The Most Common Mistakes in Marketing Automation Start Before the Build
Ask a team to describe what happens after someone fills in a demo form and you usually get three different answers. Marketing thinks the lead is scored and nurtured. Sales thinks it gets a call the same day. The founder assumes both. Nobody has written it down, so the automation gets built on whichever version the person setting it up happened to believe.
That is the first mistake, and every later one grows out of it. The platform arrives with a default pipeline, default lifecycle stages and a library of template workflows. Teams adopt them because they are there. Six months later the deal stages describe a generic company's sales process, not yours, and the reports built on them measure movement through steps your reps never actually take.
The fix is unglamorous: map the process first. Every source a lead can arrive from, every stage it passes through, who touches it at each stage, and what has to be true for it to move. When we build a client's CRM, discovery and sales-process mapping is the first step, and the pipeline is designed to mirror the real sales process rather than a default. Only then does it make sense to decide which steps a machine should take.

A useful test before anything is switched on: can you draw the whole thing on one page, with a named person beside every step that is not automated? If not, the automation will make the gaps faster, not smaller.
Scoring Activity Instead of Fit
Of all the common mistakes in marketing automation, scoring is the one that does the most quiet damage. It is where good intentions go to inflate. The standard setup adds points for every email open, every click and every page view, and declares a lead "marketing qualified" when it crosses a threshold. The result is a list of very active people, many of whom could never buy.
A student researching a thesis can out-score a VP at a target account. So can a competitor. So can an existing customer reading your help docs. Activity tells you someone is interested in the topic. It says almost nothing about whether their company fits.
Email opens have also become a weak signal in their own right. Since Apple introduced Mail Privacy Protection in 2021, Apple Mail can load messages in the background, which registers an open whether or not anyone read the email. A score built on opens is partly built on a mail client's behaviour.
Score in two separate dimensions instead:
- Fit: is this the kind of company and role you sell to? Size, industry, technology, region, and the disqualifiers that rule an account out.
- Timing: is there evidence they are looking now? A pricing page read twice, a named contact returning several times in one week, a demo request. These first-party signals are accurate and already sitting in your own systems.
Write the rule down. On our intent and signal work, the combination of signals that justifies an approach is written as a rule rather than held as a judgement, because otherwise the definition drifts and nobody can say why one account was contacted and another was not. Then route by it: demand that fits a target account goes into the account programme, the rest goes to self-serve or nurture, and nothing is thrown at sales indiscriminately.
Automating the Email and Forgetting the Person
A lot of automation is built to send things. Much less of it is built to make sure a human acts. The welcome email goes out in seconds; the sales call happens on Thursday, if someone notices the notification.
That gap is expensive because interest decays quickly. A signal that fades in two days and reaches a rep in five has been paid for and thrown away. The handoff is the step that most deserves automating, and it is the one most often left to an inbox.
Build the handoff as carefully as the nurture:
- A stated response time for each kind of lead, decided before launch, not discovered afterwards.
- Routing to a named owner, with the context attached: what they asked, which page they came from, what they downloaded. An alert with nothing in it gets ignored.
- A fallback when nobody picks it up. If a lead sits untouched past the agreed window, the system should escalate, not wait.
- An instant first touch so the lead hears something while the human is on the way. On the CRM systems we build, speed-to-lead runs automatically: a text and email go out within seconds of a form fill or an ad lead.
Sales has to agree to the response time, in writing, with marketing. If they will not, the problem is not the automation, and no workflow will fix it.
Five Tools That Do Not Talk to Each Other
The email platform has one list. The CRM has another. Webinar registrations live in a spreadsheet, ad leads in the ad platform, chat transcripts somewhere else again. Each tool automates its own corner, and none of them knows what the others did.
You can see the symptoms from outside. A prospect gets a "still interested?" nurture email the day after a sales call. A customer who signed last month is still being sent trial offers. Two people from the same account get two unrelated sequences. Every one of those is a data problem dressed up as a messaging problem.
The rule we build to is simple: every lead, from every source, routes into one CRM record, and every automation reads from and writes to that record. Forms, funnels, chat and the Google and Meta ad lead forms all feed the same place, so every lead is worked the same way.
It should also run in the other direction. Closed-won deals can be pushed back to the ad platforms as offline conversions, which is what lets LinkedIn and Google optimise toward revenue rather than form fills. Skipping that step leaves the ad platforms training themselves on the cheapest leads they can find, which is rarely the same thing as your best customers.
Treating Consent and Deliverability as Afterthoughts
For anyone emailing Canadians, Canada's Anti-Spam Legislation sets the floor, and automation makes it easy to fall through. CASL requires consent before a commercial message is sent, and implied consent expires: roughly two years after a purchase and six months after an enquiry, unless the person gives express consent in the meantime. A workflow that keeps mailing every contact it has ever collected will, sooner or later, be mailing people it no longer has permission to contact. Every message also has to identify the sender and carry a working unsubscribe, and unsubscribe requests must be honoured within 10 business days.
So store how and when consent was given as a field on the contact record, and let the automation read it. A contact whose implied consent has lapsed should drop out of commercial sequences on its own, not when someone remembers to clean the list.
Deliverability is the practical cousin of consent, and it matters as much for account-based email as for a newsletter. Since 2024, Google has required anyone sending more than 5,000 messages a day to Gmail addresses to authenticate their domain with SPF, DKIM and DMARC, offer one-click unsubscribe, and keep reported spam rates below 0.3%. Large automated programmes cross that line without noticing, and the first sign is usually a quiet fall in replies rather than an error message.
Set It and Forget It
Workflows age. The offer in email four ended last spring. The webinar it promotes already happened. A branch that depends on a field nobody fills in any more sends every contact down the same path. The person who built the sequence has moved on, and nobody else knows it exists.
The quieter version of the same problem is the list that never shrinks. Most leads will not buy this quarter. Without a rule for releasing them, a nurture list grows until the same companies are contacted again and again by people who do not know it has happened before. That is how a brand becomes the one prospects filter to spam.
Auditing for Common Mistakes in Marketing Automation
A quarterly review catches most of it. Put an owner's name on every active workflow, then check:
- Every live sequence: does each email still describe an offer, event and price that exist?
- Every branch condition: is the field it reads still populated? Run a test contact through it.
- Every list: who has been in nurture for more than a year without engaging, and what is the rule for letting them go?
- Every handoff: what was the actual response time last quarter, against the one you agreed?
- Every report: does it still match the deal stages the team really uses?
If a workflow has no owner, turn it off. A sequence nobody will defend is one nobody is checking.
Measuring Opens Instead of Pipeline
Automation platforms report what they can see: sends, opens, clicks, form fills, contacts moved between stages. Those numbers are easy to grow and almost unconnected to revenue. A team can double its open rate and leave the pipeline exactly where it was.
This is the reason so many common mistakes in marketing automation survive for years: the dashboard says everything is working. Measure the automation by what it is supposed to produce. On our demand generation work the numbers that count are sales acceptance rate and win rate by source, because they show whether what marketing hands over is something sales can close. Add the ones specific to automation:
- Median time from enquiry to first human contact.
- Share of qualified leads contacted inside the agreed window.
- Pipeline and closed revenue from contacts who went through each major sequence, compared with those who did not.
- Unsubscribe and spam-complaint rates per sequence, as an early warning.
If a sequence cannot be connected to any of those, ask what it is for.
Fixing Common Mistakes in Marketing Automation, in Order
Do not start with the content. New emails on top of a broken process just send more of the wrong thing. Work through it in this order:
- Map the process on one page: sources, stages, owners and the rule for moving between them. Agree it with sales.
- Get every source into one record, and push closed-won deals back to the ad platforms.
- Rebuild scoring as fit plus timing, written as a rule, with opens given little or no weight.
- Automate the handoff: response times, named owners, context attached, escalation when nobody acts.
- Put consent on the record and let workflows read it. Check authentication and spam rates.
- Only now, fix the sequences: retire stale ones, give every survivor an owner, set a review date.
- Report against pipeline, not activity, and review quarterly.
The first four steps usually matter more than everything else combined. A plain email sent within minutes to the right person at a company that fits will outperform a beautifully designed nurture sent to everyone.


