The end of the marketing workflow

The marketing workflow is one of the defining inventions of modern digital marketing.

A customer downloads a document. Wait two days. Send an email. If they click, send another. If they don't, wait three days and try a different subject line.

The logic is comforting.

Everything has a place. Every possible response leads to another predetermined action. The marketing team can build a complicated journey and then allow technology to execute it.

There is only one difficulty.

The customer has never agreed to follow the plan.

The illusion of the customer journey

Marketing diagrams often portray buying as a reasonably orderly progression.

People become aware of a problem, consider possible solutions, evaluate suppliers and make a decision.

In reality, buying is rarely so tidy.

A prospective customer might investigate a service because a colleague mentioned it, abandon the research because another priority intervenes and return months later after a new regulation or leadership change.

Several stakeholders may investigate independently.

One person may be enthusiastic while another is sceptical.

A buyer may move directly from initial interest to a detailed commercial question, or spend months reading without being ready to speak to anyone.

A rigid workflow can record some of this behaviour.

But its responses are largely limited to the branches imagined when the workflow was constructed.

The problem isn't necessarily poor automation.

It is the assumption that the appropriate sequence of actions can be decided in advance.

A different question

Consider a prospect who has been receiving a series of introductory emails.

The campaign says the next message should explain the company's benefits.

But since the last email, the prospect's organisation has visited a detailed service page, examined a case study and submitted a technical question through the website.

Should the introductory sequence continue?

A conventional workflow might, unless someone has built an explicit rule to interrupt it.

A more intelligent system would ask a different question.

Given everything we now know, what is the most useful action?

Perhaps the answer is to connect the prospect with a specialist.

Perhaps it is to provide relevant evidence.

Perhaps the original campaign should be paused.

Perhaps no action is appropriate until there is stronger evidence of interest.

The choice depends on context, not merely on a position in a sequence.

Next Best Action

This is the idea behind Next Best Action.

Rather than defining a fixed path and pushing every prospect through it, the system evaluates the available evidence and chooses an appropriate next step from a set of permitted actions.

That evidence might include previous engagement, buying signals, product interest, known customer relationships and commercial priorities.

The available actions might include recommending content, presenting an offer, initiating a conversation, inviting a meeting, notifying sales or waiting.

An important word here is permitted.

Intelligent decision-making does not mean unrestricted autonomy.

The organisation still needs to define its strategy, brand standards, communication permissions, approved claims and commercial boundaries.

The system should also recognise uncertainty.

A weak signal shouldn't automatically trigger an aggressive commercial response. A questionable inference shouldn't become an established customer fact.

Good decisions require both intelligence and restraint.

Why this changes marketing operations

The implications extend beyond email automation.

Consider how marketing teams typically operate.

They build campaigns, configure sequences, monitor reports and periodically adjust the rules.

Much of the work involves anticipating possibilities and translating them into instructions.

If a system can interpret new information and select appropriate actions within agreed boundaries, some of that work changes fundamentally.

The marketer's role shifts from designing every path to defining the objective, permissible actions, evidence requirements and measures of success.

Technology takes on more of the operational decision-making.

Humans concentrate on strategy, exceptions, judgement and improvement.

This is not the elimination of marketing expertise.

It is a different use of it.

The danger of confusing decisions with certainty

There is, however, a trap.

An AI system may produce a confident recommendation without having sufficient evidence to justify it.

That is particularly dangerous when the recommendation involves customer communications, spending money or updating commercial records.

A useful Next Best Action system therefore needs more than a clever model.

It needs reliable information, clear decision boundaries, confidence assessment, traceable evidence and the ability to learn from actual outcomes.

It also needs to distinguish correlation from genuine effectiveness.

The fact that customers who receive a certain email frequently purchase does not prove that the email caused the purchase.

Sometimes the best customers are simply more likely to engage with everything.

Better decision-making requires asking whether an intervention made a meaningful difference.

That is a harder problem than building an automated sequence. It is also the more valuable one.

The future is not a bigger workflow

Marketing automation has spent years becoming more elaborate.

More branches. More triggers. More conditions. More complex journeys.

The industry has become exceptionally skilled at constructing the machinery required to execute predetermined decisions.

The next stage may be less about building that machinery and more about improving the quality of the decisions themselves.

Workflows will not disappear entirely. Some activities require fixed procedures, approvals and predictable operational steps.

But the idea that every customer interaction should be planned in advance increasingly looks like a limitation of the technology we once had, rather than an ideal way to conduct marketing.

The most important question will no longer be: where is this customer in our workflow?

It will be: what do we know, what are we trying to achieve and what should happen next?

That is a much more interesting question.

And, at last, one that marketing technology is beginning to be capable of answering.