ActiveCampaign
Marketing automation and email platform for small and mid sized businesses, founded in 2003 and headquartered in Chicago. The product combines campaign email, multi step automations, a built in sales CRM, SMS and WhatsApp messaging, and an AI layer the company markets as Active Intelligence, which generates campaigns and automations from conversational prompts and, since October 2025, is available on all plans with a remote MCP server. ActiveCampaign now describes itself as an autonomous marketing platform and states more than 1,000 app integrations. It acquired AI evaluation vendor Feedback Intelligence in February 2026.
Capability Axes
Capability grades
17 of 17 axes rated · 11 graded A or B
Active Intelligence is marketed hard and shipped fast: conversational campaign generation, an automation agent, a segment agent, autonomous insights, and a remote MCP server, with a named release cadence in the help center through 2026. It sits on a marketing automation platform that predates it by two decades.
On the removal test the AI layer comes off and a complete product remains: campaign email, automation builder, forms, landing pages, CRM pipelines and reporting all continue to work. The company positions as an autonomous marketing platform; the evidence supports AI as a fast moving interface and assistance layer over conventional automation.
The posture is documented and is mostly draft for review. Autonomous Campaign Optimisation generates improved automation flows for marketer review, Autonomous Insights surfaces weekly recommendations rather than acting on them, and the April 2026 agent to user capability initiates recommendations rather than sends. Custom Instructions let a customer constrain how the AI behaves. What is not documented is failure containment: nothing public states what stops a bad generated automation once live, or what an agent can do without a further human action.
The AI engine is named and its features are enumerated by release date, which is more than most peers do, but the stack behind it is not. No public statement identifies the model providers powering Active Intelligence, no explanation of how predictive sending or performance scoring reaches its conclusions is published, and nothing addresses whether AI generated campaign content is identified as such to the businesses sending it or the people receiving it.
Outcome material is customer story and headline figure shaped. The company cites billions of platform signals behind its performance intelligence and publishes customer results, but no measurement basis is stated anywhere public: no population, no period, no metric definitions, and no holdout. Third party review sites carry saved time and lift figures that trace back to vendor or affiliate material rather than an independent measurement.
Compliance is enforced in the product rather than delegated. The platform automatically inserts an unsubscribe link in every email, requires a legitimate physical sender address, and adds the List Unsubscribe header to all sends including automation and one to one email. The anti spam policy names prohibited content categories with no exceptions, treats high bounce rates as evidence of a permission failure, and puts accounts with abnormal unsubscribe rates under review. Help documentation addresses CAN SPAM, CASL and GDPR obligations by name. No public enforcement action located.
A real and readable privacy program: the full DPA text is public and signable from account settings, it incorporates Standard Contractual Clauses, a public subprocessor page carries the change notification obligation, and the company holds EU US Data Privacy Framework certification. The gap is the one this category turns on.
Lawful basis for the contacts a customer uploads is pushed entirely to the customer, and the multi regional hosting policy expressly disclaims any representation that regional hosting is legally sufficient for the customer's own residency obligations.
ActiveCampaign is not a data vendor: contact records are customer supplied, and the platform sells no prospect database. The sourcing standard it does impose is published, namely direct permission from the subscriber, with purchased and scraped lists treated as a policy violation and high bounce rates read as evidence of one. What is not published is the legal footing behind that standard or any indemnification position, which is what separates this from the top band.
Low exposure by construction. The platform sends from its own infrastructure and reaches third party systems through official interfaces: a documented public API, a Salesforce integration with two way contact and deal sync, LinkedIn Lead Gen and Facebook lead form connectors, and a Google Ads connector launched in July 2026 that creates Performance Max campaigns and embeds Google's own generative asset builder. No headless automation of a platform that prohibits it was located. The vendor does not address platform terms conformance anywhere directly, so the grade rests on architecture and connector documentation rather than a stated position.
Security documentation is strong and single tenancy is claimed for customer data separation, but the question this axis turns on is unanswered in public material: whether contact records, campaign content and conversation data from one account inform models that serve other accounts, and whether a customer can exclude itself.
The February 2026 acquisition of an AI evaluation vendor was described in terms of the thousands of daily agent conversations the platform generates, which raises the question rather than settling it. Retention and deletion defaults are likewise not published outside the contract.
Nothing published states whether AI generated marketing content carries any indication of its origin, and no material addresses Article 50 of the EU AI Act, in force since 2 August 2026, for the agentic and conversational surfaces the platform has been shipping since 2025. Sender identity is handled well on the mechanical side, since a legitimate physical address and accurate from line are enforced by policy, which is why this is not lower. The AI disclosure question is simply not engaged.
Depth is documented rather than logo listed. More than 1,000 app integrations, a public REST API with developer documentation, two way Salesforce sync including contact push and deal stage, ownership and activity logging, and named connector behaviour for Stripe, Calendly, Docusign, Aircall, Wix, Eventbrite and Microsoft Dynamics with the specific objects and triggers stated in help articles. A remote MCP server shipped in October 2025, which is early for this category and makes the platform addressable by outside agents.
Multi tenant cloud with a published Multi Regional Data Hosting Policy. EU data centre generally available since February 2024 and Asia Pacific since March 2024, at no extra cost, with a help article showing customers where their contact data sits.
Two real limits keep this out of the top band and both are disclosed honestly: accounts are assigned to a data centre by GEO IP and the customer cannot choose, and contact data continues to be processed in the United States and other subprocessor locations regardless of hosting region.
A live trust centre gives self service access to the current SOC 2 report, penetration test attestation, compliance certifications and legal and privacy policies, which is materially stronger than the common pattern of asserting a certification without letting an outsider see it. Security practice is documented beyond the badge: single tenancy for customer data separation, in house code and infrastructure scanning, and offensive testing. HIPAA posture is stated. The DPA also commits to providing the SOC 2 Type 2 report annually on request under NDA.
Real numbers are published and the model is legible: four email tiers from roughly 15 dollars a month at 1,000 contacts on annual billing, priced by contact tier, with send limits stated as a multiple of the contact limit at 10x, 12x and 15x by plan, and included seats stated per tier. Two things stop a buyer budgeting without a call. Enterprise is quote only, and several load bearing add ons including full CRM pipelines, sales engagement, custom reporting and AI activities sit on top of the base price with no published number.
Contact, campaign and automation data can be exported from the product, and the platform documents import and export mechanics well. What is not published is everything a buyer would want to read before signing: post termination data rights, a deletion commitment with a timeline, what happens to automations and engagement history at the end of a term, and renewal notice mechanics.
The known billing behaviour that new accounts are charged for unsubscribed and bounced contacts is the kind of term that belongs in this disclosure and is documented in help material rather than in exit terms.
Sending discipline is treated as an operating practice and documented as one. Volume governance is contractual through per plan send multiples. Reputation monitoring is continuous, with complaint data surfaced in campaign reports, bounce rate and unsubscribe rate thresholds treated as permission signals, and accounts placed under review when they drift.
Dedicated IP options are documented alongside the shared pool with guidance on when a dedicated IP is the wrong choice, and a first 30 days warmup path is published. Authentication guidance covers SPF, DKIM and DMARC on a custom sending domain, and the platform maintains ISP relationships and applies List Unsubscribe on the customer's behalf.
The segment claim is clear and consistent: small and mid sized businesses, agencies and ecommerce operators rather than enterprise revenue teams, evidenced by the pricing structure itself, which starts at one seat and caps the entry tier at 10,000 contacts. Product coverage now spans email, SMS and WhatsApp. Geographic and language specifics are the gap. EU and Asia Pacific data centres and a global customer claim imply broad reach, without a published account of supported languages or regional feature parity.
What Changed
Material product, compliance, evidence and commercial changes at ActiveCampaign, each verified against a live source and tagged to the capability axis it bears on. Funding rounds and awards are not product changes and are not logged.
Active Intelligence's Industry Intelligence feature now ends a run without publishing a card when no source is relevant to the customer's market, such as off vertical, off region, listing, directory, pricing or press release pages. ActiveCampaign states that about 55% of generated Industry Intelligence cards were previously failing its quality evaluation because the fallback path published weak cards. Brand monitoring and competitive intelligence are not affected.
ActiveCampaign shipped two Active Intelligence updates in three days. On 14 September the assistant began preserving a campaign's selected audiences and reusing existing audience IDs without overwriting their definitions, and its approval cards now keep allow and skip decisions across reloads. On 16 September it started choosing brand voice examples by email type and added a dashboard widget for lifetime goal conversions.
Pricing
What this vendor charges, what it commits to in writing, and where the bill can move. Figures the vendor publishes itself are labeled Vendor Published. Figures labeled Estimated come from other sources and the vendor has not confirmed them.
- ›The company publishes no price a machine can read. Its pricing page is over a megabyte of content with all four plan names on it and not a single currency figure anywhere. The numbers only appear after you tell it how many contacts you have.
- ›Because of that, the outside guides disagree badly. The cheapest plan is reported at $15, $16 and $19 by different sources, and the top plan at 50,000 contacts is reported at $1,059, $1,169 and $1,459. None of them is being careless.
- ›There is no single price to quote anyway. What you pay depends on two things at once, which plan you pick and how many contacts you store, with about eleven contact bands from 1,000 up to 50,000.
- ›One thing to check before you sign, because it changes what you are billed for. Accounts opened before 3 November 2025 are reported to pay only for contacts they actually market to. Newer accounts are reported to be billed for unsubscribed contacts too.
- ›The extras are reported to be missing from the pricing page entirely: a pipeline module around $68 a month, custom reporting around $159, extra users around $12 each, and texting credits from about $16.83 that expire each month.
How the price works
What you are charged for, and what makes the bill go up.
Two dimensional pricing across four tiers, where the tier sets the feature ceiling and the stored contact count sets the bill. Nothing is retrievable from the vendor. The served pricing page carries the four tier names, a contact count selector and no currency figure of any kind, with prices fetched at runtime once a contact volume is chosen.
Every figure that follows is a third party estimate and the sources disagree. The entry tier is variously reported at $15, $16 and $19 per month at 1,000 contacts. The second tier at $42, $49 and $59. The third at $79 and $99. The fourth at $145, $179, and by some sources as quote only. The annual discount is reported as 15 percent by some analysts and 20 percent by others, applied to the monthly rate.
Contact bands are reported at roughly eleven steps covering 1,000, 2,500, 5,000, 7,500, 10,000, 15,000, 20,000, 25,000, 30,000, 40,000 and 50,000, with custom pricing above that and the entry tier reported as unavailable at the higher bands. Reported prices at 10,000 contacts on annual billing diverge between sources: the third tier at $375 or $419, the fourth at $589 or $665. At 50,000 contacts the top tier is reported at $1,059, $1,169 and $1,459.
Sending allowances are reported as a multiple of the contact limit, at 10, 12 and 15 times depending on tier, and included user counts are reported at one on the entry tier, three on the third and five on the fourth.
A contact counting change is reported with a date: accounts created before 3 November 2025 billed only on contacts actively marketed to, accounts created after that date billed on unsubscribed contacts as well. Since the meter is the contact count, this alters the definition of the billed unit rather than the rate.
One analysis additionally reports the product being sold in three structures covering marketing, sales and a bundle, each with the same four tier ranks, which if accurate multiplies the number of distinct published prices well beyond four.
What the contract says about your data
What the vendor commits to in writing once your data is in the product.
Substantially better than the pricing disclosure, and reachable in advance. A legal index is published carrying the material a buyer's counsel needs rather than a summary of it: a data processing addendum the customer can sign, standard contractual clauses offered as a separate signable transfer mechanism covering the European Economic Area, the United Kingdom and Switzerland, and an enumerated sub processor list published as its own document rather than referenced.
Offering the clauses as a distinct instrument the customer elects to sign, rather than folding them silently into an addendum, is the more useful arrangement of the two and relatively few vendors on this roster do it.
The privacy policy is structured by jurisdiction rather than written to one regime and footnoted for the rest, with distinct sections for the European Economic Area, the United Kingdom and Switzerland, California, Australia, Brazil and Costa Rica. For a platform sold internationally to marketers who will be processing their own customers' data, that structure is materially more usable than a single policy with a compliance appendix. Client self service rights are stated: account information can be updated, corrected or deleted from within the product or by written request, and a connected mail account can be disconnected and its data deleted by the customer directly.
What was not established: a named security certification or attestation with its scope and assessor, a retention period expressed as a duration, and a residency or hosting region statement. Those were not located from the legal index or the privacy policy.
The custody question is shaped by the product's reach. This platform holds the customer's marketing database, which at the upper contact bands runs to tens of thousands of records about people who are not the customer's users, together with site tracking data, messaging history across email and other channels, and on the sales side the pipeline. The contact counting policy described below also means the platform retains and bills for records the customer has stopped contacting.
Getting started
What it costs and what is included before the product is running.
Not published, and not retrievable. No setup fee, onboarding charge, migration rate or professional services rate appears in anything served by the pricing page, which serves no figures at all.
The trial is described consistently across independent sources as fourteen days carrying the third tier's feature set, limited to one user, 100 contacts and 100 sends, with no card required. There is no permanent free plan. A further reduction of 20 percent for non profit organizations is reported, applying across plans.
The costs that sit outside the plan are, according to multiple independent analyses, absent from the pricing page altogether. Reported figures: a pipeline module near $68 monthly, custom reporting near $159 monthly, additional users near $12 each monthly, and messaging sold as credits from roughly $16.83 monthly for an entry allocation, rising to a reported $176.83 monthly at 25,000 credits. Messaging is reported as unavailable on the entry tier and its credits are reported to expire monthly with no rollover. A per message rate near $0.025 for United States destinations is reported, varying by country. None of those figures was confirmed on a vendor surface reached here and every one of them is a third party estimate.
One further product line is reported separately: a transactional sending service starting near $15 monthly for 10,000 messages, which is a distinct subscription rather than an add on to a plan.
A buyer should expect the real first year figure to sit materially above any headline, and should obtain the add on rates in writing, since the vendor does not publish them anywhere a reader or a machine can reach.
What to watch for
Where this pricing can surprise a buyer who has not read it closely.
The most complete machine invisibility recorded in this index so far, and a second analyst has independently documented it.
The served pricing page is 1.23 megabytes. It contains all four plan names, the phrase per month eight times and the word contacts sixty nine times. It contains no currency symbol, no price figure, no structured data offer, no embedded price object and no machine readable alternative at any conventional address. Every figure is fetched at runtime by a configurator that first asks the buyer how many contacts they hold. Independently, a pricing analyst reviewing the same page in mid 2026 stated plainly that it renders its prices dynamically and that their published tables were therefore third party corroboration rather than quotations from the vendor. So the invisibility is not an artifact of my method.
The consequence is the widest third party spread this index has recorded. The entry tier is reported at $15, $16 and $19. The second tier at $42, $49 and $59. The third at $79 and $99. The fourth at $145, $179 and as quote only. The annual discount is reported as 15 percent by some analysts and 20 percent by others. At 10,000 contacts, two guides give the third tier as $375 and $419 and the fourth as $589 and $665. At 50,000 contacts the top tier is reported at $1,059, at $1,169 and at $1,459. These are not careless writers. They are people transcribing a configurator at different moments with different inputs selected.
The deeper problem is that no single headline figure can be correct here, because price is two dimensional. A tier sets the feature ceiling and a contact band sets the bill, with bands reported at roughly eleven steps from 1,000 to 50,000 and custom pricing above that. A buyer comparing an entry figure against a competitor's flat rate is not comparing like with like, and the vendor's own structure makes that unavoidable rather than accidental.
One material term deserves to be on the record because it changes the bill without the buyer doing anything, and it carries a date. Accounts created before 3 November 2025 are reported to pay only for contacts they actively market to. Accounts created after that date are reported to be billed for unsubscribed contacts as well. Since the whole meter is contact count, that is a change in the definition of the metered unit rather than a change in rate, and it does not appear on the pricing page.
The add ons are absent from the pricing page entirely according to multiple independent sources, and they are not small: a pipeline module reported near $68 monthly, custom reporting reported near $159 monthly, additional users reported near $12 each monthly, and messaging sold in credits from roughly $16.83 monthly with credits reported to expire monthly without rollover. One analysis models a buyer at 10,000 contacts sending two messaging campaigns a month needing a 25,000 credit tier at $176.83 monthly, which it calculates as adding 36 percent to the plan cost at that contact level.
What is disclosed clearly, and it is worth crediting, is the trial and the concession. Fourteen days with the third tier's features, one user, 100 contacts and 100 sends, no card required, and a stated further 20 percent reduction for non profit organizations.
No dollar figure is recorded in the numeric field. Nothing was retrieved from a vendor surface, the third party figures disagree with each other at every tier, and the metered unit is a contact band rather than a seat, so no single figure would describe a purchase.