PDCA in B2B marketing: how to close the improvement loop

The question we hear most often from marketing teams in the region is this: if we run campaigns, measure results, and review reports every month, why does the PDCA cycle still not close in our company? The answer rarely has to do with tools. It has to do with the fact that the PDCA cycle is started in the company but never closed. Plan and Do absorb all the team's energy, while Check and Act get filed under "when there's time".

Where the PDCA cycle breaks down in practice

In a typical month of work for an agency or a B2B marketing department, it looks much the same. At the start of the month an action plan is drawn up: a LinkedIn campaign, a new landing page version, two webinars, an email send to the database. Then the team executes. At the end of the month someone compiles the numbers in a spreadsheet and emails them. Nobody returns to the question of which hypothesis those numbers confirmed and which they disproved.

The problem is not a lack of data. It is that the plan contained no hypothesis, so there is nothing to verify. If the goal reads "increase the number of leads", then after a month all you can say is that there were more leads or fewer. You do not know why, so you do not know what to repeat.

The second reason is organizational. Continuous process improvement requires someone who owns the loop, not just the task. In companies where marketing is handled by one person plus an agency, that role usually does not exist. That is why the interim marketing manager model so often works, with someone from outside responsible for closing the cycle, not just for delivering materials.

It is also worth distinguishing two things that often get confused. Campaign optimization means reacting to current results while a campaign is running. PDCA is a broader mechanism that changes how the team works in the quarters that follow. The first operates on a scale of days, the second on a scale of months. Mixing them means the team is busy putting out fires and has no time to draw systemic conclusions.

Plan: a hypothesis instead of a task list

A good Plan phase in the PDCA cycle has one characteristic: it can be disproved. Instead of a list of actions it contains a sentence in the form "if we do X, then Y will increase by Z, because we believe that...". It sounds trivial, but in practice it changes everything, because it forces you to name an assumption that can be tested.

An example from work with a client in the manufacturing sector. The plan for the quarter read: "we want more inquiries from distributors". Once turned into a hypothesis it looked like this: "if we prepare a separate landing page with a margin calculator for the distributor and send it in a campaign to 400 contacts from the database, then in our example scenario the number of inquiries will rise from 6 to 12 in the quarter, because distributors do not know how much they will actually earn on our line".

The second version has three advantages. First, it points to a specific mechanism that can be verified in a conversation with a salesperson. Second, it gives a number to work from. Third, if the result comes out at 7 inquiries, it is still clear what to do next, because it is clear what failed.

It is worth keeping in mind the hierarchy of purchase criteria. In the book Scaling Your Startup Peter Cohan describes how companies win when they consistently outperform rivals on the criteria the customer themselves ranked as most important, such as quality, price, or breadth of offering (Cohan, 2019). In B2B marketing this translates into a simple thing: the hypothesis should concern a criterion the customer actually mentions in a sales conversation, not one that is convenient to measure in a tool.

Do: small scope, short time, clear owner

The Do phase of the PDCA cycle most often falls apart because of scale. The team tries to test three hypotheses at once, in three channels, across the entire database. The result is that after a month, no one knows which element worked and which did not.

A practical rule we apply in projects: one hypothesis, one channel, one audience group, four to six weeks. If a test requires a longer period, it usually means the hypothesis is too broad and needs to be broken down into smaller ones.

The second element is the owner. This is about someone who is able to say after the test ends whether the hypothesis was confirmed. In practice this means the owner must have access to sales data, not just campaign statistics. Without that, the Check phase turns into a story about clicks.

The third element is writing down the boundary conditions before the start. If the cost per inquiry exceeds a certain level, we stop the test. If after two weeks there is no inquiry at all, we check whether the problem is in the messaging or in the audience list. These conditions set in advance protect against a situation where the team plows ahead with an action only because it has already invested time in it.

Check: what to measure so the conclusions make sense

The Check phase in plan do check act is most often skipped, because it requires confronting numbers that may be uncomfortable. To be useful, it must answer three questions: what happened, why we believe so, what it means for the next step.

The first question is simple, as long as the data is collected in one place. The second is harder, because it requires reaching for qualitative sources. In practice this means talking to the salesperson who called those contacts, or reading a few responses from the form. Without this stage, the team guesses rather than draws conclusions.

The third question is the most important, because it leads directly to the Act phase. If the hypothesis was confirmed, the decision is: we scale it and write it into the standard. If it was not confirmed, the decision is: we drop it and record why. If the result is ambiguous, the decision is: we repeat the test with one element changed.

It is worth noting a trap that the literature on advertising in new formats describes well. Haiming Hang and Agnes Nairn show that the effectiveness of communication depends on how the recipient recognizes the sender's intent, and not only on the message itself (Hang and Nairn, 2018). In practice this means that if a test performed poorly, it is worth checking not only the content, but also whether the recipient even recognized that this was an offer directed at them. This often explains the difference between two campaigns with similar content.

Act: a decision instead of a report

The Act phase is where the PDCA cycle either closes or turns into an annual report. The difference is simple: a report describes what happened. A decision changes what the team will do in the next cycle.

In practice, the Act phase should end with three things written down in a single document. First, a list of actions that become part of the working standard, for example a new brief template or a changed order of sections on a landing page. Second, a list of things abandoned, so that nobody comes back to them six months later without a reason. Third, one new hypothesis for the next cycle, derived from what we learned.

This document does not have to be long. In the projects we run, it is usually one page. What matters is that it is updated at the same rhythm as budget planning, because otherwise marketing conclusions never reach decisions about allocating funds. This is the moment that connects operational work with B2B marketing strategy, which without such feedback becomes a document on a shelf.

It is also worth making sure that decisions from the Act phase are visible to the sales team. If marketing abandoned a channel that salespeople considered promising, they should know why. If a new standard was introduced, they should know what to expect. Without this communication, the PDCA cycle operates in one part of the company while the rest of the organization lives by its own rhythm.

How to implement PDCA without rebuilding the whole agency

You do not have to start by changing the team structure or introducing a new reporting system. It is enough to adopt three habits that can be sustained under the current workload.

1. One cycle per quarter, not per month. A monthly rhythm is too short to collect sales data and draw conclusions. A quarter gives time for a test and for a conversation with salespeople.
2. One hypothesis per cycle. If the team wants to test more, let them write the remaining ideas into a queue and return to them next quarter. Parallel tests without a common denominator usually produce no conclusions.
3. A Check meeting scheduled in the calendar in advance. Planned at the moment the test starts, with a specific owner and a specific question going in. If the meeting is not in the calendar, it will not happen.

On top of that, it is worth adding a simple table in which each cycle has its own row: hypothesis, scope, result, decision. After four quarters, such a table reveals more about how marketing works in the company than many an annual report. It shows which assumptions keep coming back, which channels consistently fail, and where the team repeats the same mistake.

When it is worth bringing in outside support

The PDCA cycle requires three things: time, discipline, and someone who looks at the data from a distance. In companies where marketing is handled by one person, all three are worth their weight in gold. That is when it is worth considering a model in which an experienced marketer joins the team part-time and is responsible for closing the loop rather than for producing materials.

Growth Group operates in two models: full agency service and an interim marketing manager compensated on results. Internationally, the second model works especially well for companies that already have an execution team but lack someone responsible for conclusions and decisions. You can read more about how this collaboration works on the interim marketing manager.

The point is not for an external person to take over all tasks. The point is for someone to ensure that after each cycle a decision remains, not just a presentation. That difference is what makes continuous process improvement stop being a slide in a deck and become the team's way of working.

The PDCA cycle does not require new tools. It requires one person to make sure the loop is closed, and one meeting in the calendar where a decision is made.