Product-level events: for the handful of SKUs that actually run your business

Product-level conversion events are the sharpest instrument in this set, and the easiest to overuse. Fire one for every SKU and you will have hundreds of events, each with too little volume to optimise against, and a platform that cannot learn from any of them.

The rule: hero products only

Send product events for the small number of products that meet at least one of these tests.

• It is a gateway product - a disproportionate share of new customers arrive through it and then buy other things.
• It is a subscription or replenishment anchor, so the first purchase implies a stream of them.
• It has enough weekly volume to actually train a campaign, rather than dribbling one conversion every few days.
• It has a margin, an exclusivity or a stock position that makes it strategically different from the rest.

What you can do once they exist

1. 1Optimise a dedicated campaign toward buyers of a gateway product rather than toward all purchases.
2. 2Build a lookalike from that product's buyers, which is a far tighter seed than "everyone who ever bought".
3. 3Trigger a cross-sell sequence built on what your own customers actually bought next after that product.
4. 4Watch a hero product's conversion event fall and know a stock or price problem has started before the weekly report tells you.

What to do next

Rank your products by how many of your customers made their FIRST purchase on them, not by revenue. The top three or four on that list are your gateway products, and they are where this starts.