AdPlug vs Markifact: a Markifact alternative for PPC specialists and agencies

AdPlug vs Markifact: who should pick which

Both put your ad accounts in Claude and ChatGPT, and both can change campaigns as well as read them. The difference is what you're buying, and who it's built for.

Markifact is a marketing automation suite: no-code workflows, its own agents, and an MCP connector on top, all paid for in credits. Its homepage says "AI Automation. Built for Marketers.", and its GitHub page promises to turn "any AI client into a senior performance marketer". To founders running their own ads, it puts it more bluntly: "You don't need a media buyer."

AdPlug is built for the media buyer: PPC specialists, consultants and agencies who know what good looks like and want AI to do the legwork. It plugs your accounts into the AI you already use, every call stays yours, and it costs a lot less.

And if all you're after is a client dashboard, you don't need to pay either of them for it. AdPlug's free plan reads your accounts, and Claude or ChatGPT builds the dashboard from what it reads, laid out the way you want.

The real difference: what "approval" means

Markifact says every change waits for you. If you manage other people's money, check where that wait happens, because Markifact's own pages don't agree.

How Markifact makes changes

On its marketing pages the answer is simple: "Every write operation requires explicit user approval before it executes." One of its blog posts goes further and describes "an approval queue that is independent of the MCP client".

Its docs tell a different story. In the table that compares its workflows, agents and MCP, the approval model for MCP reads "Client-controlled (approve writes per your client's settings)", and approval before writes "Depends on your client's tool settings". Its MCP page says the AI client "is instructed to describe exactly what will change and ask for your confirmation first", and the security policy in its GitHub repo says the write protocol "is enforced by the agent prompt". Its integrations page, for its part, says every write comes "always with an optional approval step".

I couldn't find a docs page that describes a hold on Markifact's own server for a change coming from Claude or ChatGPT. So there, the yes happens in the conversation, and Markifact's docs ask you to keep its write tool on Needs approval: "Do not set run_write_operation to ‘Always allow.’" Good advice. It also tells you whose job the setting is.

Inside Markifact's own app it depends on how you set it up. Its agents have three approval modes, and one of them is "Full access: auto-approve write operations without asking, for trusted use cases", which works on scheduled tasks too. The approval model its docs give for workflows is "Runs automatically once triggered", and templates such as Pause Overspending Meta Ads Campaigns do just that on a schedule.

The terms are clear about whose problem a wrong number is. An approval prompt "is your final opportunity to review before execution; once you approve, the change is made and any resulting spend is your responsibility." And Markifact is "not liable for losses caused by incorrect unit entry".

How AdPlug makes changes

AdPlug has no agents or workflows of its own, on purpose. Claude or ChatGPT does the thinking, and AdPlug is the plug socket between it and your accounts. On Google Ads a change takes two separate tool calls, and the server won't run the second without the first.

1. You ask: "Find the search terms that convert but aren't keywords yet, and add the best ones."
2. The AI reads the search terms report and works out which terms deserve a keyword of their own.
3. It calls AdPlug's preview tool. The server checks the change and stores it exactly as it will run, under an ID: which keywords, which match type, which ad groups. Nothing has changed in Google Ads yet.
4. You read it. Wrong ad group, wrong match type? Say so, and it never runs.
5. Only after your yes can the execute tool run, and only on that stored preview. It runs once and expires after 15 minutes, so last week's yes can't turn into this week's change.
6. AdPlug logs the call, and the change shows in Google Ads' own change history.

Claude

Find the search terms that convert but aren't keywords yet, and add the best ones as exact match.

✓google_ads_preview_mutationAdPlug - Google Ads

Preview · Acme Software · add 3 exact match keywords

Keyword	Match type	Ad group	Conversions, 30 days
crm for consultants	Exact	Generic · CRM	11
simple crm for agencies	Exact	Generic · CRM	8
crm with email tracking	Exact	Features	6

These three search terms brought 25 conversions in the last 30 days through broader keywords, at well under your average cost per conversion. As exact match keywords you can bid on them directly. Nothing changes until you say so. Shall I add all three?

Add them.

✓google_ads_execute_mutationAdPlug - Google Ads

Done. Three exact match keywords added, and Google Ads' change history shows them.

The AI can't skip that step, because the rule lives on AdPlug's server, not in a prompt or a client setting someone can change. HubSpot and Search Console changes work the same way. On the other ad platforms the preview tool is there too, but it isn't forced yet, so ask for it in your prompt. And AdPlug has no scheduler of its own, so nothing runs unless you, or a task you set up in your AI, start it.

Both approaches stop most accidents. Where they differ is the bad day: a long session, a vague prompt, an AI that's sure it already asked. On that day I want the check in the server, not in the model's memory. If a client or a finance director reads your change history, I suspect you do too.

What Markifact users say in their reviews

I built a rival, so don't take Markifact's faults from me. Here's what its users say, read on 25 Oct 2026. There isn't much of it.

Nobody has given Markifact fewer than 3.5 stars anywhere, which sounds better than it is: almost nobody has reviewed it at all, and four of its five G2 reviews carry G2's Incentivized badge, which means a small reward was offered for writing them. Three of those went up on the same day. Treat each voice below as a signal, not a verdict.

What users complain about

Credits to keep an eye on. The meter is the first thing a careful reviewer mentions.

Alerts that didn't fire. This reviewer uses Markifact for a wireless carrier's projects rather than ads, so weigh it accordingly.

Basic reports, missing connectors. Even a five-star reviewer wanted more.

Nobody publicly reports failed writes, wrong numbers or billing trouble. Nobody complained in public about four price rises either, which says more about how few people write reviews than about the prices.

What users like

The praise is real, and the one G2 review that wasn't incentivised is the warmest.

If you do try it, check its work the way you'd check a new hire's: Google Ads' change history for what changed, and your own conversion columns for what it reports. That goes for any tool, mine included.

AdPlug vs Markifact, side by side

Checked on each company's own pages on 25 Oct 2026. Edge marks the side that wins a row, only where that's a plain fact rather than my opinion.

AdPlug and Markifact compared: platforms, Google Ads tools, AI clients, setup, how changes happen, read-only access, logs, price, usage, refunds and team features, checked 25 Oct 2026

Compared	AdPlug	Markifact
Where it works
Platforms	AdPlugGoogle Ads, Meta Ads, LinkedIn Ads, Microsoft Ads, Reddit Ads and TikTok Ads, plus Google Search Console and HubSpot on connectors of their own.	Markifact14 ad platforms, all read and write except Apple Ads: Meta, Google, TikTok, Microsoft, LinkedIn, Reddit, Snapchat, Pinterest, Amazon, Spotify, X, DV360, ChatGPT's ads and Apple Ads. 33 other tools, from GA4 and HubSpot to Slack.
Google Ads tools	AdPlug17 tools, and 174 named kinds of change, each checked in a preview.	Markifact60+ Google Ads operations behind seven to ten general tools. The AI searches a catalogue of 1000+ operations, then runs the one it needs.
AI clients	AdPlugClaude, ChatGPT, Claude Code, Codex, Grok Bot, Hermes Agent and any AI agent that accepts a remote MCP address.	MarkifactClaude, ChatGPT, Cursor, Codex, Gemini CLI, Windsurf, Hermes Agent and others, plus its own app, Slack and WhatsApp.
Setup	AdPlugSign up, sign in through each platform's own screen, paste one address into your AI. Nothing to host.	MarkifactPaste its address into your AI, sign in, connect each platform. Nothing to host.
Built for	AdPlugPPC specialists, consultants and agencies working in Claude, ChatGPT, Grok Bot, Hermes Agent or their own AI agents.	MarkifactIn its own words, “AI Automation. Built for Marketers.”, with a promise to turn “any AI client into a senior performance marketer”.
Sources: Markifact's homepage, integrations page, docs and GitHub repo. Checked 25 Oct 2026.
How changes happen
Who does the work	AdPlugThe AI you already use, when you ask or on a schedule you set in it. AdPlug never acts on its own.	MarkifactYour AI through its connector, plus Markifact's own agents and no-code workflows, which can run on a schedule.
Approving a change	AdPlugGoogle Ads: the server won't run a change without a stored preview of that exact change. Other ad platforms: a preview when you ask.	MarkifactFrom Claude or ChatGPT, its docs call approval “Client-controlled”: your AI is instructed to ask. Its agents can be set to “Full access” and approve their own changes.
Read-only option	AdPlugThe free plan is view only, and any teammate can be set to View only on each platform.	MarkifactA Read only setting for each MCP server since 15 October 2026, plus limits on which operations and connections a server can use.
Log of calls	AdPlugEvery tool call logged, with your latest calls in the dashboard and each teammate's usage for the team owner.	MarkifactExecution logs with each operation's inputs and outputs, kept 7 days on Pro and 30 on Team.
Sources: Markifact's docs, changelog, terms and GitHub security policy. Checked 25 Oct 2026.
What it costs
Starts at	AdPlugFree, then Pro at $29 a monthEdge	MarkifactTrial credits, then Pro at $79 a month
Usage	AdPlugPro: 5,000 requests and 200 edits a month. Max: unlimited.	MarkifactPro: 2,500 credits a month, or 5,000 for $149. A report pull or a change costs about a credit, and agents bill by tokens.
A team of three	AdPlugMax at $89: unlimited requests and editsEdge	MarkifactTeam at $249: 10,000 credits and ten seats. Pro has one seat.
Free plan	AdPlugView only, one platform, unlimited ad accounts, 500 requests in the first 5 days, then 50 a month	MarkifactTrial credits only, 200 of them by its blog, with no card
Refunds	AdPlugCancel anytime, 14-day refund windowEdge	MarkifactA “Strict No Refund Policy”, even if you forget to cancel before a renewal
Sources: Markifact's pricing page, credits docs and terms, and AdPlug's pricing. Checked 25 Oct 2026.
Teams and agencies
Ad accounts	AdPlugUnlimited on every plan, the free one included.	MarkifactUnlimited on Pro and Team.
Manager accounts	AdPlugOne sign-in covers every account under an MCC. Three Google logins on Pro, five on Max and Agency.	MarkifactMCCs supported, with as many connections per platform as you like.
Access per user	AdPlugView only or Can edit for each teammate on each platform, enforced on every request.Edge	MarkifactTwo roles, team owner and team member. Read-only is set per MCP server, not per person and platform.
Team members	AdPlugThree on Max, then priced per seat on Agency.	MarkifactOne on Pro, then 10 to 50 on Team, tied to the credit tier.
CRM and pipeline	AdPlugHubSpot from Pro, with one connector reaching every client portal: three on Pro, ten on Max and Agency.	MarkifactHubSpot, read and write, plus Klaviyo to read.
Sources: Markifact's pricing page, connections docs and logs docs. Checked 25 Oct 2026.

Where Markifact is stronger

Credit where it's due. There are two good reasons to pick it, and one I'd call a draw.

• Platforms AdPlug doesn't connect. Markifact reads and changes Snapchat, Pinterest, Amazon, Spotify, DV360 and ChatGPT's ads, and reads Apple Ads, with GA4, Tag Manager, Merchant Center, Shopify and Klaviyo alongside. If those are a real part of your work, that settles it.
• It suits founders who'd rather not hire a media buyer. Its GitHub page tells solo founders "You don't need a media buyer." and promises "Pro-level decisions without learning the platform UI." If that's you, its agents and templates are made for it.
• A canvas for workflows. Markifact lets you drag boxes onto a canvas to build workflows that run on a schedule or a trigger, like its own example of a Monday report built into Slides and posted to Slack. You can get the same kind of thing with AdPlug through Claude's scheduled tasks or a Claude Code routine, and there's nothing to drag: you describe the workflow in a sentence and Claude builds it. The canvas only helps if you'd rather see every step laid out, so I'd call this one a draw.

Why PPC specialists and agencies pick AdPlug

Markifact is built for marketers in general. It charges a credit for each job, leaves the yes in Claude or ChatGPT to your AI's settings, and gives a team two roles. If PPC is your job, here's what AdPlug does instead.

If PPC is your job

• The AI you already use, with what it already knows. The work happens in Claude, ChatGPT, Grok Bot, Hermes Agent or your own agents, next to your notes, your briefs and your other tools.
• Requests you don't have to ration. Pro is $29 a month for 5,000 requests and 200 edits, and Max is unlimited. A follow-up question costs you nothing extra, so you can ask it.
• Scheduled work, no canvas. Claude's scheduled tasks and Claude Code's routines run AdPlug on a timetable: a Monday report, a daily pacing check. You describe it in a sentence and Claude builds it. I'd keep the changes for when you're there to say yes.
• Changes with names. AdPlug knows 174 kinds of Google Ads change, from negatives and budgets to responsive search ads, experiments and Performance Max asset groups, and every one of them goes through the same preview.
• Every platform you run. Meta Ads, LinkedIn Ads, Microsoft Ads, Reddit Ads and TikTok Ads sit in the same plan as Google Ads, with Google Search Console and HubSpot on connectors of their own.
• Read-only until you say otherwise. The free plan can't change anything, which is how I'd start with any new tool.

Report and optimise on pipeline, not just conversions

For B2B accounts this is the report that matters. A Google Ads conversion is a form fill. Pipeline is what the client is paying for.

AdPlug connects HubSpot next to Google Ads, so the same chat can follow each campaign from the click to the deal: the contacts it brought, how far they got through the lifecycle stages, and the pipeline they made. Then it can move budget towards the campaigns that make pipeline, with a preview before anything changes.

HubSpot comes with Pro, and one HubSpot connector reaches every client portal you connect: three on Pro, ten on Max and Agency.

Claude

How far do the contacts from each Google Ads campaign get for Acme Software? I want lifecycle stages, not form fills.

✓hubspot_reportAdPlug - HubSpot

Acme Software · paid search contacts by lifecycle stage

Campaign	Leads	MQLs	SQLs	Customers
Search · Generic	204	61	17	4
Search · Competitors	58	29	12	5
Search · Brand	139	22	5	1

Competitors turns about one lead in five into an SQL and has made the most customers, five from 58 leads. Brand brings plenty of leads, but few get past MQL. Want me to pull the spend from Google Ads and suggest budget moves as a preview?

If you run an agency

• One flat price for the team. AdPlug Max is $89 a month for three team members, with unlimited requests and edits. On Markifact, Pro has one seat, so a team of three starts on Team at $249 a month, with 10,000 credits to share.
• Access for every user, platform by platform. Each teammate gets View only or Can edit on each platform. The junior can read every client and change none of them, the senior can edit Google Ads and only read LinkedIn, and AdPlug enforces it on every request. Markifact has two roles, team owner and team member, and sets read-only per MCP server.
• Every client account, behind one sign-in. AdPlug works on unlimited ad accounts on every plan, the free one included. One Google sign-in covers every account under your manager account, and from Pro you can add Google logins for clients under other MCCs: three on Pro, five on Max and Agency.
• A record of the work. AdPlug logs every tool call, your dashboard shows your latest ones, and a team owner sees each teammate's usage platform by platform. Google Ads' change history fills in the rest.

Markifact pricing

Markifact charges in credits. Pro is $79 a month for 2,500 credits and one seat, or $149 for 5,000. Team runs from $249 a month for 10,000 credits and ten seats up to $999 for 100,000 credits and 50 seats. Paying yearly gives you two months free, and new accounts start with trial credits and no card. Checked on its pricing page on 25 Oct 2026.

What a credit buys depends on where you use it. From Claude or ChatGPT, finding operations is free and each data operation, a report pull or a change, costs about one credit. A workflow run costs a credit plus its steps, and agents bill by tokens, at up to 20 times the base rate on the priciest models. You can buy extra credits, at a price its pages don't publish.

Like for like, that's a lot more than AdPlug. Markifact's Pro is $79 for 2,500 credits, and AdPlug Pro is $29 for 5,000 requests. For a team of three, Markifact's cheapest option is Team at $249, and AdPlug Max is $89 with no meter at all.

It used to be cheaper, too. In March 2025 Markifact's entry plan cost $29 a month. By November it was $79, by way of $39 and $59, and the team plan went from $129 to $249 by January 2026. None of those changes is in its changelog. I found them in archived copies of its own pricing page.

Two lines I'd read twice. Its terms have a "Strict No Refund Policy", and it applies "even if you forget to cancel before renewal, cancel immediately after renewal, or do not use the Service during a billing cycle." Unused credits don't roll over either. AdPlug has a 14-day refund window, and the free plan never asks for a card.

Markifact alternatives, by the job you need done

The best alternative depends on why you looked at Markifact in the first place. I've compared the Google Ads connectors in more depth in my Google Ads MCP server comparison.

Switching from Markifact to AdPlug

It takes about ten minutes. Before you start, open Google Ads' change history (and Meta's activity log, if you used it there) to see what was changed through Markifact, and download any execution logs you want to keep: they expire after 7 days on Pro and 30 on Team.

1. Switch off its workflows, scheduled agent tasks and anything running on its API first, so nothing changes while you move.
2. Cancel in your Markifact account settings before the renewal date. Once a renewal has gone through there's no refund, and you have 30 days after cancelling to export your data.
3. Remove its connector from Claude, ChatGPT or Cursor and revoke any manual tokens, then remove its access in your Google account under Security, Third-party connections, and in each other ad platform's list of connected apps.
4. Sign up for AdPlug and connect your ad platforms through each one's own sign-in screen. Every account behind that login comes with it, manager accounts included.
5. Add AdPlug's address to your AI as a custom connector. The guides walk through it with screenshots for Claude, Grok Bot and ChatGPT dots.
6. Start with a question you can check, like last month's spend by campaign. Once the numbers match the platform, move on to the real work.

Who should stay with Markifact

• You've built your reports on its workflow canvas and they run the way you want.
• Snapchat, Pinterest, Amazon, Spotify or DV360 is a real part of your spend.
• You're a founder or a generalist marketer who'd rather not hire a media buyer.
• It's working, your numbers are up, and its changes match your change history. Don't switch for the sake of it.

P.S. Markifact's docs say "Workflows are the recipe, Agents are the chef, MCP is the takeout window." It's a good line. I'd just want to check the order before it leaves the kitchen.