Advertising Push

Insights

Notes from operating
the channel.

Practical observations from running permission-led push campaigns. Opinionated, specific, and free of the word "synergy".

Permission

The prompt is the product

Teams spend weeks on creative and seconds on the opt-in prompt, then wonder why the list churns. The prompt determines who joins, what they expect, and how long they stay. Treat it as the highest-leverage asset in the programme, because it is.

Frequency

Fatigue is a cliff, not a slope

Engagement holds up, holds up, then falls away sharply. By the time the average metric moves, the most valuable segment has usually already gone. Watch your best segment, not your mean.

Creative

Curiosity gaps are a loan

Vague, intriguing copy lifts click rate and lowers everything downstream. You are borrowing attention against future trust, at a poor rate of interest.

Suppression

Opt-outs are cheaper than complaints

A visible, easy unsubscribe reduces complaints, and complaints damage deliverability and reputation far more than a lost subscriber does. Making it hard to leave is a false economy.

Measurement

Click rate is a vanity metric alone

Clicks with no downstream outcome measure only tell you the notification was interesting. Pair every engagement metric with what happened next, or you are optimising for the wrong thing.

Data

A provable list is worth more

Being able to show what each person agreed to, and when, is not just a compliance position. It determines whether a list survives a change of platform, a due diligence process, or a regulator's question.

Longer note

Why permission quality shows up in every other number.

Consider two lists of the same size. The first was built with a prompt fired on page load, offering nothing specific. The second was built after a moment of demonstrated interest, with a clear statement of what would be sent.

The second list will typically show higher engagement, lower opt-out rates, fewer complaints, better downstream conversion, and a longer useful life. None of these are separate problems to solve. They are the same problem, visible from different angles.

This is why we do not treat permission as a compliance task handled at the end. It is the first design decision, and it constrains the ceiling on everything measured afterwards.

A practical test: could you show a subscriber the exact wording they agreed to, and would they recognise it? If not, the rest of the programme is built on sand.

Want these applied to your programme?

Start a conversation