How Marketing Automation Works, and What It Costs
How a campaign actually works, the condition behavior that silently breaks a first build, and what one published contract says the software costs.
A first automation can fail without looking like it has failed. The workflow runs, the report shows contacts moving through it, and nobody receives the message you built it to send. One cause sits in a piece of logic that almost no guide explains: Mautic's campaign builder documentation states that the software evaluates a condition immediately, and that the contact does not wait for the condition to become true. Marketing automation is easier to buy than to run, and the distance between the two is made of details like that one.
What Marketing Automation Is
Marketing automation is software that carries out marketing steps on its own, triggered by something a contact does or by something already true about them. You build the sequence once, and the software runs it for every contact who enters, one at a time, for as long as it stays switched on.
That definition is smaller than the category makes it sound, and the work it describes is ordinary, made of steps a person could do by hand, such as these. Send this email. Wait four days. Check whether the person opened it. Add five points to their score if they did. Tell the salesperson who owns them once the score passes a line. None of those steps needs software. Doing them for two thousand people, in the right order, on the day each person qualifies, does.
The word carrying the most weight there is trigger. A campaign is not a schedule, and it is not a broadcast. It is a set of rules that fire per contact, on that contact's own clock, which is why two people who signed up an hour apart can sit at completely different points in the same sequence.
Marketing Automation, Email Marketing and CRM Are Three Different Things
These three are sold as one product and do three separate jobs. An email platform sends messages to a list. A customer relationship management system (CRM) stores what you know about a person and what has happened with them. Marketing automation is the logic in between: it reads the record, decides what should happen next, and acts.
| The tool | What it owns | What it cannot do alone |
|---|---|---|
| Email platform | The list, the message, the send, the unsubscribe | Decide who should get the next message, and when |
| CRM | The record of a person, and everything that happened with them | Act on that record without somebody opening it |
| Marketing automation | The rules that read the record and then act | Hold data nobody put there |
One product often covers two of these, or all three, which is where the confusion starts. The useful move before buying anything is to name which of the three jobs you have a gap in. A team that already sends good email and still cannot tell who is worth calling has a logic problem, not a sending problem.
How Marketing Automation Actually Works
Marketing automation works by turning a sequence into a diagram the software executes one contact at a time. Every platform draws that diagram in its own way, and underneath they share a grammar. Mautic, the open-source engine that Acquia sells commercially as Campaign Studio, publishes its version of that grammar in full, which makes it the clearest thing to learn from. The vocabulary matches what the paid platforms use, and it is written down instead of demonstrated in a sales video.
A campaign in that model has a source, which decides who comes in, and then three kinds of event you place on the canvas: actions, decisions and conditions. Get the difference between decisions and conditions wrong and the campaign will run perfectly while doing nothing you intended.
The Source: Who Enters the Campaign
Nothing happens in a campaign until a contact enters it, and the source is the door. Mautic offers two: a segment, which is a saved group of contacts who share something, and a form, which admits a contact at the moment they submit it. Both can feed the same campaign.
That choice matters more than it looks. A form source admits people one at a time as they arrive, which suits a sequence about something the person did moments ago. A segment source admits everyone who matches. The first time you switch on a campaign built over an existing segment, the whole segment enters at once, and every one of them takes the first step on the same afternoon.
The documentation adds one restriction worth knowing before you design around it. Only public segments appear in the campaign builder, so a segment marked private cannot start a campaign at all.
Decisions Are What the Contact Does. Conditions Are What You Already Know.
A decision waits for the contact to do something. A condition tests something the record already holds. Mautic's own wording is exact: decisions are "actions that your Contacts initiate", while conditions "execute different actions based on a Contact's data".
The decisions on offer are the things a person can do to you and your site: Opens Email, Clicks Email, Replies to Email, Submits Form, Visits a page, Downloads Asset, Device visit, Request Dynamic Content. The three email ones appear only once the campaign contains a Send Email action, because until then there is nothing to open.
The conditions are the things you already know: Contact field value, Contact Points, Contact Segments, Contact Tags, Contact Campaigns, Contact owner, Contact device, Form Field value, Has valid Email address, Has active notification.
Both draw two branches on the canvas, and they look identical when you are dragging boxes around. That shared shape is why they get mixed up, and the two behave nothing alike once a contact reaches them.
The Green Path and the Red Path
Every decision and every condition splits in two, drawn green for yes and red for no. The green path belongs to the contact who did the thing or matched the test. The red path is the non-action path, and it belongs to the contact who did not.
The green path runs on the contact's own timing. Mautic's documentation says execution of the actions following a green path happens at the time the contact takes the action, or is scheduled immediately when the following action carries a delay.
The red path has no clock of its own, which is the part people miss. Its timing comes from the delay set on the action attached to it. The documentation puts it plainly: the delay becomes how long the contact has to take the action before the campaign moves them down the non-action path. Attach a reminder with no delay to a "did not open" branch, and you have told the software that the contact had no time at all in which to open it.
A Condition Does Not Wait
This is the behavior that catches people, and Mautic writes it in plain words: "Mautic evaluates Campaign conditions immediately", and "The Contact doesn't wait for the condition to be TRUE."
Say you build a birthday campaign. Your whole list is the source, a condition on the date field is the first event, and the birthday email hangs off the green path. On the day you switch it on, every contact is evaluated once, at that moment. The few whose birthday is today get the email. Everyone else fails the test, goes down the red path, and stays there. The campaign looks alive on the dashboard and does nothing for the rest of the year.
The documented fix moves the test out of the campaign and into the source. Build a segment whose filter is the date field equal to today, and start the campaign from that segment. Membership changes daily, contacts move in and out, and the campaign fires as they do. The condition can then be deleted, because the segment is doing its job.
Two limits travel with that fix. The anniversary option accepts only a day and a month, and the segment trick does not work for it. And a contact passes through a campaign once: if the date on their record changes later and they land back in the segment, they will not run through it a second time.
There is one exception to the immediacy, and it reads like a contradiction until you look at what it applies to. Mautic respects a delay set on the condition itself before passing down to a delay on any connected action, so a condition can be scheduled for later. What it cannot do is sit and wait for its own test to turn true.
Walk one contact through the diagram
Assemble the first two blocks of a campaign and see where a contact actually ends up. The two combinations that look hardest to tell apart on the canvas behave nothing alike.
1. What admits the contact?
2. What is the next block?
Where the contact ends up, with a condition as the second block
- EntryEveryone in the segment is admitted the moment you switch the campaign on.
- ImmediatelyThe condition is evaluated on the spot, against whatever the record holds at that second.
- Green pathThe contacts whose record already matches take the affirmative branch, and the action after it runs.
- Red pathEveryone else takes the non-action branch in the same instant. The contact does not wait for the condition to become true.
- TomorrowThe record changes and the test would now pass. Nothing happens: that contact has already been evaluated, and a contact passes through a campaign once.
This is the build that silently does nothing. Switch this on over an existing segment and the whole list is sorted into two piles within a second, permanently. If you built it expecting the campaign to catch people as they qualify, it never will.
The fix. Move the test out of the campaign and into the source. Build a segment whose filter is the thing you were testing for, and start the campaign from that segment. Membership changes as records change, contacts enter as they qualify, and the condition can be deleted.
Block names and both timing behaviors come from Mautic’s campaign builder documentation.
What Marketing Automation Can Do
Most of what a marketing automation platform does is not sending anything, and the fastest way to see that is to read the list of actions it can take. Of the actions Mautic's campaign builder documents, three put a message in front of anyone: Send Email, Send Marketing Message, and Send Email to User, which goes to somebody other than the contact, such as the person who owns their record. Everything else changes data, membership or ownership.
| What the action touches | The documented actions |
|---|---|
| Sends a message | Send Email, Send Marketing Message, Send Email to User |
| Scoring | Adjust Contact Points, Add to Company's score |
| Grouping | Modify Contact's Segments, Modify Contact's tags, Change Contact's Stage, Change Campaigns |
| Records and ownership | Update Contact, Update Contact owner, Add Company action, Update Contact's primary Company, Delete Contact |
| Permission | Add Do Not Contact, Remove Do Not Contact |
| Other systems | Push Contact to Integration, Send a Webhook |
| Flow control | Jump to Event |
That distribution answers a question the category never says out loud. A marketing automation platform is not an email tool with a timer bolted on. It is a system for changing what you know about a person and who is responsible for them, with sending as one of the things it can do about that.
Lead Scoring and Routing
Lead scoring is a number on the contact record that moves as the person behaves. Mautic implements it as an action, Adjust Contact Points, which adds or subtracts points from the contact's total or group, and as a condition, Contact Points, which checks whether the contact has reached a threshold. There is a company-level version, Add to Company's score, which adjusts the score for every company linked to the contact.
Routing is what the score is for. Update Contact owner assigns the person to a named user, and Contact owner as a condition lets a later branch check who that is. Between them they describe the whole handoff: watch, score, and when the score crosses a line, move the record to a human and tell them why.
Two things reliably break scoring. Points awarded for behavior that predicts nothing, an email open being the usual offender, since the reader may not have opened it on purpose. And a threshold set once and never compared against what happened next. A score is a hypothesis about who is ready to buy, and the only evidence it is a good one is the salesperson agreeing with it.
Personalization Past the First Name
Personalization inside a campaign happens two ways, and only one of them is a merge tag. The first is a token. Mautic's documentation shows the shape in its webhook action, where {contactfield=firstname} is replaced with the value on the record, and any field you store can be dropped into a message the same way.
The second is branching, and it is the one that changes results. The Contact field value condition checks whether information on the contact record, on their primary company, or in the UTM tags that record where a visit came from, matches criteria you set, and sends them down a different path accordingly. That is not a different name in the greeting. It is a different email.
Where the weak and strong forms of personalization sit is a question the email marketing post settles. The mechanic here is what makes the strong form buildable at all. If the plan a customer is on lives in a field, a condition can read that field, and the campaign can say something that is only true for people on that plan.
The Integrations That Carry the Data
An automation platform is worth no more than the data reaching it, and integrations are how the data arrives and leaves. Mautic's campaign builder documents two ways out. Push Contact to Integration creates or updates the contact inside a connected system. Send a Webhook sends a request to a URL you choose, with headers and data you control, and it accepts tokens such as the contact's fields and their address.
The webhook is the escape hatch, and it is worth understanding before you judge a platform on the length of its connector directory. Anything that can receive a web request can be wired to a campaign, whether or not the vendor lists it as a supported integration.
What integrations never fix is the direction of truth. Two systems holding the same person will disagree eventually, and the question to settle before connecting them is which one is allowed to overwrite the other.
Where AI Fits, and Where It Does Not
What vendors currently ship under the AI label sits on top of the decision logic, not inside it. ActiveCampaign's homepage leads on taking action on your marketing data in ChatGPT and Claude, on memory and custom instructions, and on scheduled tasks. Those are useful things. None of them changes what a decision or a condition does when a contact arrives at it.
The one place machine judgment has moved into the mechanism is timing. Mautic's smart event schedule reads a contact's own history of email reads, page hits and form submissions and picks when to run a send, instead of firing at a fixed interval. That is a real change and a narrow one. It decides when, not whether.
So the practical read for somebody buying now is that these features speed up the making and the asking. The diagram you draw is still the diagram that runs.
Marketing Automation Examples
An example of marketing automation is any sequence where something a person does causes the software to act, with nobody deciding case by case. Welcome sequences and cart reminders are the examples everyone reaches for, and both of them are email, which is part of why the category gets mistaken for email software. These five show the rest of it.
A visit that changes a score and a queue. Source: a segment of existing customers. Decision: Visits a page, pointed at your pricing page. Green path: Adjust Contact Points, then Update Contact owner to the account manager for that region. Nothing is sent to anyone. A salesperson finds a customer sitting in their queue with a reason attached.
A download that tags and hands off. Source: a form. Decision: Downloads Asset. Green path: Modify Contact's tags to record what they took, then Push Contact to Integration so the record shows up in the system the sales team lives in. Red path: nothing, which is a legitimate branch and an underused one.
A form answer that splits the sequence. Source: the form. Condition: Form Field value, checking what the person selected when they filled it in. One path per answer, and the contact is never asked to identify themselves twice.
A renewal date that nobody has to remember. Source: a segment whose filter is a renewal date equal to today, which is what makes the date arrive as an entry instead of a test. Action: Send Email to User, so the person who owns the account is told, instead of the customer being mailed a reminder they did not ask for.
A reply that stops everything. Decision: Replies to Email. Green path: Change Campaigns, removing the contact from the sequence they are in. A human is now in the conversation and the machine should get out of the way.
The anatomy repeats every time. Something admits the contact, something watches or tests, and one or more actions fire on each branch. Once you can name those three parts in an example, you can build it.
Why Companies Buy It, and What It Does Not Fix
Companies buy marketing automation for five reasons, and every one has a limit that rarely gets printed beside it.
| What it does | The limit attached to it |
|---|---|
| Runs the same sequence for everyone, at any volume | It runs a bad sequence at that volume too |
| Reacts soon after something happens, at any hour | Only to events it has been wired to notice |
| Keeps a record of what each person did and when | The record is worth exactly what you feed it |
| Hands a person to a salesperson with context attached | It cannot turn an unqualified person into a qualified one |
| Removes repeated manual work | It adds new work: designing, testing and maintaining the diagrams |
What it does not fix is anything upstream of itself. It will not create demand for something nobody wants, repair an offer that people understand and decline, or grow a list, because nothing inside a campaign builder adds a contact who was never there. Automation multiplies whatever is already happening: if a sequence you run by hand converts, running it automatically converts more, and if it does not, you have bought a faster way to be ignored.
What You Need Before You Turn It On
Four things have to exist before a workflow can do anything, and a platform bought ahead of them sits idle while the invoice runs.
Contacts who agreed to hear from you. A campaign source is a segment or a form, and both are downstream of somebody deciding to give you their address. Nothing in this software builds an audience.
Data on those contacts that the software can read. Conditions test fields. If the field you want to branch on is empty for half your list, the branch sends half your list down the same path, and the personalization you designed collapses back into a broadcast.
Something worth sending. The category assumes the messages exist and only the timing is missing. For a team starting out that assumption is usually backwards.
One outcome you can name. A campaign with no defined end has no way to be wrong, which means it can run for a year without anyone being able to say whether it worked.
The small-list question belongs here, and the answer is arithmetic instead of size. A license is a fixed annual cost whatever the list does, so the test is whether the repeated decision you would hand over is worth that much of somebody's year. Say a founder sends the same follow-up to twenty people every week and loses an afternoon to it. That afternoon is the thing being bought. If there is no such afternoon, there is nothing for the software to take over.
Choosing a Platform, and What It Costs
No ranked list of platforms survives contact with a real decision. Many of the rankings that exist are published by companies that sell one of the tools, and even an honest ranking ranks against a use case that may not be yours. What helps instead is a real price, since the category hides prices, and a short set of questions that separate one product from another.
What One Published Price List Shows
Vendor pages in this category leave the price off the page. Government procurement does not. New York State's Office of General Services (OGS) publishes the contract price schedule for Acquia Inc, and its Appendix E workbook, created in July 2026, lists Acquia Campaign Studio at seventeen prices, one per contact tier, running from 5,000 contacts to 2,500,000.
| Contacts | List price a year | Cost per 1,000 contacts |
|---|---|---|
| 5,000 | $16,480 | $3,296 |
| 20,000 | $19,795 | $990 |
| 30,000 | $21,630 | $721 |
| 50,000 | $23,795 | $476 |
| 100,000 | $28,120 | $281 |
| 250,000 | $40,560 | $162 |
| 500,000 | $59,485 | $119 |
| 1,000,000 | $97,335 | $97 |
| 2,500,000 | $210,895 | $84 |
Acquia Campaign Studio list prices, from the New York State OGS contract price schedule for Acquia Inc, Appendix E. Nine of the seventeen published tiers are shown. The right-hand column is our own division of the list price by the tier size, rounded to the dollar.
Two readings come out of that ladder. The first is that the entry price is the wall. Two hundred times the contacts, from 5,000 up to 1,000,000, costs a little under six times the money, and the cost of carrying a thousand contacts falls from $3,296 to $97 across the same range. A small list is not what makes this expensive.
The second is that the sticker is not the bill. The same schedule prices Acquia Campaign Studio Quick Start, an onboarding engagement, at $5,500. Setup is a separate line, and it is almost never quoted next to the software.
Three limits on those numbers, stated so nobody reads them as a market benchmark. They are one vendor's schedule under one state contract. New York's own net price sits 4.46% below the list price shown. And this is the enterprise end of a category whose other end serves a small list for a modest monthly fee. The ladder is useful because it is published at all, and because its shape, costly to start and cheap to grow, is the shape of the whole category.
One fact joins the price to the mechanics. Acquia Campaign Studio is the commercial distribution of Mautic, and Acquia's own product page invites buyers to "Tap Into the Mautic Community" and describes it as the biggest, most active open-source marketing automation community on earth. The grammar of sources, decisions and conditions is the grammar running underneath that price.
The Questions That Separate Platforms
Five questions do more work than any feature grid.
- What can start a campaign, and how long after the trigger does it act? Some platforms fire on the event. Mautic's actions and decisions require a scheduled job that runs its trigger command at an interval you choose, so "immediately" means "the next time that job runs". Ask, because a five-minute gap is fine for a nurture sequence and useless for a cart.
- Which conditions can it test, and against which fields? The branch you want to build has to exist as a condition. If you need to split on a custom field, a company attribute or a UTM tag, check that the condition list reaches those and not only the built-in ones.
- What can it do besides send a message? The action list is the honest feature list. Scoring, tagging, ownership, stage changes and webhooks are what separate a workflow tool from a scheduler with branches.
- What does it take to run, and who runs it? Somebody has to build the diagrams, maintain the fields and read the reports. One vendor answers that question in dollars, with a priced onboarding engagement; your answer is a name and a share of somebody's week.
- How does data get in, and how does it get out? Native connectors are convenient and finite. A generic webhook is the difference between a platform you can extend and one you have to wait on.
How to Start
Start with one automation, and pick it by asking which decision you are already making by hand more than once a week. That is the whole selection method, and it beats a feature audit because it names something you can check afterwards.
Then, in this order:
- Write the sequence on paper. For every step, name three things: what admits the contact, what is watched or tested, and what happens on each branch. If you cannot fill in all three, the step is not ready to build.
- Fill the fields the branches need. A condition on a field that is empty for most of your list is a branch that never splits.
- Build it for a small segment. Small enough that a mistake reaches a few dozen people and not a few thousand.
- Run yourself through it. Enter your own record at the source and walk both paths, including the red one, with the delay shortened so you see it fire.
- Watch it for a month before adding a second. The failures that matter show up in the second and third week, when contacts start reaching branches nobody tested.
Setup costs time before it costs money. At the enterprise end the New York schedule prices onboarding as its own engagement, and at the small end that same work is your own weeks, spent before the first campaign sends anything to anyone.
What to Measure
Measure the campaign, not the channel. Four numbers tell you whether an automation is doing its job, and all four are about the sequence itself.
Entry rate. How many contacts enter each week. A campaign nobody enters is a source problem, and it is invisible if you only look at what the campaign sent.
The split at each branch. What share take the green path where the diagram forks. A branch where everyone goes the same way is not a branch, it is a delay with extra steps. Say a condition splitting on a field almost nobody has filled in: the reading is a near-total red path, and what it measures is the field, not the message.
Completion. How many contacts reach the end of the sequence against how many entered. The gap tells you where people stop, and the step immediately before the gap is the one to rewrite.
The outcome the campaign exists for. Named before the build: a reply, a booking, a renewal, a purchase. If the campaign moves that number, the branch statistics are diagnostics. If it does not, they are decoration.
One diagnostic that is not a metric: open the campaign and look at where contacts are sitting right now. A pile at one step is the quickest way to find a branch behaving in a way you did not draw.
Where Marketing Automation Goes Wrong
Marketing automation goes wrong in a small number of ways, and each one is a mechanism doing exactly what it was told.
A condition where a decision belongs. You want the software to wait for something to become true, and a condition never waits. The campaign looks alive, the negative branch fills up, and nothing you designed ever fires.
A red path with no delay. The non-action branch takes its timing from the action attached to it, so a reminder with no delay set fires against a contact who was given no time to act.
The whole segment entering at once. A campaign switched on over an existing segment admits everybody the same afternoon. Say a shop builds one over its entire customer list and step one is an email: harmless as a scoring rule, and unmissable as a send.
Scoring nobody reviews. Points accumulate against rules written months ago, a threshold gets crossed by activity that means nothing, and a salesperson learns to ignore the queue.
A sequence with no way out. Replies to Email and Change Campaigns exist to pull somebody out of a flow when a human takes over. Campaigns built without an exit keep mailing a person who is already in a conversation with you.
Automating something you have never done by hand. The diagram encodes a process. If the process has never run, the diagram encodes a guess, and it will run that guess against your whole list.
Nobody owning the diagram. Workflows outlive the person who built them. A campaign whose owner has left keeps running, and nobody left in the building knows what it does or dares switch it off.
Common Questions About Marketing Automation
Three questions come up about marketing automation more often than the others, and all three are about fit instead of features: what it takes to run one, whether a small list justifies it, and where it stops being a CRM. The answers are short on purpose.
What Skills Are Needed for Marketing Automation?
The skills follow the parts of a campaign. You need to write a message worth reading, define an audience precisely enough that it can become a segment or a form, reason about branching logic without losing track of the second path, and read a report closely enough to see which branch people took. None of that needs code, though a webhook and a tidy set of contact fields are the two places a technical hand pays for itself.
Is Marketing Automation Worth It for a Small List?
The list size is the wrong test. What decides it is whether the same decision is being made by hand often enough to be a real cost, and whether the data that decision reads is already on the record. A list of three hundred people where you recognize most of the names rarely clears that bar, and the same three hundred with a renewal date on every record and a weekly follow-up ritual might.
Is Marketing Automation Part of CRM?
No, though the two are sold together often enough that the line blurs. A CRM stores the record. Marketing automation reads the record and acts on it. Several vendors ship both inside one product, which is convenient and does not make them the same function: you can run automation against a CRM you did not buy from the same company, and plenty of CRM users never automate anything.
The Short Version
Marketing automation is a set of rules that read a contact record and act on it on their own. The rules are made of four parts, and knowing them is most of what separates a workflow that runs from one that only appears to. A source admits the contact. A decision waits for something they do. A condition tests something you already know, and resolves on the spot. An action does the thing, and sending a message is one action among many.
The published prices say the rest. Ten times the list, from 5,000 contacts to 50,000, moves one vendor's annual list price from $16,480 to $23,795, so a small list is not what makes this expensive, and setup is a separate line that nobody quotes.
Two questions settle whether to buy. Which decision are you making by hand often enough that a written rule could make it instead? And is the data that rule would read already sitting on the contact record?
With answers to both, the first build is small and the software earns its price. Without them, a platform will run your empty diagram faithfully, for a year, at whatever the contract says.
Sources
- Acquia Inc, Appendix E, contract price schedule, New York State Office of General Services, workbook created July 2026.
- Using the Campaign Builder, Mautic Documentation.
- Acquia Campaign Studio, Acquia.
- ActiveCampaign, ActiveCampaign homepage, 2026.