Good that you noticed these .....I was about to point out these nuances - then kept to myself.

Industrial Production has basically stalled post the globalization - but the YoY% measure is a good way to judge output trajectory.
The accompanying Capacity Utilization - provides a 2nd degree view - and its been falling. However, its noisier as a Recession marker.

St. Louis Fed (FRED): Capacity Utilization: Total Index (TCU) | FRED

The reason why RRSFS looks noisy is because of inflation.( that series is inflation adjusted)

Ray .... GTT is a good paper - but its Economic overlay and parameters are all inclusive of 2014 data ie In-Sample. And the ONLY OOS data point of recession since then is COVID and it was practically set up for failure against that kind of an environment.

Here's the main unadjusted series - noisier but it judges actual $s people are spending:

St. Louis Fed (FRED): Advance Retail Sales: Retail Trade and Food Services (RSAFS) | FRED

In this one - the edge ie MoM trajectory matters a bit more -but there is no definitive good-fit. Its slowing clearly.

This one is a Coincidental one - but you need to see at least some rise on it St. Louis Fed (FRED): Smoothed U.S. Recession Probabilities (RECPROUSM156N) | FRED

The supposedly best one newyorkfed.org - Capital markets
This is the ONE I track and have also implemented on my side.
NOTE: Its in sample till the bottom in 2009. But this is ALSO the ONLY ONE - which called the COVID ( Not exactly the pandemic) -Recession. And has not had any success with the recent inversion rates.

Thumbrule: It needs to be above 30% - which at current 27% is Below. NOTORIOUSLY - this also gets adjusted ie revised (but not a lot- its main input is the spread) - so you need to capture it real-time. Hence I decided to track it on my side too.

Hope this helps!

best