Check whether past actions worked and decide which ones to keep, change or stop.
Start every monthly marketing review with the list of actions from last month's report. For each one, check whether it was done and what it produced, and do that before you look at the new month's numbers.
Jun runs the monthly marketing reports for a business that sells sofas and advertises on Meta and Google. In early December he opens his draft of the November report. One of his three proposed actions is "refresh the Meta ad images". He has a feeling he has written it before, so he searches his old reports and finds it in August and again in October. In August the refresh was done and Meta sales did not move. In October nobody did it. Nobody noticed either time, because every report looked forward and none looked back.
A monthly routine that only sets new actions becomes a list of good intentions. When reports only look forward, repeated or failed actions go unnoticed.
Put the previous report's actions at the top of the review and ask two questions about each one: was it done, and what happened?
Done is a yes or no answer with a date. If an action was not done, say so plainly and note why, for example no time, waiting on someone else, or deprioritised. An action that keeps slipping tells you something. Either it matters and needs a different owner or more time, or it does not matter and should be dropped.
What happened compares the actual result with the result you expected when you set the action. One of Jun's October actions was adding the sofa range pages to the Google Ads campaign. He did it on 6 November, expecting more sofa enquiries from paid search. After three weeks they had risen from 4 to 11 (example figures). The result was clear enough to keep the change.
Review last month's actions before the new month's numbers because it keeps the review honest. It is harder to set three ambitious new actions when the page shows that two of last month's were never started.
Not every action shows its effect within a month, and judging too early is how good changes get abandoned. How long you wait depends on the channel. A change to an ad, a landing page or an offer usually shows something within days or weeks, provided there is enough traffic. Search rankings can take months to respond to new or improved pages, and SEO and AI search explains why. Email changes need several sends before a pattern is clear, especially with a small list. Anything that builds reputation, such as reviews or referral habits, works slower still.
When you set an action, write down the action, the date it will be done by, and what you will measure and when you will judge it. Match the judging date to how fast the channel responds. For example: rewrite the three sofa range pages, done by 15 November, and judge organic enquiries to those pages at the end of February. The December and January reviews mark it as done and still waiting. Until the judging date arrives, that is its status, and it is not recorded as a failure.
It works the other way too. A quick win in the first week may fade, so check it again a month later before you treat it as settled.
A decision log is a simple running record of your marketing decisions and the reasoning behind each one. It answers "why did we do that?" It has five columns: date, decision, reason, expected result and actual result. Leave the actual result empty when you write the entry, and fill it in once the result is known. Each entry takes about a minute.
Two rows from Jun's log show how it reads:
15 August: refreshed the Meta ad images because the click rate had been falling for six weeks. Expected the click rate and sales to recover. Actual result: the click rate recovered briefly and sales did not change. 1 September: cut the Meta budget from S$1,500 to S$1,000 a month because Meta payback was 11 months against 4 for Google search, from the unit economics sheet. Expected fewer Meta sales and roughly the same total sales. Actual result: total sales stayed flat and Meta sales fell from 18 to 14, which confirmed the expectation.
The log pays off months later. A new team member might suggest raising the Meta budget, or the owner might wonder why the newsletter went from weekly to fortnightly. The answer, and the evidence you had at the time, are in the log. It also stops the same debate from being re-run every quarter from memory.
Failed actions often follow the same course. An action fails, someone suggests trying it again with small changes, it fails again, and a few months later it comes back because it feels like the obvious thing to do. Jun's Meta ad refresh follows that course.
The stop rule is this: if the same action has been done properly twice and failed both times, stop doing it and write down why in the decision log, along with the next step you will try. Stopping is a decision like any other, and recording it means nobody has to rediscover the lesson. Jun's entry reads: stopped refreshing Meta images as a fix for falling sales, because it was done twice with no effect on sales either time. Next step: test a different offer instead.
The rule does not mean you can never revisit an idea. Markets change, and something that failed last year may work next year. If you come back to it, make that a deliberate choice with the log open, so the idea doesn't come back just because nobody remembered it had failed.
Jun's November report now opens with a short table of October's actions, showing whether each was done and what it produced. Getting your own log started with the last few decisions you can remember is the quickest way to make that table possible next month.
Set up a decision log with five columns: date, decision, reason, expected result and actual result. Enter your last three decisions.
Junxiong-WFG Organisation is an authorised representative of AIA Financial Advisers Private Limited (Reg. No. 201715016G).