How to prioritise marketing tasks when everything looks urgent

Aug 22
3 min read

Updated: 6 days ago

Introduction

The list is always longer than the week. Redo the website, run ads, fix the reviews, start posting again, sort out the pricing, build an email list.

Every item is defensible, which is exactly the problem. Defensibility is not a ranking, and without one the list gets worked in the order things were remembered.

1. How to prioritise marketing tasks: start with what unblocks other work

The first question is not impact. It is what else depends on this.

Recording enquiry sources is a small, dull task that determines whether any channel decision you make for the next year is informed. It ranks above a website redesign that impresses nobody but the owner.

Dependency ordering is unglamorous and it is where most of the available gain sits.

2. Prefer reversible over irreversible

Between two tasks of similar value, do the one that is easier to undo.

A price test on one product is reversible. Rebuilding the site is not. Trying a channel with a capped budget is reversible. Signing a twelve-month contract is not.

Reversible work lets you be wrong cheaply, which matters because you will be wrong regularly. This is also why "test it small" beats "commit properly" as a default.

3. Rank by evidence produced, not by effort saved

A task that answers a question is worth more than a task that produces output.

Two hours working out what a customer is worth over a year changes every budget decision afterwards. Two hours writing social posts produces two hours of posts.

When the list is long, prefer the item that will make the rest of the list shorter by ruling things out.

4. Distinguish the leaking bucket from the empty one

Fixing conversion beats adding traffic, and it is not close.

If half your enquiries never get a reply, doubling enquiries doubles the waste. If the form does not work on a phone, every pound of ad spend is discounted by however much of your audience is mobile.

Check the bucket for holes before deciding it needs more water. The holes are usually free to fix.

5. Weight by frequency, not by size

A small task that recurs beats a large one that happens once.

Asking every customer for a review is a trivial change with a compounding result. A launch campaign is a large effort with a single result. Over a year the trivial recurring habit usually wins.

So when ranking, ask how often the benefit repeats. Anything that becomes a routine ranks above anything that becomes an event.

6. Cap the list at three

Not because three is magic, but because a list of fifteen is a way of not choosing.

Three items, each with a named owner and a date. Everything else goes on a second list you are allowed to look at only when one of the three finishes.

Long lists feel responsible and produce the same output as no list, spread thinner.

7. Schedule the boring maintenance separately

Some tasks never win a priority contest and still have to happen.

Checking that automations ran. Confirming the reporting connection is live. Reviewing what the tools cost. None of these will ever beat a growth initiative on a scoring matrix, and skipping them is how a working system stops working.

Give maintenance its own recurring slot so it does not compete with new work.

8. Re-rank on a schedule, not continuously

Constant re-prioritising is indistinguishable from having no priorities.

Set the three items, work them, and re-rank at a fixed point — weekly or monthly. Between those points, new arrivals go on the second list unless something is actually broken.

This is the difference between responding to evidence and responding to whoever spoke last.

Conclusion

Order the list by what unblocks other work, prefer reversible tasks, and favour the ones that produce evidence over the ones that produce output.

Fix leaks before adding volume, weight recurring habits above one-off events, cap the active list at three with names and dates, give maintenance its own protected slot, and re-rank on a schedule rather than every time something new is suggested.