Guides

A two-minute guide to giving Bytown admin access to your Google Ads account — find Access and security in the new or old layout, invite us, and pick the right permission tier.

Step-by-step guide to granting Bytown access to your Meta Business Manager — create an account, add us as a partner, or invite a user, with the exact toggles to enable.

There is no official taxonomy of ad formats, and grouping them by layout produces a useless list. Here they are grouped by the job the creative does instead — which is what predicts where it belongs and whether it survives a cold audience.

Targeting used to be the lever. The platforms took it back, and what is left in your control is the ad itself. Here is what the evidence says about making one — including the four famous numbers that fall apart the moment you look for the source.

You have under two seconds in a mobile feed, recall roughly doubles between a glance and ten seconds, and the share of ads getting full attention has collapsed since the early 1990s. Here is what is measured and what is folklore.

Account to live campaign in six steps, with the exact fields, the character limits, and the four things that catch people out the first time.

Restricted is not prohibited. OpenAI's ad policy names two legitimate ways into a gated category, and one very fast way to lose the account. Plus what actually happened when we applied: refused every time, appealed every time, restored every time.

Most write-ups say healthcare, finance and legal are banned from ChatGPT Ads. OpenAI's policy says something more useful — those three are approved-advertiser categories reviewed case by case. Here's who is genuinely blocked, how the door opens, and what to run while you wait.

Every roundup publishes the prompts. Almost none publish what happens when you run them on a live account. These ten are the ones I actually use — each with the failure I have hit, because knowing where a prompt lies to you is worth more than the prompt.

Two independent bodies of research rank creative far above targeting, and they were published years apart using completely different methods. Then Apple removed the signal and the platforms automated the buying, and the gap stopped being academic.