The one line difference
A CRM records what happened and tells a person what to do next. Marketing automation decides what gets sent and sends it.
Everything else follows from that. The CRM is filled deliberately by people doing work: calls, quotes, jobs, stage changes. Marketing automation is filled by behaviour: opens, clicks, page views, form submissions, and it acts on that behaviour without anyone involved.
| CRM | Marketing automation | |
|---|---|---|
| Primary user | Sales, service, operations | Marketing, often one person |
| Opened | Every morning, by everyone | Weekly, by a specialist |
| Core object | A record with an owner and a stage | A contact in a list, with a score |
| Core verb | Follow up | Send |
| Success looks like | Nothing goes quiet, forecast is trusted | Right message, right moment, at scale |
| Fails through | Neglect, so data goes stale | Mess, so sends go wrong |
| Reports on | Revenue, pipeline, activity | Delivery, engagement, conversion to enquiry |
One sentence test. If your problem is that people are not being contacted, you need a CRM. If your problem is that too many people need contacting to do it by hand, you need marketing automation.
Which one are you actually short of
Four symptoms, and each points clearly at one side.
Quotes go quiet and nobody notices for weeks. That is a CRM gap. Automation will not help, because the problem is that no record has an owner and a dated next action, and a sending tool has no opinion about ownership.
You have five hundred enquiries a month and three salespeople. That is an automation gap. There is no human way to give five hundred people a timely, relevant response, and adding a CRM alone just gives you a very well organised backlog.
You cannot say which marketing produced which revenue. This is usually a CRM gap disguised as a marketing one, because attribution requires source captured on a record that later carries a value. An email platform can tell you who clicked, and only a system of record can tell you who paid.
Your emails all go to everyone. Automation gap, but check the data first. Segmentation is limited by what you know about people, and most businesses that cannot segment have a data problem rather than a tooling problem.
The pattern worth noticing: two of the four look like marketing problems and are solved in the record. That is why we recommend the CRM first in almost every case.
Why the suites cost more than expected
The obvious answer to owning both is a suite that does both, and for marketing led businesses it can be a good answer. Two pricing mechanics catch buyers out, and neither is hidden, just rarely modelled.
Marketing tiers are priced on contacts, not on seats. Your bill grows with your database, which grows whether or not you hire anybody, and it keeps growing with contacts who will never buy again. The cost of a bad import is not just messy data, it is a permanent line item.
The feature you will certainly need sits two tiers up. Multi step workflows, custom reporting, removal of the vendor's branding, or the number of sending domains. Find out which one applies to you before you sign, because the jump between tiers in this category is frequently three to five times, not twenty percent.
Model three years at your expected contact growth, not at today's list size. We have seen a suite quoted at four hundred a month reach three thousand a month in year three, entirely predictably, without the business gaining a single extra user.
Two tools, connected properly
The alternative is a CRM as the system of record with a sending tool beside it. Done well this is cheaper and more flexible. Done badly it produces two databases and a permanent argument about which is right.
Four rules make it work, and skipping any one of them produces the second database.
- The CRM owns the contact. Nobody creates or edits a contact in the sending tool. If someone needs a new contact, they make it in the CRM and wait for the sync, which is usually minutes.
- Segments are computed in the CRM and pushed out as audiences. Not rebuilt by hand in the marketing tool with slightly different criteria, which is how two definitions of active customer come to exist.
- Unsubscribes and bounces flow back immediately and are treated as authoritative. This is a legal requirement, not a preference, and it must be tested rather than assumed.
- Engagement comes back summarised, not in full. The CRM does not need every open event. It needs last engaged, a simple score and the campaign that produced an enquiry. Syncing raw events into a CRM makes it slow and no more useful.
Consent is the part that must live in the CRM
Whichever architecture you choose, consent belongs with the record, and this is the single most common thing we have to fix in existing setups.
You need to be able to answer three questions about any person quickly and with evidence: what they agreed to, when, and through which form or conversation. A checkbox whose history is lost the moment somebody toggles it is not evidence. Consent should be stored as an event with a timestamp and a source, so the answer is a row rather than a recollection.
The reason it must sit in the CRM rather than the sending tool is that the sending tool is replaceable and the record is not. Businesses change email platforms every few years, and a migration that leaves consent history behind creates a database you cannot lawfully use, which is an expensive way to discover where your evidence was living.
One practical detail worth building on day one: a single do not contact flag on the record that suppresses everything, everywhere, and is respected by every outbound channel including a salesperson's own sequence. Businesses get into trouble not through malice but through a person being suppressed in one system and not in another.
What to buy first, by situation
| Situation | First purchase | Why |
|---|---|---|
| Under 500 contacts, few enquiries | Neither | A shared sheet and a habit. Revisit in a year |
| Quotes go quiet, low volume | CRM | The gap is ownership, not sending |
| High enquiry volume, small team | CRM, then automation quickly | Organise first, then scale the response |
| Marketing led, content driven, long cycles | Suite or a tight pair | Nurture is the core activity here |
| Field or job based business | CRM, built around jobs | Most events never touch an inbox |
| Both exist and disagree | Fix the seam | You have two databases, not a tooling gap |
The last row describes more businesses than any of the others, and it is the cheapest problem on the list to fix. One afternoon deciding who owns the contact, one direction for the sync, and a tested unsubscribe path usually removes a complaint that has been running for years.
How this is built when the CRM is yours
In systems we build, the split is deliberate and the CRM stays the record. We do not build a sending platform, because deliverability, unsubscribe handling and reputation management are a specialised job done well by tools that cost very little.
What we build instead is the part that is specific to your business: segments computed from your own data on a schedule, consent stored as events, a single suppression flag that everything respects, and a two way connection to whichever sending tool you prefer, monitored for twelve months as part of the build.
The automations that live in the CRM are the operational ones, and they are usually worth more than the marketing sequences: when a quote has had no response in three days, create a task. When a job is completed, request a review. When a maintenance agreement is thirty days from renewal, raise the invoice and notify the customer. Those are the rules described on the CRM automation page, and none of them belong in an email platform.
If you already own both and they disagree, that is a scoping conversation rather than a build, and it is usually short. Tell us which two systems and what each one currently edits, through the contact page, and you will get a written answer about where the seam should sit within two business days.
Lead scoring, and why most of it is theatre
Scoring is the feature demonstrated most often in this category and the one abandoned most often after purchase, so it is worth being blunt about when it works.
The usual setup assigns points for behaviour: five for an email open, ten for a pricing page visit, twenty for a demo request, minus points for inactivity. Sales are told to call anyone above a threshold. Within a quarter, nearly everyone has stopped looking at the number.
Three reasons it fails. The weights are invented rather than derived, so the score measures somebody's theory about buying rather than actual buying. Open tracking is unreliable now that mail clients preload images, so a large share of points come from machines. And a score is a single number describing a person, which collapses the two things sales actually need to know: are they the right kind of customer, and are they doing something right now.
What works is splitting those apart. Fit is a small set of factual attributes you can check, such as industry, size and location, and it barely changes. Intent is a short list of behaviours that genuinely correlate with buying in your business, usually visiting pricing, requesting something, or replying. Two axes, four boxes, and a rule for each box. It is less impressive in a demo and considerably more useful on a Monday.
The one thing that makes either version worth having is derivation from your own closed deals. Look at fifty customers you won and fifty enquiries you lost, find what the winners did that the losers did not, and score that. Without the exercise, the model is decoration. That exercise needs a CRM holding outcomes, which is another reason the record comes first.
AI tools for marketing and CRM: what works in 2026
Every CRM and every marketing platform now sells AI. Some of it saves real time and some of it is a demo. After building these features for clients, the line falls in a fairly predictable place.
| Use | Works today? | Why |
|---|---|---|
| Drafting a reply from the record | Yes | The record supplies the facts, a person checks before sending |
| Summarising a long thread or call | Yes | Low risk, saves minutes on every handover |
| Cleaning and deduplicating imports | Yes | Suggestions a person approves in bulk |
| Writing whole campaigns | Partly | Fine for a first draft, weak on your actual offer |
| Predicting which leads will buy | Rarely | Needs thousands of closed deals to learn from |
| Sending without review | No | One wrong message costs more than the hours saved |
The useful pattern is AI that prepares and a person who decides. That keeps the speed and keeps the brand. It also means the AI needs the CRM record to work well, which is one more reason the record should be complete. Our AI CRM page shows how we wire it in, using your own API key so the usage bill is visible and yours.
Three industries where the line falls differently
The general advice in this guide holds for most businesses. Three industries we work with often bend it in different directions.
Home builders run the longest cycles on this list. A buyer may register interest a year before a plot is released, so marketing automation carries real weight: plot release announcements, show home invitations, finance and part exchange guides. The CRM has to hold the plot, the buyer's stage and the reservation, and the sending tool only ever works from segments the CRM produces. See the home builder CRM.
Insurance agencies live on renewal dates, so the most valuable automation is not marketing at all. It is a renewal sequence triggered from the policy record at ninety, sixty and thirty days. Because the business is regulated, every message sent needs to be recorded against the client, which is a strong argument for sending from the CRM rather than from a separate platform. The insurance agent CRM is built that way.
Video production and creative agencies barely need marketing automation. They win a few large clients through referrals and pitches, and the work is project shaped. A pipeline, a proposal tracker and a light newsletter are enough. Spending on an automation suite here is the clearest case of buying the wrong tool first. Our agency CRM puts the effort into projects and retainers instead.
What a small team should automate, in order
Automation is sold as campaigns and earns its money as chores. For a business with a handful of people, the order that returns the most per hour of setup is consistent.
- The instant acknowledgement. Every enquiry gets a reply within seconds saying what happens next and when. Trivial to build, and it measurably reduces the number who contact a competitor while waiting.
- The internal chase. A quote with no response after three days creates a task for its owner. This one recovers more revenue than any marketing sequence, and it lives in the CRM rather than the sending tool.
- Appointment and arrival messages. Confirmation, reminder, and an arrival window on the day. In any business that visits people, this is the highest value automation there is.
- The review request. Sent when work is completed, to the right person, once. Local search visibility is built on this and almost nobody does it consistently by hand.
- Renewal and reorder timing. Anything with an anniversary: maintenance agreements, insurance, annual services, consumables. The system knows the date, so nothing should depend on remembering it.
- Reactivation. Customers with no activity for a set period, contacted with something specific rather than a newsletter.
Only the last two are marketing automation in the conventional sense. The first four are operational rules that belong with the record, which is the practical version of the whole argument in this guide: most of what businesses want from automation is not sending, it is remembering.
