Sixty new creatives in a Drive folder. A sheet with the copy, the landing pages, and the UTMs. Three ad accounts that all need the same batch by Monday. And Ads Manager, where each of those ads still gets built one at a time, in each of the three accounts.
If that week sounds familiar, this piece is for you. We compared nine tools that launch ads in bulk on at least one of Meta, TikTok, Snapchat, or Google from a sheet, a Drive folder, or a feed, ours included and held to the same lines as everyone else. We also spent a section on the thing the existing lists skip: launches that happen without a person pressing Launch, on a schedule, from new sheet rows, from a watched folder, or from a feed.
How we compared
We make one of these tools, so the method matters more than usual. Every entry is built from three layers: what the vendor says on its own site and in its documentation; what users say on G2, Capterra, TrustRadius, and Trustpilot; and the gap between the two. Scope and limits come only from documentation, never from a single review. We didn't run launches through the other eight tools, and we don't claim to have.
The review layer is thin in this category. Over the past 12 months Birch, Smartly, AdManage, and Hunch each have under ten reviews (Birch at 3.8★); Madgicx has around 40 (1.4★). Adnova, Ads Uploader, and Blip have none we can place in the period, and Kitchn's 20 on G2 carry no dates. Where reviews are absent the entry says so and the gap layer is empty. Review counts were pulled Oct. 27, 2026; everything else below is as of Oct. 30, 2026.
One observation we can make with its basis: we read the 22 bulk-launch lists that AI answers cite today, and none of the 22 covers launches that run without a person pressing Launch, so we added that in as a separate section.
What "bulk launch" has to mean before you compare tools
- Sources. Where the creatives and the copy come from: a sheet, a Drive or Dropbox folder, a product feed, an existing Post ID.
- Launch platforms. The platforms ads are created on.
- Auto-launch. Whether anything launches without a person pressing the button, and what starts it: a schedule, new sheet rows, a watched folder, a feed sync. Then the approval model: per run, per ad, or none.
- After launch. Whether the tool manages what it launched, with rules, and on which platforms.
- AI. Whether it suggests, writes, or acts, and on which platforms.
- Agent access. Whether an agent outside the tool can work through it: an MCP server or an API, read-only or acting, and on which platforms.
- Pricing model. Per upload, per ad account, flat, or a percentage of spend. The model changes less often than the figures and tells you more about the bill.
The free option comes before any of these. Meta Ads Manager duplicates campaigns, ad sets, and ads in bulk and imports new ads from a spreadsheet; Google Ads Editor is a free desktop app that makes bulk changes offline and imports files before you post them. Each stops at its own platform, and neither reads a planning sheet you already keep or launches anything on a schedule. If your week is one platform, one account, and one batch, you may not need software yet. What happens to those ads after launch is a separate question, and we've written about it for Meta and for Google.
The nine at a glance, in alphabetical order
What they cost
Published figures as of Oct. 30, 2026, each linked to the vendor’s page.
The tools
AdManage
Platforms: 15 channels, listed on every plan: Meta, TikTok, Google Ads, Snapchat, Pinterest, Reddit, X, LinkedIn, AppLovin, Taboola, ChatGPT Ads, Mintegral, Moloco, Unity Ads, and Apple Ads. The entry plan includes one ad account; more cost about $133 a month each.
Sources: Google Drive, Dropbox, Frame.io, Box, SharePoint, Canva, Figma, and a dozen more cloud and asset tools; Google Sheets through an add-on and CSV import for row-based batches; Post IDs for relaunching existing posts and creator ads; a built-in media library.
Launch: creatives are grouped and named by filename, with placement sizes paired into one ad, and go out to every channel from one flow. The Sheets add-on maps your columns, generates templates, and pushes campaigns from inside the spreadsheet; an Ad Status column sets each ad active or paused on creation. Meta and TikTok have dedicated Sheets templates.
Auto-launch: a batch can be prepared now and launched at a later time. New files in a connected Drive or Dropbox folder show up ready to launch, and automations can run on triggers from Drive files, sheet changes, or ad performance. The approval model is the paused batch: launch everything paused, check it in Ads Manager, then switch it on.
After launch: rules on Meta that pause losers, scale winners, and hide comments on a schedule. The entry plan includes one monitoring or comments rule; performance rules start on the mid-tier plan with 10 rules and are unlimited on the agency tier.
AI: an assistant that researches ads, creates variations, prepares drafts, and launches; AI tags on every creative; AI naming from the mid-tier plan; image and video generation on monthly credits.
Agent access: an MCP server for Claude and ChatGPT on every plan, which launches and reports with drafts reviewed before publishing, and a documented public REST API on every plan: launch ads on Meta, TikTok, Snapchat, Pinterest, Reddit, X, Google, LinkedIn, and the rest, poll batch status, and pull reports.
Best for: teams for whom the launch is the job, across many channels and markets. The cross-channel part is the launch; the rules afterwards run on Meta. Agencies sizing a plan by ad accounts rather than by spend.
Adnova
Platforms: Meta.
Sources: Google Drive and Dropbox, with files pulled in without downloading and re-uploading; existing ads and organic posts.
Launch: set defaults and optimization rules once, then launch 100 or more ads in one click.
Auto-launch: none.
After launch: the launch defaults only.
AI: Adnova's AI sits in its other products, sold separately: creative analytics, search over a 75-million-ad library, competitor reports, and asset tagging. None of it acts in the ad account.
Agent access: none.
Best for: Meta-only teams that already run creative research in Adnova and want the launcher next to it.
Ads Uploader
Platforms: Meta.
Sources: Google Drive import, direct upload in bulk, public links, and Post IDs for duplicating existing ads; an XLSX configuration for launch settings, and an XLSX export for Ads Manager if you'd rather launch there.
Launch: bulk through the Marketing API, with size variations and thumbnails matched automatically, ad sets grouped, and names generated from a pattern. Ads can be published paused, and ad sets take start and end dates.
Auto-launch: nothing recurring is built in. The start date on an ad set is the only scheduling in the product. If you're comfortable with a command line, the CLI lets a script or an AI agent run a whole launch, and your own scheduler can repeat it.
After launch: none.
AI: none in the product itself.
Agent access: an MCP server, hosted or local, for Claude, ChatGPT, Cursor, and Claude Code: list ad accounts, browse campaigns and creatives, upload media, preview exactly what would be created, create ads from a preset or a saved build, and duplicate by Post ID. The agent can run the launch end to end, or you review the preview and press Create Ads yourself. MCP access is for paid plans only, not during the trial.
Best for: solo buyers and small Meta teams who launch from the terminal or from an agent and don't need rules afterwards.
Birch
Platforms: Meta, TikTok, and Snapchat.
Sources: Google Sheets and Google Drive folders shared view-only, with subfolders synced; links to files or whole folders without a sheet. One sheet connects to one platform.
Launch: Stage reads your sheet. Each row can set its own ad account and template ad, so one sheet launches into many ad accounts in a single run, or every row goes to one destination with only the media link required. A validation preview flags bad cells before anything is created. New ads are built from a template ad set and ad, and launch active, paused, or with a start date. Files whose names differ only by a placement suffix are grouped into one ad. Formats cover image, video, carousel, lead, and partnership ads, by platform.
Auto-launch: a Routine runs on a schedule you set, hourly, daily, on weekdays, or weekly. The approval mode is Manual: each sheet waits for a one-click approval per run.
After launch: automated rules on Meta, Google, TikTok, and Snapchat, built the same way on each, with your own numbers from a connected attribution tool or a Google Sheet as inputs.
AI: Birch AI reads performance across the four platforms and creates and edits rules on all four from a plain-language description; Routines run any of those prompts on the same schedules.
Agent access: Birch MCP connects Claude, ChatGPT, Codex, or any other MCP client. It reads Stage tasks with their uploads and created ads, reads rules and their logs, and creates and edits rules on all four platforms. What to ask once it's connected.
Best for: teams and agencies that plan launches in a sheet they already keep and launch into many ad accounts on Meta, TikTok, or Snapchat, then manage the ads with rules.
Blip
Platforms: Meta.
Sources: Google Drive, Dropbox, and desktop uploads; for launches from chat, a CSV and a public Drive folder.
Launch: reusable templates and saved settings, Post ID scaling for whitelisted and partnership content, and assets grouped by aspect ratio into one ad.
Auto-launch: a launch started from chat can be scheduled or set live. No watched folder, sheet trigger, or recurring launch.
After launch: recommendations rather than rules: suggestions to scale, reduce, or pause a campaign or ad set, flags on ads that have run 14 or more days with high spend and a below-average CPA, and anomaly alerts on CPA spikes and fast overspend. You apply a recommendation in one click or skip it.
AI: the recommendations and anomaly detection above, and placement grouping.
Agent access: launch ads, check analytics, and apply budget recommendations from Claude and ChatGPT.
Best for: Meta teams launching a high volume on a flat price who want pause-and-scale suggestions next to the launch rather than a rules engine.
Hunch
Platforms: Meta and Snapchat for direct launch.
Sources: Google Sheets, Shopify and other product feeds, and catalogs; the feed can carry live pricing and discounts into the creative.
Launch: feed-based. A product catalog becomes thousands of localized image and video variants, built in bulk from one sheet, and published to Meta and Snapchat.
Auto-launch: an automated campaign is set up once and then creates ad sets and ads from the feed. When an item newly matches the filter, an ad is created; when an item's values move it to another ad set, the ad is turned off and recreated there; when an item leaves the filter, its ad is archived and a fresh one is created if it returns. Scheduled rules add a second trigger that fires on a fixed schedule regardless of performance.
After launch: rules on Meta: performance triggers on metric conditions or scheduled triggers, actions limited to a status change or a budget adjustment, checks from every 30 minutes to monthly.
AI: background removal and color matching for product images, animated catalog product videos, product sets built from GA4 and conversion data, and creative insights.
Agent access: none.
Best for: e-commerce teams with a large catalog and a creative bottleneck, running paid social, whose ads should follow the feed rather than a launch calendar.
Kitchn
Platforms: Meta, TikTok, Google Ads for Demand Gen, Pinterest, and Snapchat; X connects for data only. Meta is the most complete channel by Kitchn's own description.
Sources: files from the desktop, pasted Drive or Dropbox links, and Frame.io; a bulk import from a Drive folder that matches assets by filename, so a 1:1 and a 9:16 of the same creative land in one row; headlines and primary text pasted from Google Sheets. CSV import and a Google Sheets add-on are Enterprise-only, and Drive is their only asset source.
Launch: a Launch Sheet where each row is one ad, up to 300 ads a sheet, edited like a grid and checked by a QA engine for typos, asset specs, and the 86 creative enhancements Ads Manager may switch on by default. Reusable templates on every plan; custom naming from the mid-tier plan.
Auto-launch: start and stop dates per ad, set row by row or in bulk, on Meta's Sales and App Promotion objectives. The mid-tier and Enterprise plans list "automated folder scanning and asset import" and "fully automated ad launches."
After launch: the pre-launch checks above; rule-based campaign changes and ad duplication are an Enterprise feature.
AI: none named on the product or pricing pages.
Agent access: API access on Enterprise; no MCP server.
Best for: paid social teams launching high volumes who want every ad checked before it goes out and don't automate anything after the launch. Plans are sized by accounts and uploads a month.
Madgicx
Platforms: Meta.
Sources: uploads, and variations the AI ad generator produces from one product image or a proven ad. No sheet, folder, or feed source.
Launch: an ad launch tool deploys creatives across campaigns in bulk.
Auto-launch: no sheet, folder, or feed trigger. What Madgicx has instead is the AI Marketer, which you set to autonomous execution or to a human-approved workflow; with approvals, you review each recommendation and its prepared action before it goes live.
After launch: rules, AI bidding, and budget shifts on Meta.
AI: the AI Marketer above, which audits the account and prepares or executes actions, and a Copilot chat.
Agent access: an MCP server for Claude, ChatGPT, and other clients that pauses underperformers and scales winners on Meta from the conversation. Included on paid plans.
Best for: Meta-only accounts that want the AI to run the account, with launching as one of its jobs rather than the job.
Smartly
Platforms: Meta, Google (Demand Gen, App, and Video campaigns, no Search), TikTok, Snapchat, Pinterest, Reddit, Amazon DSP for connected TV, and the major DSPs.
Sources: a data sheet, a feed, or a launch document; Google Sheets as a feed; product catalogs.
Launch: templated creative generated and localized at scale, then bought and reported in the same system.
Auto-launch: Automated Ads build, revise, and archive ads as the source changes, from a data sheet, a feed, or a launch document.
After launch: optimization triggers and predictive budget allocation across channels.
AI: Creative Predictive Potential, which scores creative before launch, and Creative Insights after it.
Agent access: a reporting API with account-level tokens; no MCP server.
Best for: brands producing localized creative variants at scale across markets, with a procurement process and an ad-ops team to run the platform.
The agent-first launchers
Three more tools build the launch for you from a prompt rather than from a sheet, and they are reviewed too thinly to compare on the lines above. AdAmigo runs Meta and Google from a chat agent that can bulk-launch whole campaigns; nothing goes live without your approval by default, with an auto-apply setting on the $349 plan. AdStellar builds a Meta campaign from your past performance and launches variants in bulk on a credit-based $100-a-month plan. Scalemate launches to Meta and TikTok from templates with hourly rules afterwards, on a free plan of one account and 100 launches a month.
Launches that run without you
The creative team fills the sheet by Thursday, and the launch runs Friday morning. New sheet rows suit a sheet the team keeps filling all week. Whoever adds a row shouldn't have to find whoever launches. A watched folder suits a creative team that drops files and keeps no sheet. A feed suits a catalog. The ad exists as long as the product does and goes away when it doesn't.
The tools differ most on approval. AdManage launches a batch paused and leaves the switch to you in Ads Manager. Birch runs the sheet on the schedule and waits for a one-click approval per sheet per run. Smartly and Hunch run on the feed. Madgicx and AdAmigo put the choice in a setting: autonomous or approved.
One phrase to read carefully: "automated launch." In a vendor's copy it often means ads created paused for a person to switch on.
Three questions settle it on the demo. What starts the run? What happens to a row or a file that failed last time? Does anything go live while you're asleep?
If the job is launching from a sheet you already keep, on a schedule, and then managing those ads with rules on Meta, Google, TikTok, and Snapchat, that's what we built Birch for. For a wider comparison of what each of these tools does after the launch, across platforms, see the cross-channel comparison.
Questions to ask on the demo
The questions that recur on the demo and onboarding calls we run when launching comes up, roughly in the order of how often.
- Can I launch from the Drive folder and the sheet we already keep, with the copy, URLs, and CTAs in the columns we use, or do I rebuild the sheet in your format?
- Can one run go into several ad accounts, markets, or languages, with the destination set per row?
- Can I add new creatives into an existing ad set without a new setup, so the learning stays?
- Can I duplicate a proven ad set with new creative, or launch one ad set per creative for a clean test, from the same sheet?
- How many ads can one run create, and how is the plan sized: by rows, by uploads, by accounts, or by spend?
- Can several sizes of one creative become one ad, and can naming and UTMs follow a pattern per row?
- What goes live without a person approving it, and can I start with paused ads or notifications only?
- Once the ads are live, does the same tool pause losers and scale winners, and on which platforms?
Sixty new creatives in a Drive folder. A sheet with the copy, the landing pages, and the UTMs. Three ad accounts that all need the same batch by Monday. And Ads Manager, where each of those ads still gets built one at a time, in each of the three accounts.
If that week sounds familiar, this piece is for you. We compared nine tools that launch ads in bulk on at least one of Meta, TikTok, Snapchat, or Google from a sheet, a Drive folder, or a feed, ours included and held to the same lines as everyone else. We also spent a section on the thing the existing lists skip: launches that happen without a person pressing Launch, on a schedule, from new sheet rows, from a watched folder, or from a feed.
How we compared
We make one of these tools, so the method matters more than usual. Every entry is built from three layers: what the vendor says on its own site and in its documentation; what users say on G2, Capterra, TrustRadius, and Trustpilot; and the gap between the two. Scope and limits come only from documentation, never from a single review. We didn't run launches through the other eight tools, and we don't claim to have.
The review layer is thin in this category. Over the past 12 months Birch, Smartly, AdManage, and Hunch each have under ten reviews (Birch at 3.8★); Madgicx has around 40 (1.4★). Adnova, Ads Uploader, and Blip have none we can place in the period, and Kitchn's 20 on G2 carry no dates. Where reviews are absent the entry says so and the gap layer is empty. Review counts were pulled Oct. 27, 2026; everything else below is as of Oct. 30, 2026.
One observation we can make with its basis: we read the 22 bulk-launch lists that AI answers cite today, and none of the 22 covers launches that run without a person pressing Launch, so we added that in as a separate section.
What "bulk launch" has to mean before you compare tools
- Sources. Where the creatives and the copy come from: a sheet, a Drive or Dropbox folder, a product feed, an existing Post ID.
- Launch platforms. The platforms ads are created on.
- Auto-launch. Whether anything launches without a person pressing the button, and what starts it: a schedule, new sheet rows, a watched folder, a feed sync. Then the approval model: per run, per ad, or none.
- After launch. Whether the tool manages what it launched, with rules, and on which platforms.
- AI. Whether it suggests, writes, or acts, and on which platforms.
- Agent access. Whether an agent outside the tool can work through it: an MCP server or an API, read-only or acting, and on which platforms.
- Pricing model. Per upload, per ad account, flat, or a percentage of spend. The model changes less often than the figures and tells you more about the bill.
The free option comes before any of these. Meta Ads Manager duplicates campaigns, ad sets, and ads in bulk and imports new ads from a spreadsheet; Google Ads Editor is a free desktop app that makes bulk changes offline and imports files before you post them. Each stops at its own platform, and neither reads a planning sheet you already keep or launches anything on a schedule. If your week is one platform, one account, and one batch, you may not need software yet. What happens to those ads after launch is a separate question, and we've written about it for Meta and for Google.
The nine at a glance, in alphabetical order
What they cost
Published figures as of Oct. 30, 2026, each linked to the vendor’s page.
The tools
AdManage
Platforms: 15 channels, listed on every plan: Meta, TikTok, Google Ads, Snapchat, Pinterest, Reddit, X, LinkedIn, AppLovin, Taboola, ChatGPT Ads, Mintegral, Moloco, Unity Ads, and Apple Ads. The entry plan includes one ad account; more cost about $133 a month each.
Sources: Google Drive, Dropbox, Frame.io, Box, SharePoint, Canva, Figma, and a dozen more cloud and asset tools; Google Sheets through an add-on and CSV import for row-based batches; Post IDs for relaunching existing posts and creator ads; a built-in media library.
Launch: creatives are grouped and named by filename, with placement sizes paired into one ad, and go out to every channel from one flow. The Sheets add-on maps your columns, generates templates, and pushes campaigns from inside the spreadsheet; an Ad Status column sets each ad active or paused on creation. Meta and TikTok have dedicated Sheets templates.
Auto-launch: a batch can be prepared now and launched at a later time. New files in a connected Drive or Dropbox folder show up ready to launch, and automations can run on triggers from Drive files, sheet changes, or ad performance. The approval model is the paused batch: launch everything paused, check it in Ads Manager, then switch it on.
After launch: rules on Meta that pause losers, scale winners, and hide comments on a schedule. The entry plan includes one monitoring or comments rule; performance rules start on the mid-tier plan with 10 rules and are unlimited on the agency tier.
AI: an assistant that researches ads, creates variations, prepares drafts, and launches; AI tags on every creative; AI naming from the mid-tier plan; image and video generation on monthly credits.
Agent access: an MCP server for Claude and ChatGPT on every plan, which launches and reports with drafts reviewed before publishing, and a documented public REST API on every plan: launch ads on Meta, TikTok, Snapchat, Pinterest, Reddit, X, Google, LinkedIn, and the rest, poll batch status, and pull reports.
Best for: teams for whom the launch is the job, across many channels and markets. The cross-channel part is the launch; the rules afterwards run on Meta. Agencies sizing a plan by ad accounts rather than by spend.
Adnova
Platforms: Meta.
Sources: Google Drive and Dropbox, with files pulled in without downloading and re-uploading; existing ads and organic posts.
Launch: set defaults and optimization rules once, then launch 100 or more ads in one click.
Auto-launch: none.
After launch: the launch defaults only.
AI: Adnova's AI sits in its other products, sold separately: creative analytics, search over a 75-million-ad library, competitor reports, and asset tagging. None of it acts in the ad account.
Agent access: none.
Best for: Meta-only teams that already run creative research in Adnova and want the launcher next to it.
Ads Uploader
Platforms: Meta.
Sources: Google Drive import, direct upload in bulk, public links, and Post IDs for duplicating existing ads; an XLSX configuration for launch settings, and an XLSX export for Ads Manager if you'd rather launch there.
Launch: bulk through the Marketing API, with size variations and thumbnails matched automatically, ad sets grouped, and names generated from a pattern. Ads can be published paused, and ad sets take start and end dates.
Auto-launch: nothing recurring is built in. The start date on an ad set is the only scheduling in the product. If you're comfortable with a command line, the CLI lets a script or an AI agent run a whole launch, and your own scheduler can repeat it.
After launch: none.
AI: none in the product itself.
Agent access: an MCP server, hosted or local, for Claude, ChatGPT, Cursor, and Claude Code: list ad accounts, browse campaigns and creatives, upload media, preview exactly what would be created, create ads from a preset or a saved build, and duplicate by Post ID. The agent can run the launch end to end, or you review the preview and press Create Ads yourself. MCP access is for paid plans only, not during the trial.
Best for: solo buyers and small Meta teams who launch from the terminal or from an agent and don't need rules afterwards.
Birch
Platforms: Meta, TikTok, and Snapchat.
Sources: Google Sheets and Google Drive folders shared view-only, with subfolders synced; links to files or whole folders without a sheet. One sheet connects to one platform.
Launch: Stage reads your sheet. Each row can set its own ad account and template ad, so one sheet launches into many ad accounts in a single run, or every row goes to one destination with only the media link required. A validation preview flags bad cells before anything is created. New ads are built from a template ad set and ad, and launch active, paused, or with a start date. Files whose names differ only by a placement suffix are grouped into one ad. Formats cover image, video, carousel, lead, and partnership ads, by platform.
Auto-launch: a Routine runs on a schedule you set, hourly, daily, on weekdays, or weekly. The approval mode is Manual: each sheet waits for a one-click approval per run.
After launch: automated rules on Meta, Google, TikTok, and Snapchat, built the same way on each, with your own numbers from a connected attribution tool or a Google Sheet as inputs.
AI: Birch AI reads performance across the four platforms and creates and edits rules on all four from a plain-language description; Routines run any of those prompts on the same schedules.
Agent access: Birch MCP connects Claude, ChatGPT, Codex, or any other MCP client. It reads Stage tasks with their uploads and created ads, reads rules and their logs, and creates and edits rules on all four platforms. What to ask once it's connected.
Best for: teams and agencies that plan launches in a sheet they already keep and launch into many ad accounts on Meta, TikTok, or Snapchat, then manage the ads with rules.
Blip
Platforms: Meta.
Sources: Google Drive, Dropbox, and desktop uploads; for launches from chat, a CSV and a public Drive folder.
Launch: reusable templates and saved settings, Post ID scaling for whitelisted and partnership content, and assets grouped by aspect ratio into one ad.
Auto-launch: a launch started from chat can be scheduled or set live. No watched folder, sheet trigger, or recurring launch.
After launch: recommendations rather than rules: suggestions to scale, reduce, or pause a campaign or ad set, flags on ads that have run 14 or more days with high spend and a below-average CPA, and anomaly alerts on CPA spikes and fast overspend. You apply a recommendation in one click or skip it.
AI: the recommendations and anomaly detection above, and placement grouping.
Agent access: launch ads, check analytics, and apply budget recommendations from Claude and ChatGPT.
Best for: Meta teams launching a high volume on a flat price who want pause-and-scale suggestions next to the launch rather than a rules engine.
Hunch
Platforms: Meta and Snapchat for direct launch.
Sources: Google Sheets, Shopify and other product feeds, and catalogs; the feed can carry live pricing and discounts into the creative.
Launch: feed-based. A product catalog becomes thousands of localized image and video variants, built in bulk from one sheet, and published to Meta and Snapchat.
Auto-launch: an automated campaign is set up once and then creates ad sets and ads from the feed. When an item newly matches the filter, an ad is created; when an item's values move it to another ad set, the ad is turned off and recreated there; when an item leaves the filter, its ad is archived and a fresh one is created if it returns. Scheduled rules add a second trigger that fires on a fixed schedule regardless of performance.
After launch: rules on Meta: performance triggers on metric conditions or scheduled triggers, actions limited to a status change or a budget adjustment, checks from every 30 minutes to monthly.
AI: background removal and color matching for product images, animated catalog product videos, product sets built from GA4 and conversion data, and creative insights.
Agent access: none.
Best for: e-commerce teams with a large catalog and a creative bottleneck, running paid social, whose ads should follow the feed rather than a launch calendar.
Kitchn
Platforms: Meta, TikTok, Google Ads for Demand Gen, Pinterest, and Snapchat; X connects for data only. Meta is the most complete channel by Kitchn's own description.
Sources: files from the desktop, pasted Drive or Dropbox links, and Frame.io; a bulk import from a Drive folder that matches assets by filename, so a 1:1 and a 9:16 of the same creative land in one row; headlines and primary text pasted from Google Sheets. CSV import and a Google Sheets add-on are Enterprise-only, and Drive is their only asset source.
Launch: a Launch Sheet where each row is one ad, up to 300 ads a sheet, edited like a grid and checked by a QA engine for typos, asset specs, and the 86 creative enhancements Ads Manager may switch on by default. Reusable templates on every plan; custom naming from the mid-tier plan.
Auto-launch: start and stop dates per ad, set row by row or in bulk, on Meta's Sales and App Promotion objectives. The mid-tier and Enterprise plans list "automated folder scanning and asset import" and "fully automated ad launches."
After launch: the pre-launch checks above; rule-based campaign changes and ad duplication are an Enterprise feature.
AI: none named on the product or pricing pages.
Agent access: API access on Enterprise; no MCP server.
Best for: paid social teams launching high volumes who want every ad checked before it goes out and don't automate anything after the launch. Plans are sized by accounts and uploads a month.
Madgicx
Platforms: Meta.
Sources: uploads, and variations the AI ad generator produces from one product image or a proven ad. No sheet, folder, or feed source.
Launch: an ad launch tool deploys creatives across campaigns in bulk.
Auto-launch: no sheet, folder, or feed trigger. What Madgicx has instead is the AI Marketer, which you set to autonomous execution or to a human-approved workflow; with approvals, you review each recommendation and its prepared action before it goes live.
After launch: rules, AI bidding, and budget shifts on Meta.
AI: the AI Marketer above, which audits the account and prepares or executes actions, and a Copilot chat.
Agent access: an MCP server for Claude, ChatGPT, and other clients that pauses underperformers and scales winners on Meta from the conversation. Included on paid plans.
Best for: Meta-only accounts that want the AI to run the account, with launching as one of its jobs rather than the job.
Smartly
Platforms: Meta, Google (Demand Gen, App, and Video campaigns, no Search), TikTok, Snapchat, Pinterest, Reddit, Amazon DSP for connected TV, and the major DSPs.
Sources: a data sheet, a feed, or a launch document; Google Sheets as a feed; product catalogs.
Launch: templated creative generated and localized at scale, then bought and reported in the same system.
Auto-launch: Automated Ads build, revise, and archive ads as the source changes, from a data sheet, a feed, or a launch document.
After launch: optimization triggers and predictive budget allocation across channels.
AI: Creative Predictive Potential, which scores creative before launch, and Creative Insights after it.
Agent access: a reporting API with account-level tokens; no MCP server.
Best for: brands producing localized creative variants at scale across markets, with a procurement process and an ad-ops team to run the platform.
The agent-first launchers
Three more tools build the launch for you from a prompt rather than from a sheet, and they are reviewed too thinly to compare on the lines above. AdAmigo runs Meta and Google from a chat agent that can bulk-launch whole campaigns; nothing goes live without your approval by default, with an auto-apply setting on the $349 plan. AdStellar builds a Meta campaign from your past performance and launches variants in bulk on a credit-based $100-a-month plan. Scalemate launches to Meta and TikTok from templates with hourly rules afterwards, on a free plan of one account and 100 launches a month.
Launches that run without you
The creative team fills the sheet by Thursday, and the launch runs Friday morning. New sheet rows suit a sheet the team keeps filling all week. Whoever adds a row shouldn't have to find whoever launches. A watched folder suits a creative team that drops files and keeps no sheet. A feed suits a catalog. The ad exists as long as the product does and goes away when it doesn't.
The tools differ most on approval. AdManage launches a batch paused and leaves the switch to you in Ads Manager. Birch runs the sheet on the schedule and waits for a one-click approval per sheet per run. Smartly and Hunch run on the feed. Madgicx and AdAmigo put the choice in a setting: autonomous or approved.
One phrase to read carefully: "automated launch." In a vendor's copy it often means ads created paused for a person to switch on.
Three questions settle it on the demo. What starts the run? What happens to a row or a file that failed last time? Does anything go live while you're asleep?
If the job is launching from a sheet you already keep, on a schedule, and then managing those ads with rules on Meta, Google, TikTok, and Snapchat, that's what we built Birch for. For a wider comparison of what each of these tools does after the launch, across platforms, see the cross-channel comparison.
Questions to ask on the demo
The questions that recur on the demo and onboarding calls we run when launching comes up, roughly in the order of how often.
- Can I launch from the Drive folder and the sheet we already keep, with the copy, URLs, and CTAs in the columns we use, or do I rebuild the sheet in your format?
- Can one run go into several ad accounts, markets, or languages, with the destination set per row?
- Can I add new creatives into an existing ad set without a new setup, so the learning stays?
- Can I duplicate a proven ad set with new creative, or launch one ad set per creative for a clean test, from the same sheet?
- How many ads can one run create, and how is the plan sized: by rows, by uploads, by accounts, or by spend?
- Can several sizes of one creative become one ad, and can naming and UTMs follow a pattern per row?
- What goes live without a person approving it, and can I start with paused ads or notifications only?
- Once the ads are live, does the same tool pause losers and scale winners, and on which platforms?
