Loading…

On strategy, where the expensive mistakes live

• Volume before architecture is the most expensive default in marketing. Publishing on a cadence into a site with no structure produces pages that compete with each other. The pages are not wasted individually, they are wasted collectively.
• The channel that works is rarely the channel you enjoy. Preference masquerades as strategy more often than anyone admits, and it is very hard to spot in yourself.
• Deciding what not to do is the harder half. Most plans are lists of everything that might help. A plan that does not name what you are deliberately skipping has not made any decisions yet.
• Positioning that has not stopped moving cannot be compounded. Content written against shifting positioning has to be rewritten, and deleted content that ranked costs you the links too.
• If the market is small enough, no amount of content creates demand. Check the arithmetic before committing a year to a channel.

Several of these are the cost of breadth without depth, which the full stack marketer piece takes apart properly.

On measurement, where the comfortable mistakes live

• Traffic is the metric most likely to rise while the business does not. It is also the easiest to show, which is exactly why it dominates reporting.
• Confirm the conversion is being counted correctly before optimising against it. A meaningful share of diagnosed traffic problems are measurement problems wearing a disguise.
• One blended number hides every segment that matters. Blended conversion rate, blended CAC, blended organic sessions. Split them and most problems name themselves.
• Brand traffic flatters everything it touches. Report it separately or your acquisition engine can be failing while every dashboard looks healthy.
• Agree which metric gets reported when, before the programme starts. Otherwise a good programme gets judged on lagging numbers during the quarter when only leading ones exist.

For the version of this that is a working system rather than a warning list, see the metrics worth reporting.

On execution, where the recurring mistakes live

• Specify linking rules in the brief, not after publishing. Links added at proofreading reflect what the writer remembered, which correlates with recency rather than with structure.
• Gating a guide removes its ability to rank, be cited, or earn links. Gate the artefact, let the explanation run free.
• The unglamorous tool beats the impressive one. A narrow utility that replaces a spreadsheet takes a week and tells you whether the channel works. The ambitious build takes six weeks and answers nothing.
• Comparison pages are uncomfortable to write and out-convert everything else. The discomfort and the performance have the same source: conceding where a competitor genuinely wins is what makes the rest credible.
• Updating a page at position eleven beats writing a new one from nothing. It is cheaper, faster, and it never feels like progress, which is why it gets skipped.

The audit version of this same mistake, buying a snapshot instead of a system, is covered in why most SEO audits do not compound.

On working with people, which is most of the job

• Set the timeline expectation on day one, not at the first review. Organic programmes get cancelled at month five, roughly one month before the curve turns. That conversation is only useful before the work starts.
• An engagement with no internal owner produces generic output. Not because the external team is weak, but because generic is all that is possible without access to the product and the customers.
• If your team cannot explain why a page is structured that way without asking you, you built a dependency rather than a system.
• Report the thing that is not working before someone finds it. Credibility survives bad numbers and does not survive discovering you knew.
• Most disagreements about whether marketing works are disagreements about what is being measured. Settle the definition and the argument usually dissolves.

Three of these are about saying an uncomfortable thing early. That is not a coincidence, and it is the pattern I would most want to hand to someone starting out.

The one that took longest

The lesson underneath most of the others: activity is not the same as compounding, and only one of them is worth funding.

A marketing function can be extremely busy, produce a large volume of defensible work, and build nothing that is worth more this year than last. Each individual piece survives scrutiny. The portfolio has no accumulating asset in it.

The question that separates the two is boring and it works: what will be more valuable in twelve months because we did this. If nothing on the list clears that bar, the list is maintenance. Maintenance is sometimes correct, but it should be named rather than mistaken for strategy.

Working out what is actually compounding is most of what I do in growth consulting engagements.