performance Max. It now shows how much each channel contributes (Search, YouTube, Display, Discover, Maps) and allows structured experiments with creative assets.
- Display moves into Demand Gen. Standalone Display campaigns are being folded into Demand Gen, which is more focused on visual creative and video.
- AI creative inside the platform. Tools like Asset Studio generate images and videos from your brand guidelines, and a Gemini-powered assistant helps build and review campaigns.
Meta Ads
- Andromeda. The new engine that decides which ads each person can see. It puts the emphasis on creative relevance rather than manual targeting.
- Advantage+ by default. Sales, app and lead campaigns now start from the automated setup.
- Conversions API is practically mandatory. The browser pixel captures fewer and fewer conversions, so sending data from your server or CRM is what lets the system optimise properly.
The conclusion is the same on both platforms: the algorithm does the targeting, and what guides it is your creatives and your conversion signals.
Why refreshing creatives is no longer optional
With systems like Andromeda, creative works as targeting: each piece finds its audience. If you only have one «hero» ad, the algorithm has very little to work with and the campaign burns out sooner.
Creative fatigue shows in clear signals: frequency goes up, CTR drops, CPM gets more expensive and cost per result worsens without you changing anything. By the time you see it, you’re already late.
What works:
- Real variety, not copies. Ten versions of the same ad in a different colour aren’t diversity. Change the angle (price, speed, trust), the format (video, carousel, static, UGC), the hook in the first seconds and the person you’re speaking to.
- A refresh calendar. Don’t wait for performance to drop: plan new pieces every few weeks depending on your spend.
- Learn from what wins. Analyse which messages and formats work and produce more variations of those, not of everything.
- Use AI to produce, not to think. Generative tools speed up versions and format adaptations, but the idea and the message are still brand work.
Connect your CRM: optimise for sales, not forms
This is where the most money is lost. If you tell Google or Meta that a conversion is «someone filled in a form», the algorithm will bring you lots of forms. Some will be good customers; others, contacts who will never buy. The platform can’t tell them apart unless you tell it.
Connecting the CRM changes that. When a lead moves through your pipeline (qualified, opportunity, customer), that data goes back to the ad platform and the algorithm learns to look for more people like the ones who actually buy.
- Google Ads: enhanced conversions for leads and offline conversion import, to send your CRM stages as conversions with their value.
- Meta: Conversions API, including its CRM version, to optimise lead campaigns for quality leads.
- HubSpot, Salesforce and other CRMs have native integrations or connectors that sync these stages without custom development.
- Audiences from the CRM: exclude current customers, build lookalikes of your best customers and re-engage leads that stalled halfway.
I’ve seen it first-hand: when launching an insurance product from scratch, setting up the CRM with lead scoring first and connecting it to sales operations is what made it possible to scale acquisition without costs spiralling. Here’s the case.
A checklist for your campaigns
- Do your campaigns optimise for a business conversion (qualified lead, sale, value) and not just form submissions?
- Do you have Conversions API and enhanced conversions active and verified?
- Does your CRM report back to the platforms which leads move forward and which don’t?
- Do you have at least 4 or 5 genuinely different creative concepts per campaign?
- Is there a calendar for renewing creatives before fatigue sets in?
- Have you reviewed your dynamic search campaigns before the migration to AI Max?
If the answer to several of these is «no», you’re probably paying for the algorithm to learn things you don’t care about. If you’ve already connected your CRM to your ads, I’d love to hear how it went: get in touch.
Sources: Google: DSA upgrades to AI Max · Google Ads updates 2026 · Meta Ads updates 2026
