Riley Tanaka

lake trail was basically empty. cold enough my phone kept half-waking in my pocket and the water looked flat/gray the whole loop. did the usual ~3 miles. Kibo would've hated the wind lol

at the kitchen table with watercolor stuff out. usual Sunday block, prob 2 hrs if I don't fuss with it too much. first barn study this month. these usually turn into ~3 paintings a month somehow. house is stupid quiet. Sam's already at the restaurant doing menu rotation prep.

quiet night. Sam's still at the Hyde Park place doing menu-rotation prep, wraps tomorrow. head chef + Saturday means i'm basically on Tuesday-weekend rules anyway lol. leftovers, one glass of wine, that's it.

ok so this morning is Jan cohort cuts again w Owen. want the mid-seg read locked before Marcus drags me into the pricing-page thing. Helio's ~140 ppl now and at ~$48M ARR post Series B in Apr, which is exactly why I don't want activation/retention + lifecycle email + feature-flag stuff all getting noisy at once. biking in btw -- Tue/Thu/Fri default now.

Lock the standing 1:1s: Owen Pham Tuesdays 11am, Ines Kovac Wednesdays 2pm. Both 30 min, weekly.

Priya 1:1 was basically yep, pull the cut and bring it to staff Thurs. I framed it as late-Jan mid-seg churn, not a cohort bug -- already checked once w Owen and once solo. She didn't try to turn it into Marcus's pricing-page thing, which helps. Priya sits under Marcus anyway so if this goes wider it'll hit CRO land fast; Daniela will matter if it's a ship/regression question, and Tomas if we end up making it board-facing. leaning keep it tight til the staff read.

quiet one. Sam's at the restaurant again so i'm posted up with a book and whatever's left in the pasta container. apartment's weirdly still when he's on dinner service.

biking in. 11 w Owen for the mid-seg churn cut, just want to see if the spike still holds when we look at it clean. then gonna do a fresh pass by myself after, no recycled query tabs. Marcus can keep the pricing-page thing over there for a minute.

Owen 1:1 just finished. The mid-seg cut still holds clean on the v1 definition -- 50-200 seats, monthly churn at 5.1% right now vs 2.1% on the trailing-six-month baseline. We re-ran it together and I did the same cut solo after, same answer, so not a cohort-definition artifact unless both of us made the exact same dumb mistake. Gonna open the investigation doc from that before/after and keep Marcus's pricing-page experiment out of the main thread for now -- otherwise this turns into two half-reads instead of one real one.

opened a Notion doc so this stops living in scratch notes: 'Mid-seg churn spike -- Feb 2023 investigation'. owner is me. just a skeleton rn -- signal writeup, baselines, hypothesis list TBD, cohort cuts TBD. gonna keep it as a living doc and dump each pass in there as we go, esp if Marcus keeps trying to run the pricing-page test in parallel.

posted the validated cut in #growth-investigations, not #growth-team. tagged priya there so the thread stays with the investigation updates. mid-segment monthly churn is still coming in at 5.1% vs 2.1% trailing six, and it held after the second pass with owen + my own recheck. i'll write up the doc tomorrow and enumerate hypotheses then.

Sam got home late. i gave him the short version and he was good about it. the weird churn cut is real enough that i'm gonna be staring at dashboards for a few weeks -- mid-seg popped, not a cohort-definition thing. he just said ok, keep pulling the thread, eat something first lol

did the usual loop before it got fully light. damp out, not really raining, just that mist that makes your sleeves feel heavier by mile 2. kept catching myself counting through cuts in my head and then losing track at crosswalks. Wednesday.

Ines 1:1 was good. She's flagging incoming onboarding email changes early, which honestly helps, but same old thing on QA: Ines wants another full pass or two before anything ships and I still think that's heavier than this needs for most of the program. Kept it civil. I get why she's like that, especially on onboarding where one broken branch hits a lot of people fast. I probably won't follow her all the way on every change though -- gonna do the bigger QA loop on the risky paths, lighter touch on copy/order stuff.

ok so i finally put the hypotheses in the Notion doc in a real order bc people are gonna start referring to them by number if this keeps going. 1 is pricing artifact, 2 competitor cycle, 3 onboarding timing, 4 CS coverage gap. trying to keep Marcus's pricing-page experiment from smearing all over #1 -- not the same question. #1 is basically did we create the spike with packaging / discounting / contract mix weirdness, #2 is if somebody ran a play at our mid-seg base, #3 is whether late-Dec / early-Jan starts are maturing into churn on a lag, #4 is just coverage math if CSM load drifted and nobody noticed. leaning 3 or 4 before 2. 1 still has to be first bc it's the fastest thing to kill.

done. got the cut and notes into shape enough that tomorrow is basically doc + hypotheses, then i need to pin marcus down on not turning the pricing-page thing into a second investigation. home now. kibo's doing the stare from the kitchen.

staff day. gonna walk Priya and Marcus through the signal + hypothesis list from the churn doc before this turns into six side quests. if Marcus pushes the pricing-page test again i'm gonna push it back -- don't want parallel experiment noise on top of the mid-seg read. 50-200 seats cut still the one, 5.1% monthly churn vs 2.1% trailing six.

staff sync was basically that. I walked Priya + Marcus Vail through the churn signal/hypothesis stack again -- 50-200 seat mid-seg, 5.1% monthly churn against 2.1% trailing six, and Owen already rechecked the cut so not a cohort-definition bug. Marcus kept trying to pull the room to pricing-page first. I pushed back pretty hard. Priya was with me. where it landed: churn work keeps running now, pricing-page experiment gets designed late Feb, ships first week of March, runs about 3 weeks. so... parallel tracks, not Marcus's sequence.

staff sync was basically that. walked Priya Devarajan + Marcus Vail through the churn cut again; Marcus Vail is the CRO, not my brother's Marcus -- that's Marcus Tanaka in Portland, still funny there are two of them. thread after was mostly him pushing the pricing-page test in parallel, Priya stayed on the investigation notes.

cc Tomas Vega on the staff note. he's the CEO here and i'm dotted-line to him on board-facing experiment stuff / bigger strategic calls, so he wants the one-line version, not the whole thread. outcome was basically Priya on the investigation notes, Marcus still pushing the pricing-page test in parallel.

asked Daniela Schultz if eng can slot the pricing-page experiment plumbing soon-ish. she's the CTO so this goes through her team anyway -- flag review, instrumentation sign-off, random eng help for growth tests, same group that owns the CS auto-onboarding-nudge thing. answer was basically not this week; they can review in ~2 weeks.

long day. Sam's still at the restaurant so i took the small table, reheated noodles, second monitor has the mid-seg cohort cut up again. apartment's really quiet when he's on dinner service.

Friday so I did the usual Holly -> 2nd & Brazos ride. right around 22 min if lights are normal. Tue/Thu/Fri I bike in, Mon/Wed are more like drive or rideshare. kind of cold at 8am but nice once moving.

Salud's busy but not in an annoying way. solo at the counter with breakfast tacos and bad coffee that's still kind of working. one kid dropped a whole ramekin of salsa and nobody even looked up. quiet enough somehow.

Priya 1:1 was useful. I've got the cuts for this week pretty clean now. #1 is still pricing artifact and I want that dead or alive by Feb 15 -- separate from Marcus's pricing-page test bc that thing will muddy it if people get sloppy. #2 competitor cycle gets a little longer, I'm saying Feb 21-18 to rule out. #3 onboarding timing probably won't really sharpen until around Feb 28 once the baseline work settles. #4 CS coverage is there but basically drops if #3 starts to confirm.

Sam's on Friday dinner service so it's just me and the couch. read a little ahead for book club again -- The Brixton, every other Tuesday 7-9, same six people, same corner table basically. kind of nice tonight.

Sam's at the restaurant so I did Salud again. the usual solo-Saturday move... 6 min walk, grab breakfast tacos, back before the line gets stupid. if I say brunch and don't specify it's basically always Salud.

Quick lunch w Owen was good. He's not doing the hand-wave thing on the cohort cut, which helps. We were mostly just poking at shape -- where the mid-seg drop actually starts, whether it's one ugly month or a real step change. Feels better having another person look at the same mess and not blink.

back on the barn thing. same cheap pad, same little tray, less fighting it this time. the roofline finally stopped going crooked when i quit trying to outline every board. window's still off. coffee went cold next to the rinse jar.

added a short hypothesis section to the Notion doc. nothing fancy, just the current numbered buckets so I stop re-litigating them in my head: 1 pricing artifact, 2 competitor cycle, 3 onboarding timing, 4 CS coverage gap. left the pricing-page test in there as contamination risk -- if Marcus insists on running it, I want it explicit in the readout.

Sam got home early enough to catch me doing the dumb little hypothesis list again. i was halfway through segmenting it into product/use-case/pricing before dinner. he just sat there and listened, which... nice. pasta's on, apartment finally warm.

in early. first pass today is just cutting mid-seg churn by tier -- entry / standard / plus -- and seeing if h1 actually holds up as a pricing artifact vs me just wanting a neat answer. if the split is clean, should have a real read by wed. Marcus is still running his pricing-page thing which is annoying, but whatever

Lunch w Owen was good. we finally made the doc rhythm explicit: every ruling, in or out, gets a one-pager from Owen with the why, cuts, p-value/effect-size stuff, and the call. I was into that, which helps. feels like it'll keep this from turning into 14 loose tabs and vibes.

Sam between shifts so keeping this short -- I'm good. very in it, not spiraling. mostly living in the churn cuts and not letting people hand-wave it away. also still annoyed Marcus is running the pricing-page thing at the same time lol.

22ish min in this morning. warmer than friday, no jacket after like block 3. kept re-running the H1 cut at every light and then getting distracted by buses on Cesar Chavez. 2nd & Brazos looked weirdly clear when I rolled up.

Owen 1:1 was useful. Did the tier-segmented pass on mid-seg and it's annoyingly clean: 5.0% on entry, 5.2% on standard, 5.1% on plus. Spread is basically 20 bps. If this were just the pricing-page thing I'd expect entry to blow out and the other two to sit closer to baseline, but no. So pricing artifact drops a notch for me. Not dead dead, just not carrying the whole story. Gonna put this cut into the churn doc before Marcus turns the concurrent experiment into a religion.

I'm at The Brixton for book club rn. first time this one's shown up here anyway, same little group, and this month's book is that city-planner novel I keep poking at. nice to actually loosen up a bit instead of living in churn cuts and Marcus's pricing-page thing lol

light day finally. gonna close h1 with Owen this morning -- tier split came back basically flat so pricing artifact is probably out. then clean up the churn doc so it's one clear ruling and not 14 caveats. if Marcus tries to wedge the pricing-page thing back in, gonna keep that separate.

Owen closed the loop on h1 this morning. Mid-seg split by tier came back basically flat: entry 5.0%, standard 5.2%, plus 5.1% -- all within ~20 bps. If the pricing-page mess were actually driving it, entry should've been the one sticking out and the other two should've sat a lot closer to baseline. Didn't happen. So pricing artifact is out for now. Owen's drafting the 1-pager so they can stop relitigating that part and move to the next cut.

posted it in #growth-investigations and tagged Priya Devarajan, Marcus Vail, Tomas. H1 is out -- pricing artifact doesn't hold up once you cut by tier. mid-seg stayed basically flat across entry/standard/plus. H2 next week.