Email and automation

How do you build marketing automation, and what should you automate first?

The ScaleLab teamPublished 9 min read

In short: Marketing automation is built in a set order: you map the customer’s path from first contact to order, bring customer data together in one place, and only then set up rules that take over the repetitive steps. Automate first the steps where delay loses customers, such as replying to a new enquiry. Conversations, proposals and decisions about what to stop stay with your team.

Why does automation start with the process, not the software?

Automation often starts with buying software. The company pays for a subscription, sets up a few emails and, months later, finds that nobody knows which messages go out or to whom. The reason is simple: the software runs a process, and the process was never written down.

Good automation does three things:

  • Removes manual work: steps that repeat every day in the same way happen without anyone having to start them.
  • Treats every customer the same: each customer goes through the same steps, whoever is on shift or on holiday.
  • Keeps the know-how: the process lives in the system, so a new team member does not start from scratch.

None of this happens if the steps are unclear before they are moved into software. That is the difference between a system that works and a subscription that simply gets paid.

What should you prepare before the first rule?

1. Pick one goal

Start from the business goal. It might be more orders from existing customers, faster replies to enquiries or fewer customers who drop away after their first purchase. Each goal leads to a different first automation, so do not try to solve all three at once.

2. Draw the customer’s path

Take a sheet of paper and note where the enquiry comes from, who receives it, and what happens in the first hour, the first week and after the order. Mark where there is manual work today and where customers are lost because nobody called back in time.

3. Check the data

If enquiries from the website, the phone and social media sit in different places, the automation will work from an incomplete picture. Bring them together before you write any rules.

ScaleLab tip: Send an enquiry yourself through every form on your website and follow where it goes: to whom, how quickly and whether the details arrive complete. This ten-minute test often reveals more than a week of discussions about software.

What is the system made of?

Software names change, but the parts stay the same. Each part does not have to be a separate tool: in a smaller company, one tool often covers several of them.

  • Customer system (CRM): where every contact lives: where they came from, what they asked, what they bought and who on the team looks after them.
  • Email platform: sends sequences of messages when a customer takes a certain action or when a set amount of time has passed.
  • Website tracking: records enquiries, orders and important visitor actions, so the system knows what has happened.
  • Connections between tools (integrations): when a customer fills in a form, their details reach the customer system and the email platform on their own, without anyone copying them across.
  • Report: brings together how many enquiries and orders there are, where they come from and what they cost.

The connections are the part most often skipped. Without them, each tool knows only its own piece and the team is back to moving data by hand. If some enquiries come in through a chat on your website, see also how an AI chat assistant can take enquiries without losing them on the way to your team.

Why is your email domain set-up part of automation?

Automated emails only make sense if they reach the inbox. The large mailbox providers now require bulk senders to prove that their emails really come from their domain. That makes authentication not a separate technical chore but the first step of any set-up.

Provider Who it applies to What is required Since
Gmail close to 5,000 or more messages a day to personal Gmail accounts authentication, spam rate below 0.1% and never reaching 0.3%, one-click unsubscribe on marketing emails from November 2025, non-compliant messages face temporary and permanent rejections
Outlook, Hotmail, Live 5,000 or more messages a day SPF, DKIM and DMARC (at least p=none, aligned) from 5 May 2025, rejected with code 550 5.7.515

What does this mean for your business? If you send newsletters or sequences to a large list, check your settings before you switch the automation on. With Gmail, bulk sender status is permanent: once you have reached the threshold, the requirements apply to you even when you send less. Even below the threshold, authentication helps keep your emails out of spam. How to set up the three records is explained in the article on SPF, DKIM and DMARC.

ScaleLab tip: Add a one-click unsubscribe link to every marketing sequence as soon as you create it. Someone who cannot find how to unsubscribe presses the spam button instead, and that harms all of your emails.

The rules are set by Google and Microsoft and can change. The details in the table are as of September 2026, so check the current requirements on both providers’ pages.

Which steps should you automate first?

Choose steps that repeat often, follow a clear rule and cost money when they are delayed.

1. Replying to a new enquiry

Someone who has filled in a form expects to know it has arrived and what happens next. An automatic confirmation with a clear next step, plus a notification to the person responsible, is the rule worth setting up first. The system can also remind that person if the enquiry is left unanswered.

2. A sequence after an enquiry

Not everyone who asks is ready to buy straight away. A few emails spread over time can answer the questions people have before deciding: how the work runs, what they need to prepare, what the common mistakes are. When the person is ready to talk, they already know the basics, and the conversation is shorter and more focused.

3. A new customer’s first weeks

After a first order or a signed contract, the customer needs guidance: how to use the product, who to contact, what to expect. A sequence like this can cut repeated questions to the team and helps the customer see the value of the purchase sooner.

4. Customers who have not ordered for a while

A rule that spots a customer with no order after their usual interval can send a reminder or something useful. The interval depends on the product: short for consumables, long for services with a yearly cycle.

5. Customers who have left

When a customer cancels a subscription or contract, a short message asking why gives you information that is rarely available otherwise. The answers show whether the reason lies in the price, the product or the service. Whether to offer a way back is decided by a person after reading the answers.

What stays with people, and which mistakes should you avoid?

Automation follows rules. It cannot judge when a rule no longer applies. That is why the team keeps conversations with customers about larger decisions, proposals tailored to a specific business, the wording of messages and changing it when customers stop responding, decisions about which sequences continue, and replies to complaints. A good sign is when, after automation, the team has more time for these things, not more notifications to watch.

Common mistakes when building it:

  • Software before process: the subscription is paid, but the customer’s path has not been mapped.
  • Too many sequences at once: nobody knows exactly what the customer is receiving.
  • Stale data: a customer who has already bought keeps getting invitations to buy.
  • Unread messages: emails go out in the company’s name without anyone on the team having read them once.
  • No one responsible: nobody reviews the results at least once a month.

How can you tell whether the system is working?

Compare the metrics before and after each new automation, not for the whole system at once. That way you know which step changed what.

  1. Time to reply to an enquiry: shows whether customers get a reply in time, including outside working hours.
  2. Share of people who send an enquiry or order: shows whether sequences lead to action, not just to opened emails.
  3. Repeat orders: shows whether messages to existing customers bring in revenue.
  4. Cost per order: shows whether an order costs less on average, counting both the software and the work on it as cost.
  5. Hours of manual work: count them before and after. It is your own estimate, but without it the time saved remains an impression.

How cost per order and the share of visitors who order are calculated is explained in the metric definitions. Our recommendation: if a sequence changes nothing after a few months, stop it and put the effort elsewhere.

ScaleLab tip: Keep a simple spreadsheet listing every active sequence, its goal, its start date and the person responsible. When the list gets too long to take in at a glance, you have more sequences than you can keep track of.

Marketing automation is not a project you finish but a system that starts with one well-chosen step and grows as each next step proves it serves the goal. That is how we approach email marketing and automation: process, data and an authenticated domain first, sequences after that.

Frequently asked questions

Which step should I automate first?

For most companies it is the reply to a new enquiry: an automatic confirmation to the customer and a notification to the person responsible. The step repeats often, the rule is clear, and delay directly loses customers.

Do I need expensive automation software?

Not necessarily. In a smaller company, one tool often covers the customer system, email and reporting. It matters more to map the customer’s path and bring the data together first, then choose software that does exactly that.

Do the Gmail and Microsoft requirements affect me if I send few emails?

The strict requirements apply to bulk senders, around 5,000 or more messages a day to the provider in question. Authentication with SPF, DKIM and DMARC still helps any sender reach the inbox, so set it up from the start.

How do I know a sequence is not working?

Compare the share of people who send an enquiry or order, and repeat orders, before and after it went live. If nothing has changed after a few months, stop the sequence or change its wording and timing, one thing at a time.

Sources

Related articles

  • Email and automation

    Why do your company's emails land in spam, and how do SPF, DKIM and DMARC fix it?

    SPF, DKIM and DMARC are three DNS records that prove to mailbox providers that your emails really come from your company. Gmail and Microsoft require all three from senders of around 5,000 or more emails a day, and Gmail also watches spam complaint rates and easy unsubscribing. Set them up step by step, start DMARC at p=none and read the reports before you tighten the policy.

    9 min read

  • Email and automation

    AI chat assistants on your website, Messenger or WhatsApp: when do they help enquiries, and when do they hurt?

    An AI chatbot helps when it answers repeat questions, collects the details of an enquiry, books appointments and works out of hours. It hurts when it hides the human, gives wrong answers or collects data with no clear purpose. Connect it to your enquiry process and CRM, measure the share of conversations that become enquiries, handovers to a person and response time, and bear in mind that from 2 August 2026 the AI Act requires chatbots to say they are AI, unless that is obvious from the context.

    8 min read