I'll publish all of it when the fortnight ends.
Including if the answer turns out to be "not much", which is a real possibility.
https://t.co/D2BSFXn3b5
https://t.co/IU66sLPrGs has a distribution channel that opens by default for anyone installing a plugin.
It's been running since March.
Nobody has ever published what it actually does to a plugin's numbers.
I'm in it right now, so I'm going to.
One thing I've already found that I don't think is documented anywhere:
The order of the eight plugins is reshuffled every day. Same eight, different sequence.
So position in that list is a draw, not a ranking. Which hands me a free randomised experiment.
Type "X without Y" into your own site search. Then type just "X".
If the first number is bigger, your search engine just treated the thing your
customer excluded as something they wanted more of.
I tested 5 WordPress sites this week. 4 of 5 did exactly that.
One went from 1,340
Queryra just became the #1 result for "semantic search" on https://t.co/IU66sLPrGs — ahead of plugins many times its size.
Algorithmic ranking, not an award. Verify: https://t.co/Ual7dcCALN, search "semantic search". Screenshot dated, because rankings move.
Before you buy an AI search plugin for WordPress, three questions worth asking:
Who pays for the API calls?
What happens when you hit the search limit?
Is it actually semantic, or keyword matching with an AI label on it?
I wrote up the fine print here: https://t.co/1MvuLBKYCg
Your search log is the cheapest customer research you own.
It also lies until you clean it.
Full write-up, including what to do with your own log this week:
https://t.co/8jqztYGvWi
I logged every search that hit my API for four months.
155,730 queries.
Under 3% came from a real store.
Here is what the other 97% was, and why it matters if you run a shop.