Equipo Creatify
COMPARTIR
EN ESTE ARTÍCULO
Some of you had a Meta feature literally named Automated Ads and want to know where it went. Meta retired it in October 2025 and moved everyone to Advantage+. The next section covers what changed and what to do about it.
The rest of you are asking the bigger question: how much of running ads can a machine handle now, and what's left for you? That's the rest of the piece, from automated buying through to digital ad creation.
Both questions share same backdrop. 8 million advertisers now use at least one of Meta's AI creative tools, up from 4 million at the end of 2024, according to CFO Susan Li on the Q1 2026 earnings call reported by PPC Land. Most of them are small and medium-sized businesses.
What happened to Meta's Automated Ads
Automated Ads was Meta's guided campaign builder for small businesses. You answered a few questions about your goal, and Meta put together the targeting and pulled creative from your Facebook Page. It existed so you'd never have to open Ads Manager.
Meta began phasing it out in October 2025 and moved those advertisers to Advantage+. Existing campaigns kept delivering through the transition.
Three things changed for anyone making that move.
More gets decided for you, not less. Advantage+ automates a wider set of decisions than Automated Ads did, and it keeps expanding. eMarketer reported in February 2025 that Meta is removing some manual targeting controls from Advantage+ catalog ads, keeping retargeting and custom audiences.
Source: Meta
Creative generation is a much bigger part of it. Automated Ads mostly rearranged what was already on your Page. Advantage+ sits alongside Meta's generative tools, which write copy and produce video from your assets.
Budget and objective are still yours. That hasn't moved. You set what you're optimizing toward and how much you'll spend, and everything downstream follows from those two numbers.
If Automated Ads was your whole setup, the practical version of the move is this: you now have fewer knobs and more output, so the quality of what you feed in matters more than it used to.
Automated ads, the broader meaning
Outside Meta's product name, automated ads means any system where the platform makes buying, targeting, placement or creative decisions for you. Almost every advertiser is in that category now, whether or not they opted in.
It's three separate automations, and most confusion comes from treating them as one.
Automated buying is the oldest layer: software purchasing inventory through real-time auctions instead of insertion orders and phone calls. This is programmatic, and it's been settled for over a decade.
Automated optimization is the middle layer: machine learning setting bids, pacing budgets, expanding audiences and rotating creative toward a goal you defined.
Automated ad creation is the newest and least settled: systems that write copy, generate images, and assemble video from a brief or a product URL. This is the layer most people mean when they say digital ad creation.
Knowing which layer a given feature belongs to tells you what it can and can't do for you.
What gets automated
Here's what the platforms say they're doing with automated ads, from their own documentation rather than their marketing.
Targeting and audience
Google's Performance Max runs keywordless. You supply what Google's documentation calls audience signals, described as sharing "customer intent preferences to guide learning". Signals guide the model. They don't fence it in.
TikTok's Smart+ "automatically finds the right audience with minimal input, such as location, age, and language settings", per TikTok's own product blog, though advertisers can "switch to manual targeting for full control".
Meta has gone furthest, stripping manual targeting controls out of parts of Advantage+ as covered above.
Bidding, budget and placement
Performance Max "automatically funnels your budget toward the highest performing inventory and placements" across Search, Shopping, YouTube, Display, Discover, Gmail and Maps. TikTok's Smart+ Automatic Placement spans TikTok, Pangle, Lemon8 and the Global App Bundle, "automatically allocating budget to placements where your creative performs best".
You still set the target and the ceiling. Target CPA, target ROAS, daily budget. Those numbers have never been automated away, and they remain the most consequential inputs you control.
Creative assembly and automated ad creation
Performance Max: "Generative AI tools combine your assets and creatives to help generate engaging images and videos."
Google's AI Max for Search goes further. Text customization writes ad copy from "your existing ads, landing page copy, and assets along with generative AI", and Final URL expansion picks which page on your site to send the click to.
Meta's roadmap is the most aggressive. Marketing Brew reported in April 2026 on the Andromeda retrieval system matching diverse creative to broad audiences, the Manus agent Meta acquired in December 2025 surfacing inside Ads Manager, and Zuckerberg's stated ambition that advertisers eventually supply an objective and a payment method and nothing else.
The measured gains are smaller than the ambition. On that same Q1 2026 call Meta reported a 3% conversion-rate gain from video generation in a large-scale test, more than 6% from Lattice enhancements on landing-page-view ads, and 1.6% from its Adaptive Ranking Model. Real, worth having, and a long way from transformation. They're also Meta's own numbers
This layer is worth more attention than the other two, because creative does more work than targeting does. NCSolutions' study of nearly 450 campaigns, updated in August 2023, put creative at 49% of short-term sales lift against 11% for targeting. That research covers CPG campaigns rather than digital-native ones, so treat the exact split as directional. The ordering is the point.
One limit to understand about the platform versions specifically. Their optimizers pick among the assets you handed over. They'll find the best arrangement of the territory you supplied and won't wander to a different one. Moving to a genuinely new angle is a decision that gets made outside the auction, by you or by a system built to originate rather than recombine.
Why platforms want the black box
Worth being clear-eyed about the incentives, because they explain the direction of travel better than any product announcement.
Bundling creative, bidding, placement and targeting into one campaign type gives the platform more auction liquidity and more freedom to fill long-tail inventory that advertisers would otherwise refuse. It also drops the barrier to entry, which matters when most of those 8 million advertisers are small businesses with no media buyer on staff. Those goals overlap with yours often enough to be useful and diverge often enough to need watching.
The divergence shows up in the data. PPC Land reported research putting Search Partner Network return on ad spend 37% below Google Search proper, on inventory advertisers had no way to exclude from Performance Max.
What stays yours
Production is not the boundary any more. Research, concepts, scripts and finished video all get generated end to end by tools built for it. What doesn't get automated is the set of decisions that require knowing your business.
The offer. Nothing in any ad system decides what you sell, at what price, or to whom it matters. Automation distributes a proposition it didn't author.
Brand truth. What your product does, which claims you're allowed to make, what your brand will and won't say. A system can write a hook in seconds. It can't know your compliance position or what shipped last week unless you tell it, which is why the tools worth using make you lock that in as a constraint.
What counts as a conversion. The optimization target is an input. Point a very good system at the wrong event and it will hit that event with impressive efficiency.
Your margin. Ad platforms optimize toward platform ROAS. None of them knows your gross margin by SKU, your return rate, or which customers cost you money at full price. That translation is yours.
Whether it was incremental. Platforms report credit, not causality, and each one credits itself. Holdouts and geo tests remain a human design decision.
The exclusion lists. Google's documentation lists brand exclusions, negative keywords, URL exclusions and search themes as advertiser inputs. Somebody has to decide what goes on them.
Knowing when the machine is wrong. Practitioners quoted in Marketing Brew's April 2026 report on Meta's AI push are blunt about this. Hayley Owen, SVP at Deutsch, described the ongoing work as "We're constantly having to go through and play Whac-A-Mole to figure out what's the new thing they didn't tell us about." Aaron Edwards of The Charles Group said Meta has been "keeping audiences broad, giving advertisers less control." Daniel Johnson at We Scale Startups reported that with Meta's creative AI, "We consistently see worse results" than with third-party systems. Jeremy Schulkin at Hawke Media put Advantage+ at 60% to 70% of their Meta spend while noting "it has not yet replaced a full account structure."
The line, function by function
Function | What the machine does | What stays yours |
|---|---|---|
Auction bidding | Evaluates signals and sets bids per impression | The CPA or ROAS target and the budget ceiling |
Placement | Allocates spend across inventory | Exclusion lists and brand safety boundaries |
Audience | Expands beyond your signals using conversion feedback | First-party data quality, suppression lists, who you're really for |
Asset variation | Combines, resizes, reformats, generates variants | Brand-approved source material and what counts as on-brand |
Creative concept | Platform tools recombine the assets you upload; dedicated creative systems generate concepts, scripts and finished ads | Brand truth, the claims you can stand behind, which direction to back |
Conversion definition | Optimizes toward whatever you specify | Choosing the event that predicts real value |
Measurement | Reports in-platform attribution | Incrementality design, margin maths, budget allocation |
The compliance line moved in August 2026
If you're relying on automated ad creation and you touch the EU market, something changed three weeks before you read this.
EU AI Act Article 50 transparency obligations took effect on 2 August 2026. Providers of generative systems must mark outputs as AI-generated or manipulated in a machine-readable format. Deployers must label deepfakes, and must disclose AI-generated text on matters of public interest that gets published without human review. The grace period is narrower than it's often reported: it covers only the machine-readable marking duty, running to 2 December 2026 for generative systems already on the market before August. The deepfake labelling and AI-interaction disclosure duties started on 2 August with no transition. Penalties reach 15 million euros or 3% of worldwide turnover.
The exemption worth reading twice sits inside that text obligation: it falls away where the text has had human review or editorial control and a person holds editorial responsibility for it. That obligation covers text published to inform the public on matters of public interest, so most ad copy sits outside it, and most advertisers won't be leaning on the exemption.
The marking duty on the tools applies regardless, which is the part to check with your vendors. And the underlying principle is one you probably wanted anyway: a pipeline with a named person accountable for what ships sits in a different position from one running unattended.
Automation gives control back, too
The story everyone tells is that control only disappears. The documentation says something more interesting.
Google is testing Search Partner Network and Display exclusions for Performance Max in a limited alpha, per PPC Land, with access rationed and some agencies refused. Those exclusions answer the original complaint from when the format launched.
AI Max ships with brand inclusions and exclusions, location controls, URL inclusions and exclusions, and text guidelines, which Google describes as controls that "allow you to steer AI with greater precision". Its reporting added a search terms report showing AI Max as a match type, plus headlines, URLs, and a landing pages report with a "Selected by" column. TikTok's Smart+ modules toggle on and off individually.
The shape this is settling into: a machine that executes and a human that constrains. The skill moves from operating the levers to specifying the boundaries.
One caveat on how automation arrives. Google is upgrading Dynamic Search Ads to AI Max. From September 2026 you can't create new DSA campaigns, and campaigns running automatically created assets or campaign-level broad match get upgraded then. Google has since pushed the DSA sunset and auto-upgrade itself back to February 2027, saying advertisers needed more time. Upgrading voluntarily, Google says, keeps "full control over your campaign setup". Default-on migration is the mechanism by which most automation reaches your account.
Six rules for automating ads without losing the plot
Audit what's already on. Most automation arrives switched on. Go through every campaign and write down which automated features you chose and which chose you.
Spend your hours upstream. Offer, concept, conversion definition, exclusion lists. Those are the inputs the machine amplifies, and they're where your time now earns the most.
Give it something to choose between. Optimization systems select among candidates. One creative gives them nothing to work with, and the selection advantage you're paying for goes unused.
Set guardrails, not review queues. An automated system touches thousands of placements a day. A person can review dozens. That arithmetic means control has to live in exclusion lists, budget caps, brand safety settings and audit logs, rather than in real-time approval. Match review depth to risk: a headline variant needs less scrutiny than a claim about your product.
Feed it real revenue. Send qualified, downstream revenue back to the platform rather than raw lead counts, so the optimizer learns from outcomes that matter to your business.
Re-check quarterly. Controls change in both directions and mostly without an announcement.
Where Creatify fits in digital ad creation
The ad platforms stop at recombination. They arrange the assets you hand them and won't originate anything, which is why their in-house creative AI is the piece practitioners complain about most.
Creatify runs the full cycle instead. Creatify Agent researches your brand, studies competitors and what's working in your category, writes the scripts, casts the avatars, generates the shots, scores the audio, composites the cut and runs QA against the brief before anything ships, at $5 to $8 per finished ad. URL to Video takes a product link and returns 5 to 10 script variations in under 60 seconds. Creatify Studio adds a strategist who develops personas, concepts and hooks alongside the AI and reviews every output before delivery.
So concept generation isn't the line. The line is who answers for what the concept says, which is where the guardrails earn their keep. A vision-capable critic judges every scene against the brief before it enters the cut, verified product features and claims are locked as constraints, and logos, brand colours and on-screen text get validated against confirmed brand data, with failures routed back for regeneration. AdFlow adds step-by-step workflow replay, so anyone on the team can open a workflow and see exactly how an ad was built. On the buying side, Performance Agent connects to Meta, Google, TikTok and AppLovin and audits accounts for wasted spend.
Qula360 is a useful example of the human half doing its job. The e-commerce agency had no in-house video team and was watching static image ads decay, so it designed a controlled side-by-side test for a mobile car detailing client: one static image ad, one Creatify-generated vertical video, same audience and same budget. Click-through rate went from 2.24% to 6.74%, and link clicks from 124 to 434 on less spend. The automation made the video. A person decided to run the test.
Read also: Ad analytics tools compared: dashboards, optimizers, and agents that close the loop
Frequently Asked Questions
What are automated ads?
Automated ads are campaigns where the ad platform makes some of the buying, targeting, placement or creative decisions for you. In 2026 that covers three separate things: automated buying through real-time auctions, automated optimization of bids and budgets, and automated ad creation that generates copy and video from a brief.
Did Meta discontinue Automated Ads?
Yes. Meta began phasing out its Automated Ads product in October 2025 and moved advertisers to Advantage+ campaigns, with existing campaigns continuing during the transition. Search results still surface help pages for the retired product, which is why the term causes confusion.
What is automated ad creation and how does it work?
Automated ad creation generates ad assets from inputs you supply: a brief, a product URL, brand assets, or an existing ad. Platform versions combine your uploaded assets into permutations, while dedicated tools generate scripts, voiceover and video. Both work from material and direction you provide.
What can't be automated in advertising?
Less than people assume. Research, concepts, scripts and finished video all get generated end to end now. What stays human is business judgment: what you sell and at what price, which claims you can stand behind, which event counts as a conversion, your margin maths, whether the spend was incremental, what goes on your exclusion lists, and who answers for what an ad says.
Are automated ads better than manual campaigns?
They're better at execution speed and at evaluating more combinations than a person can. They're worse when the inputs are thin, when the optimization target is poorly chosen, or when nobody's watching placement quality. The reported gains from platform AI features tend to be single-digit percentages rather than step changes.
Do I have to disclose AI-generated ads?
If you touch the EU market, EU AI Act Article 50 transparency obligations applied from 2 August 2026, requiring machine-readable marking of AI-generated output and labelling of deepfakes. The text-labelling duty is narrower than it sounds: it covers text published to inform the public on matters of public interest, and falls away where the text had human review and someone holds editorial responsibility. Check your own position with counsel rather than relying on a summary.
How do I keep control when automating ads?
Audit which automated features are switched on, set your targets and budget ceilings deliberately, maintain exclusion and suppression lists, keep a human approval gate on generated creative, measure incrementality outside the platform, and re-check the available controls every quarter.
Wrapping up
If you came for Meta's Automated Ads: it's Advantage+ now, you have fewer manual controls than before, and what you feed it matters more than it used to.
If you came for the broader question: machines take execution and, increasingly, production. Auctions, pacing, placement, asset generation, whole finished ads. What stays with you is the offer, the claims you can stand behind, the conversion definition, the margin maths, and the boundaries.
One thing worth sitting with. The advertisers who get hurt by automation are usually the ones who kept doing the job it took, hand-adjusting bids and building audience lists, rather than moving up to the job it can't do. Those upstream decisions always mattered most. Automation just removed the busywork that used to hide that fact.
Some of you had a Meta feature literally named Automated Ads and want to know where it went. Meta retired it in October 2025 and moved everyone to Advantage+. The next section covers what changed and what to do about it.
The rest of you are asking the bigger question: how much of running ads can a machine handle now, and what's left for you? That's the rest of the piece, from automated buying through to digital ad creation.
Both questions share same backdrop. 8 million advertisers now use at least one of Meta's AI creative tools, up from 4 million at the end of 2024, according to CFO Susan Li on the Q1 2026 earnings call reported by PPC Land. Most of them are small and medium-sized businesses.
What happened to Meta's Automated Ads
Automated Ads was Meta's guided campaign builder for small businesses. You answered a few questions about your goal, and Meta put together the targeting and pulled creative from your Facebook Page. It existed so you'd never have to open Ads Manager.
Meta began phasing it out in October 2025 and moved those advertisers to Advantage+. Existing campaigns kept delivering through the transition.
Three things changed for anyone making that move.
More gets decided for you, not less. Advantage+ automates a wider set of decisions than Automated Ads did, and it keeps expanding. eMarketer reported in February 2025 that Meta is removing some manual targeting controls from Advantage+ catalog ads, keeping retargeting and custom audiences.
Source: Meta
Creative generation is a much bigger part of it. Automated Ads mostly rearranged what was already on your Page. Advantage+ sits alongside Meta's generative tools, which write copy and produce video from your assets.
Budget and objective are still yours. That hasn't moved. You set what you're optimizing toward and how much you'll spend, and everything downstream follows from those two numbers.
If Automated Ads was your whole setup, the practical version of the move is this: you now have fewer knobs and more output, so the quality of what you feed in matters more than it used to.
Automated ads, the broader meaning
Outside Meta's product name, automated ads means any system where the platform makes buying, targeting, placement or creative decisions for you. Almost every advertiser is in that category now, whether or not they opted in.
It's three separate automations, and most confusion comes from treating them as one.
Automated buying is the oldest layer: software purchasing inventory through real-time auctions instead of insertion orders and phone calls. This is programmatic, and it's been settled for over a decade.
Automated optimization is the middle layer: machine learning setting bids, pacing budgets, expanding audiences and rotating creative toward a goal you defined.
Automated ad creation is the newest and least settled: systems that write copy, generate images, and assemble video from a brief or a product URL. This is the layer most people mean when they say digital ad creation.
Knowing which layer a given feature belongs to tells you what it can and can't do for you.
What gets automated
Here's what the platforms say they're doing with automated ads, from their own documentation rather than their marketing.
Targeting and audience
Google's Performance Max runs keywordless. You supply what Google's documentation calls audience signals, described as sharing "customer intent preferences to guide learning". Signals guide the model. They don't fence it in.
TikTok's Smart+ "automatically finds the right audience with minimal input, such as location, age, and language settings", per TikTok's own product blog, though advertisers can "switch to manual targeting for full control".
Meta has gone furthest, stripping manual targeting controls out of parts of Advantage+ as covered above.
Bidding, budget and placement
Performance Max "automatically funnels your budget toward the highest performing inventory and placements" across Search, Shopping, YouTube, Display, Discover, Gmail and Maps. TikTok's Smart+ Automatic Placement spans TikTok, Pangle, Lemon8 and the Global App Bundle, "automatically allocating budget to placements where your creative performs best".
You still set the target and the ceiling. Target CPA, target ROAS, daily budget. Those numbers have never been automated away, and they remain the most consequential inputs you control.
Creative assembly and automated ad creation
Performance Max: "Generative AI tools combine your assets and creatives to help generate engaging images and videos."
Google's AI Max for Search goes further. Text customization writes ad copy from "your existing ads, landing page copy, and assets along with generative AI", and Final URL expansion picks which page on your site to send the click to.
Meta's roadmap is the most aggressive. Marketing Brew reported in April 2026 on the Andromeda retrieval system matching diverse creative to broad audiences, the Manus agent Meta acquired in December 2025 surfacing inside Ads Manager, and Zuckerberg's stated ambition that advertisers eventually supply an objective and a payment method and nothing else.
The measured gains are smaller than the ambition. On that same Q1 2026 call Meta reported a 3% conversion-rate gain from video generation in a large-scale test, more than 6% from Lattice enhancements on landing-page-view ads, and 1.6% from its Adaptive Ranking Model. Real, worth having, and a long way from transformation. They're also Meta's own numbers
This layer is worth more attention than the other two, because creative does more work than targeting does. NCSolutions' study of nearly 450 campaigns, updated in August 2023, put creative at 49% of short-term sales lift against 11% for targeting. That research covers CPG campaigns rather than digital-native ones, so treat the exact split as directional. The ordering is the point.
One limit to understand about the platform versions specifically. Their optimizers pick among the assets you handed over. They'll find the best arrangement of the territory you supplied and won't wander to a different one. Moving to a genuinely new angle is a decision that gets made outside the auction, by you or by a system built to originate rather than recombine.
Why platforms want the black box
Worth being clear-eyed about the incentives, because they explain the direction of travel better than any product announcement.
Bundling creative, bidding, placement and targeting into one campaign type gives the platform more auction liquidity and more freedom to fill long-tail inventory that advertisers would otherwise refuse. It also drops the barrier to entry, which matters when most of those 8 million advertisers are small businesses with no media buyer on staff. Those goals overlap with yours often enough to be useful and diverge often enough to need watching.
The divergence shows up in the data. PPC Land reported research putting Search Partner Network return on ad spend 37% below Google Search proper, on inventory advertisers had no way to exclude from Performance Max.
What stays yours
Production is not the boundary any more. Research, concepts, scripts and finished video all get generated end to end by tools built for it. What doesn't get automated is the set of decisions that require knowing your business.
The offer. Nothing in any ad system decides what you sell, at what price, or to whom it matters. Automation distributes a proposition it didn't author.
Brand truth. What your product does, which claims you're allowed to make, what your brand will and won't say. A system can write a hook in seconds. It can't know your compliance position or what shipped last week unless you tell it, which is why the tools worth using make you lock that in as a constraint.
What counts as a conversion. The optimization target is an input. Point a very good system at the wrong event and it will hit that event with impressive efficiency.
Your margin. Ad platforms optimize toward platform ROAS. None of them knows your gross margin by SKU, your return rate, or which customers cost you money at full price. That translation is yours.
Whether it was incremental. Platforms report credit, not causality, and each one credits itself. Holdouts and geo tests remain a human design decision.
The exclusion lists. Google's documentation lists brand exclusions, negative keywords, URL exclusions and search themes as advertiser inputs. Somebody has to decide what goes on them.
Knowing when the machine is wrong. Practitioners quoted in Marketing Brew's April 2026 report on Meta's AI push are blunt about this. Hayley Owen, SVP at Deutsch, described the ongoing work as "We're constantly having to go through and play Whac-A-Mole to figure out what's the new thing they didn't tell us about." Aaron Edwards of The Charles Group said Meta has been "keeping audiences broad, giving advertisers less control." Daniel Johnson at We Scale Startups reported that with Meta's creative AI, "We consistently see worse results" than with third-party systems. Jeremy Schulkin at Hawke Media put Advantage+ at 60% to 70% of their Meta spend while noting "it has not yet replaced a full account structure."
The line, function by function
Function | What the machine does | What stays yours |
|---|---|---|
Auction bidding | Evaluates signals and sets bids per impression | The CPA or ROAS target and the budget ceiling |
Placement | Allocates spend across inventory | Exclusion lists and brand safety boundaries |
Audience | Expands beyond your signals using conversion feedback | First-party data quality, suppression lists, who you're really for |
Asset variation | Combines, resizes, reformats, generates variants | Brand-approved source material and what counts as on-brand |
Creative concept | Platform tools recombine the assets you upload; dedicated creative systems generate concepts, scripts and finished ads | Brand truth, the claims you can stand behind, which direction to back |
Conversion definition | Optimizes toward whatever you specify | Choosing the event that predicts real value |
Measurement | Reports in-platform attribution | Incrementality design, margin maths, budget allocation |
The compliance line moved in August 2026
If you're relying on automated ad creation and you touch the EU market, something changed three weeks before you read this.
EU AI Act Article 50 transparency obligations took effect on 2 August 2026. Providers of generative systems must mark outputs as AI-generated or manipulated in a machine-readable format. Deployers must label deepfakes, and must disclose AI-generated text on matters of public interest that gets published without human review. The grace period is narrower than it's often reported: it covers only the machine-readable marking duty, running to 2 December 2026 for generative systems already on the market before August. The deepfake labelling and AI-interaction disclosure duties started on 2 August with no transition. Penalties reach 15 million euros or 3% of worldwide turnover.
The exemption worth reading twice sits inside that text obligation: it falls away where the text has had human review or editorial control and a person holds editorial responsibility for it. That obligation covers text published to inform the public on matters of public interest, so most ad copy sits outside it, and most advertisers won't be leaning on the exemption.
The marking duty on the tools applies regardless, which is the part to check with your vendors. And the underlying principle is one you probably wanted anyway: a pipeline with a named person accountable for what ships sits in a different position from one running unattended.
Automation gives control back, too
The story everyone tells is that control only disappears. The documentation says something more interesting.
Google is testing Search Partner Network and Display exclusions for Performance Max in a limited alpha, per PPC Land, with access rationed and some agencies refused. Those exclusions answer the original complaint from when the format launched.
AI Max ships with brand inclusions and exclusions, location controls, URL inclusions and exclusions, and text guidelines, which Google describes as controls that "allow you to steer AI with greater precision". Its reporting added a search terms report showing AI Max as a match type, plus headlines, URLs, and a landing pages report with a "Selected by" column. TikTok's Smart+ modules toggle on and off individually.
The shape this is settling into: a machine that executes and a human that constrains. The skill moves from operating the levers to specifying the boundaries.
One caveat on how automation arrives. Google is upgrading Dynamic Search Ads to AI Max. From September 2026 you can't create new DSA campaigns, and campaigns running automatically created assets or campaign-level broad match get upgraded then. Google has since pushed the DSA sunset and auto-upgrade itself back to February 2027, saying advertisers needed more time. Upgrading voluntarily, Google says, keeps "full control over your campaign setup". Default-on migration is the mechanism by which most automation reaches your account.
Six rules for automating ads without losing the plot
Audit what's already on. Most automation arrives switched on. Go through every campaign and write down which automated features you chose and which chose you.
Spend your hours upstream. Offer, concept, conversion definition, exclusion lists. Those are the inputs the machine amplifies, and they're where your time now earns the most.
Give it something to choose between. Optimization systems select among candidates. One creative gives them nothing to work with, and the selection advantage you're paying for goes unused.
Set guardrails, not review queues. An automated system touches thousands of placements a day. A person can review dozens. That arithmetic means control has to live in exclusion lists, budget caps, brand safety settings and audit logs, rather than in real-time approval. Match review depth to risk: a headline variant needs less scrutiny than a claim about your product.
Feed it real revenue. Send qualified, downstream revenue back to the platform rather than raw lead counts, so the optimizer learns from outcomes that matter to your business.
Re-check quarterly. Controls change in both directions and mostly without an announcement.
Where Creatify fits in digital ad creation
The ad platforms stop at recombination. They arrange the assets you hand them and won't originate anything, which is why their in-house creative AI is the piece practitioners complain about most.
Creatify runs the full cycle instead. Creatify Agent researches your brand, studies competitors and what's working in your category, writes the scripts, casts the avatars, generates the shots, scores the audio, composites the cut and runs QA against the brief before anything ships, at $5 to $8 per finished ad. URL to Video takes a product link and returns 5 to 10 script variations in under 60 seconds. Creatify Studio adds a strategist who develops personas, concepts and hooks alongside the AI and reviews every output before delivery.
So concept generation isn't the line. The line is who answers for what the concept says, which is where the guardrails earn their keep. A vision-capable critic judges every scene against the brief before it enters the cut, verified product features and claims are locked as constraints, and logos, brand colours and on-screen text get validated against confirmed brand data, with failures routed back for regeneration. AdFlow adds step-by-step workflow replay, so anyone on the team can open a workflow and see exactly how an ad was built. On the buying side, Performance Agent connects to Meta, Google, TikTok and AppLovin and audits accounts for wasted spend.
Qula360 is a useful example of the human half doing its job. The e-commerce agency had no in-house video team and was watching static image ads decay, so it designed a controlled side-by-side test for a mobile car detailing client: one static image ad, one Creatify-generated vertical video, same audience and same budget. Click-through rate went from 2.24% to 6.74%, and link clicks from 124 to 434 on less spend. The automation made the video. A person decided to run the test.
Read also: Ad analytics tools compared: dashboards, optimizers, and agents that close the loop
Frequently Asked Questions
What are automated ads?
Automated ads are campaigns where the ad platform makes some of the buying, targeting, placement or creative decisions for you. In 2026 that covers three separate things: automated buying through real-time auctions, automated optimization of bids and budgets, and automated ad creation that generates copy and video from a brief.
Did Meta discontinue Automated Ads?
Yes. Meta began phasing out its Automated Ads product in October 2025 and moved advertisers to Advantage+ campaigns, with existing campaigns continuing during the transition. Search results still surface help pages for the retired product, which is why the term causes confusion.
What is automated ad creation and how does it work?
Automated ad creation generates ad assets from inputs you supply: a brief, a product URL, brand assets, or an existing ad. Platform versions combine your uploaded assets into permutations, while dedicated tools generate scripts, voiceover and video. Both work from material and direction you provide.
What can't be automated in advertising?
Less than people assume. Research, concepts, scripts and finished video all get generated end to end now. What stays human is business judgment: what you sell and at what price, which claims you can stand behind, which event counts as a conversion, your margin maths, whether the spend was incremental, what goes on your exclusion lists, and who answers for what an ad says.
Are automated ads better than manual campaigns?
They're better at execution speed and at evaluating more combinations than a person can. They're worse when the inputs are thin, when the optimization target is poorly chosen, or when nobody's watching placement quality. The reported gains from platform AI features tend to be single-digit percentages rather than step changes.
Do I have to disclose AI-generated ads?
If you touch the EU market, EU AI Act Article 50 transparency obligations applied from 2 August 2026, requiring machine-readable marking of AI-generated output and labelling of deepfakes. The text-labelling duty is narrower than it sounds: it covers text published to inform the public on matters of public interest, and falls away where the text had human review and someone holds editorial responsibility. Check your own position with counsel rather than relying on a summary.
How do I keep control when automating ads?
Audit which automated features are switched on, set your targets and budget ceilings deliberately, maintain exclusion and suppression lists, keep a human approval gate on generated creative, measure incrementality outside the platform, and re-check the available controls every quarter.
Wrapping up
If you came for Meta's Automated Ads: it's Advantage+ now, you have fewer manual controls than before, and what you feed it matters more than it used to.
If you came for the broader question: machines take execution and, increasingly, production. Auctions, pacing, placement, asset generation, whole finished ads. What stays with you is the offer, the claims you can stand behind, the conversion definition, the margin maths, and the boundaries.
One thing worth sitting with. The advertisers who get hurt by automation are usually the ones who kept doing the job it took, hand-adjusting bids and building audience lists, rather than moving up to the job it can't do. Those upstream decisions always mattered most. Automation just removed the busywork that used to hide that fact.
¿Listo para convertir tu producto en un video atractivo?
