JIT Weekly Report — July 12, 2026
Version 2. Rebuilt after Sundeep caught a real mistake: v1 wrote the advice first and stuck expert names on afterward. This version loaded the expert playbooks before deciding anything — and two of the answers changed. Window: Jul 5–11 vs Jun 28–Jul 4. Written in plain English.
The whole report in 6 sentences:
You spent $189 on ads and got $107 back. You lost about $82.
The ads are great. The signup page is great. More people joined your list than ever.
The problem is at the very end: people reach the checkout page and don't finish paying. That's over half of them, worth about $839 a month.
The email that's supposed to chase those people has never sent. Not once. Zero people.
That's the one thing to fix. Everything else on this list is smaller.
You spent $189 on ads and got $107 back. You lost about $82.
The ads are great. The signup page is great. More people joined your list than ever.
The problem is at the very end: people reach the checkout page and don't finish paying. That's over half of them, worth about $839 a month.
The email that's supposed to chase those people has never sent. Not once. Zero people.
That's the one thing to fix. Everything else on this list is smaller.
1 · The Numbers
🔴 The big one: more people came in, fewer people bought.
88 new people joined your email list this week. Only 5 of them bought anything. Last week 73 people joined and 8 bought.
So the number that matters — out of everyone who joins, how many buy — dropped from 11 out of 100 to about 6 out of 100 MEASURED — Kit + Stripe, N=5.
It now costs you $34.76 to get one sale. Last week it cost $17.44. That's twice as expensive.
Important honesty note: 5 sales vs 8 sales is a tiny difference. At these numbers, luck moves things around a lot. This might be a real problem, or it might be noise. Do not panic and do not rebuild anything based on this one week. But it does mean: don't raise the ad budget. The rule was "raise budget after two good weeks in a row." We got one good week, then a bad one. Budget stays where it is.
88 new people joined your email list this week. Only 5 of them bought anything. Last week 73 people joined and 8 bought.
So the number that matters — out of everyone who joins, how many buy — dropped from 11 out of 100 to about 6 out of 100 MEASURED — Kit + Stripe, N=5.
It now costs you $34.76 to get one sale. Last week it cost $17.44. That's twice as expensive.
Important honesty note: 5 sales vs 8 sales is a tiny difference. At these numbers, luck moves things around a lot. This might be a real problem, or it might be noise. Do not panic and do not rebuild anything based on this one week. But it does mean: don't raise the ad budget. The rule was "raise budget after two good weeks in a row." We got one good week, then a bad one. Budget stays where it is.
🟢 Your emails are on fire.
Three emails actually went out this week (the auto-scheduling fix is holding). Tuesday's got opened by 46.9% of people and 6.4% clicked. Normal for this industry is 21–28% opened and 2.5–3.5% clicked. That's your best email ever OBSERVED — Kit broadcast 24840951. Your list is now 632 people, and all 88 new ones this week are real (no fake test emails).
Three emails actually went out this week (the auto-scheduling fix is holding). Tuesday's got opened by 46.9% of people and 6.4% clicked. Normal for this industry is 21–28% opened and 2.5–3.5% clicked. That's your best email ever OBSERVED — Kit broadcast 24840951. Your list is now 632 people, and all 88 new ones this week are real (no fake test emails).
🟢 The Instagram robot works, and ManyChat is gone.
Someone commented "Work" on your Instagram on Jul 9. The new robot saw it, sent them a DM with the free guide, and logged it as OK OBSERVED — responder log, row 12. That's the first real one since Jun 27. ManyChat (the old $17/month tool) is cancelled and you're not missing anything.
Someone commented "Work" on your Instagram on Jul 9. The new robot saw it, sent them a DM with the free guide, and logged it as OK OBSERVED — responder log, row 12. That's the first real one since Jun 27. ManyChat (the old $17/month tool) is cancelled and you're not missing anything.
2 · What Changed This Week
| What | When | What happened | Did it get written down? |
|---|---|---|---|
| EV-2026-07-08-01 | Jul 8 | Fixed a real bug: buyers who lost their download link had no way to get it back. Added the link to 6 emails and the thank-you page. | ✅ Yes |
| Creative test built | Jul 8 | New ad campaign with 4 separate budgets so each new video gets a fair shot. 2 turned on. | ❌ No entry in the change log |
| Test shuffled | Jul 11 | You paused the layoff video, doubled the "Stay Positive" picture ad to $10/day, and turned on the "Stay Positive" video. | ❌ No |
| ManyChat cancelled | Jul 11 | Subscription ended. Billing stops Jul 18. | ❌ No |
| Next content wave built | Jul 10–11 | 27 posts scheduled for Jul 17–26. All checks passed. | ❌ No |
| Friday planner email | Jul 10 | Went out to 14 people as planned. 5 opened, 1 clicked. | ✅ Yes |
Small problem: 4 real changes shipped with nothing written in the change log. And the notes file points to event IDs (EV-2026-07-08-01/-02) for the ad test that don't exist — that ID belongs to the download fix. The change log is how we figure out why a number moved. It's drifting. (Fix in §9, R6.)
3 · The Ads
| Campaign | Spent | People who clicked the link | Cost per click | Signups | Sales | |
|---|---|---|---|---|---|---|
| SALES ENGINE (your main one) | $139.02 | 7.4% of viewers | $0.24 | 91 | 4 | |
| CREATIVE-TEST (new videos, started Jul 8) | $44.15 | 2.4% | $1.03 | 10 | 0 | |
| Planner test | $6.26 | 1.3% | $3.13 | 0 | 0 | |
| Normal for this kind of business: 1.5% of viewers click (2.5% is "really good", 4% is "amazing"), and each click costs 80¢ to $1.50. You get 7.4% of viewers clicking at 24¢ a click. Your ads are not the problem. They're one of the best parts of the business. | ||||||
The four ads still running
| Ad | Spent | Link clicks | Sales | Read | |
|---|---|---|---|---|---|
 |
TrialWinner | $138.22 | 7.4% | 4 | Still your whole business. It ate 99.4% of the ad money. |
 |
TEST — StayPositive picture | $19.32 | 3.2% | 0 | Best of the new ones. Still less than half as good as TrialWinner. |
 |
TEST — StayPositive video | $8.51 | 2.0% | 0 | Only started showing Jul 11 night. Too early to judge. |
 |
TEST — Layoff video YOU PAUSED IT JUL 11 | $16.32 | 1.2% | 0 | Good call. $5.44 per click. Way too expensive. |
4 · Is Any Ad Being Starved? (the rule we added last week)
"Starved" means: an ad is switched ON but nobody is seeing it, because another ad in the same pot of money is hogging everything. A starved new ad isn't a bad ad — it just never got a chance.
| Ad | How old | Views in 3 days | Verdict |
|---|---|---|---|
| TrialWinner | 37 days | 1,660 | ✅ Fine — it's the one making money |
| StayPositive video — the copy inside SALES ENGINE | 6 days | 0 | 🐌 STARVED — and worse, it's fighting itself (see below) |
| POVTrust | 37 days | 0 | 😴 Old and asleep. Leave it. That's the rule. |
| TEST — StayPositive picture | 4 days | 413 | ✅ Getting seen. The separate-budget fix worked. |
| TEST — StayPositive video | 4 days | 153 | ✅ Getting seen now |
The same video is running in two places at once, and the two copies are competing.
The plan on Jul 8 said: when you turn a video ON in the test campaign, turn OFF the copy of it in the main campaign — so it isn't bidding against itself. That got done for two videos. It got missed for the "StayPositive" video when it was switched on Jul 11 OBSERVED — live ad list, re-checked today.
Why this matters (Denney): Facebook now fingerprints each ad. Two identical videos = one fingerprint = one ticket to the auction. They don't get two shots — they split one.
Fix: turn OFF the copy in SALES ENGINE (
The plan on Jul 8 said: when you turn a video ON in the test campaign, turn OFF the copy of it in the main campaign — so it isn't bidding against itself. That got done for two videos. It got missed for the "StayPositive" video when it was switched on Jul 11 OBSERVED — live ad list, re-checked today.
Why this matters (Denney): Facebook now fingerprints each ad. Two identical videos = one fingerprint = one ticket to the auction. They don't get two shots — they split one.
Fix: turn OFF the copy in SALES ENGINE (
120248739739550465). Leave the test copy running. This does not kill the test — the test is the one with its own budget.
5 · Where People Actually Fall Off
| Step | This week | How that compares |
|---|---|---|
| Saw your ad | 5,159 times | — |
| Clicked the link | 330 people (6.4%) | 🟢 4x better than normal |
| Page actually loaded | 299 (91% of clicks) | 🟢 Great — nothing broken |
| Gave their email | 90 (30% of visitors) | 🟢 Top of the normal range (15–30%) |
| Bought something | 5 | 🔴 This is where it breaks |
| Every single step is healthy until the last one. The ads work. The page works. The signup works. Then people don't pay. Also: zero people bought the $19.95 Templates this week (3 did last week). The middle of your product ladder went silent. | ||
6 · What The Experts Actually Said (loaded BEFORE any advice was written)
Why this section exists. Last night's version of this report wrote the advice first and pasted expert names on top afterwards. That's backwards, and it's against your own Rule #7. So this time the playbooks were opened first — and two conclusions flipped.
Product / Growth CPO lens — find the ONE thing blocking growth
The rule: at any moment there's exactly one broken link in the chain, and work spent anywhere else is mostly wasted. Walking your funnel:
• Getting traffic? Excellent. Not it.
• Getting signups? Excellent (30%). Not it.
• Getting paid? Over half the people who reach the checkout page never finish. That's ~$839 a month walking away — about the same as everything you actually earn.
• Keeping people? Also weak, but it can't matter until they pay you once.
The binding constraint is checkout completion. Not the ads. Not the creative. Not the coach.
• Getting traffic? Excellent. Not it.
• Getting signups? Excellent (30%). Not it.
• Getting paid? Over half the people who reach the checkout page never finish. That's ~$839 a month walking away — about the same as everything you actually earn.
• Keeping people? Also weak, but it can't matter until they pay you once.
The binding constraint is checkout completion. Not the ads. Not the creative. Not the coach.
Dara Denney — performance creative
Her diagnosis chart, read literally on your numbers:
You have a 7.4% click rate and a $34.76 cost per sale. So by her own tree, making more ad videos is the wrong project right now.
Second thing she'd flag: her minimum test budget is $30–50/day per ad set for Meta's system to learn anything. Your test ad sets are at $5 and $10/day. The test literally cannot give you a trustworthy answer at that budget — you're paying $15/day for a result you can't use.
Third: her budget split is 70% winners / 20% tweaks to winners / 10% brand-new tests. You're at roughly 57% winner / 43% test. Way over-testing for your size.
Good click rate + bad cost-per-sale → it's an OFFER or PAGE problem. NOT a creative problem.You have a 7.4% click rate and a $34.76 cost per sale. So by her own tree, making more ad videos is the wrong project right now.
Second thing she'd flag: her minimum test budget is $30–50/day per ad set for Meta's system to learn anything. Your test ad sets are at $5 and $10/day. The test literally cannot give you a trustworthy answer at that budget — you're paying $15/day for a result you can't use.
Third: her budget split is 70% winners / 20% tweaks to winners / 10% brand-new tests. You're at roughly 57% winner / 43% test. Way over-testing for your size.
Lifecycle email operator
"Abandoned checkout is often the highest-earning email flow in the whole business, because the person already showed you they wanted it. Fire it within about 1 hour, then again at +1 day, +2 days."
Yours fires on Stripe's
Also: healthy businesses get 15–25% of revenue from email. You get roughly 0%. Your emails get opened at 47% — the audience is there. Nothing is being sold to them.
Yours fires on Stripe's
checkout.session.expired — which Stripe only sends roughly 24+ hours later. So even on the day it starts working, it's built to be late. And right now it isn't working at all: 0 people have ever entered it OBSERVED — Kit, re-checked today: total_subscriptions = 0.Also: healthy businesses get 15–25% of revenue from email. You get roughly 0%. Your emails get opened at 47% — the audience is there. Nothing is being sold to them.
Joanna Wiebe — conversion copy
"The best copy is not written. It's found, in the words your customers already use." She spends 60–70% of copy time on research.
Here's what nobody noticed: you started recording exactly that, 7 days ago. The Bailey coach now saves what people type. This week people typed things like:
Here's what nobody noticed: you started recording exactly that, 7 days ago. The Bailey coach now saves what people type. This week people typed things like:
"Getting ahead of the game before i take a management position."
"The whole team." (when asked what slipped)
That is free, first-party customer language — the thing she says is the whole ballgame — and it has never been used in a single ad, headline, or email. That's a real opportunity we were sitting on.
Ramit Sethi — buyer psychology
"Write to the real fear, not the stated want." The stated want is "help me plan my week." The real fear underneath your coach transcripts is "I'm about to be responsible for people and I don't know if I can do it."
He'd also say your coach's first question is a qualifying question asked at the wrong moment — you're asking someone to do homework before you've given them a single reason to trust you.
He'd also say your coach's first question is a qualifying question asked at the wrong moment — you're asking someone to do homework before you've given them a single reason to trust you.
Alex Hormozi — the value equation
Half your buyers quitting at a $9.95 checkout is not a price problem. Nobody hesitates over ten dollars because of the money. It's the bottom half of that equation: effort and doubt. Fix friction and belief at checkout, don't touch the price.
Value = (Dream outcome × Belief it'll work) ÷ (Time it takes × Effort it costs)Half your buyers quitting at a $9.95 checkout is not a price problem. Nobody hesitates over ten dollars because of the money. It's the bottom half of that equation: effort and doubt. Fix friction and belief at checkout, don't touch the price.
7 · Bailey the Coach — What People Actually Typed
Bailey opens by giving people homework, and they leave in 4 seconds.
People who clicked in from your email: 49 opened it. 45 got a reply from Bailey. 1 got a plan. They stayed about 4 seconds.
People who came from the signup page: 27 opened. 6 got a plan. They stayed 131 seconds MEASURED — coach-stats, 7 days.
Why? Everyone from the email lands on the same starter, and Bailey's very first line back is:
The proof is almost funny. One person (
People who clicked in from your email: 49 opened it. 45 got a reply from Bailey. 1 got a plan. They stayed about 4 seconds.
People who came from the signup page: 27 opened. 6 got a plan. They stayed 131 seconds MEASURED — coach-stats, 7 days.
Why? Everyone from the email lands on the same starter, and Bailey's very first line back is:
"What were 1–2 wins from this week — and what's one thing that slipped?"
That's a two-part written essay question, asked before Bailey has given them anything. So they close the tab.The proof is almost funny. One person (
anon:8dca5501) fired that starter three separate times across two days, got the exact same question every time, and never typed a single word back OBSERVED — daily digests Jul 10 + Jul 11. They told you three times.
| Person | What they said or did | What happened |
|---|---|---|
anon:8dca5501 | Fired the canned planner starter 3× in 2 days. Never replied once. | 🔴 Left, 3 times |
anon:cb675182 | Said "The whole team". Bailey gave its best answer of the week — named two likely causes, handed over an exact question to ask the team. | 🟡 Still left |
anon:87e0f141 | "Getting ahead of the game before i take a management position." Then asked for a reusable template. | 🟢 Got a plan. Templates offered. |
| 4 more people (Jul 9) | All got the same wins/slips question. All went silent. | 🔴 Left |
The pattern is simple: people who bring a real problem in their own words stay and finish. People who get handed a canned planning question leave. Every single time.
8 · Email, Website Health, Content
Email — the best channel you have
| Sent to | Opened | Clicked | |
|---|---|---|---|
| "The day I stopped answering my team's questions" | 450 | 46.9% | 6.4% |
| "She got promoted 3× in 3 years" | 537 | 27.4% | 2.6% |
| "60 seconds to plan next week?" (planner nudge) | 14 | 35.7% | 7.1% |
| Normal: 21–28% opened, 2.5–3.5% clicked. All three beat it. And all three sold nothing. That's not a copy problem — there's no selling flow attached to any of them. | |||
| Automatic sequence | People in it | Status |
|---|---|---|
| Soap Opera (new signups) | 205 | ✅ Running |
| Abandoned checkout | 0 | 🔴 Has never fired. 13 days. |
Website health (Clarity, last 3 days)
212 visits. 0 rage-clicks (nobody angrily mashing a button). People scroll 61–96% down the page. Everyone is from the US, Australia or Canada — exactly who you're paying for. Nothing is broken on the site. MEASURED — N=212
Content
21 posts went out on Instagram and Facebook this week. They got 4 likes and 0 comments total. Organic is flat. That's not new, and it's not a crisis — organic is where you find hooks (your best ad started as an organic post), not where you get customers. Treat it as a lab, not a shop.
Comment keywords all check out. Your posts used WORK (×5) and PLAN (×1). Both are live in the robot's list OBSERVED — live diag check today. Nothing is orphaned.
Format check (Rule #10): next two waves both PASS. This week's published feed was 8 videos out of 13 posts (62%) — over the limit — because the daily trial-reel lane posts outside the plan and dodges the check. Same finding as last week. Fix in R7.
9 · What To Do — In Order
Two of these changed after reading the experts. Last night I said "hold the ad test until Jul 15" and "the coach fix is P1." Denney's own numbers say the test can't work at $5–10/day, and the CPO rule says only one thing is the real bottleneck. So: the ad test gets cut, and the coach fix drops to #4. Honest correction.
P1 — DO THIS FIRST
R1 · Make the abandoned-checkout email actually work — and make it fast
OBSERVED — 0 people in the sequence, re-checked todayLifecycle operatorThis is the highest-earning email flow most businesses have, because the person already proved they wanted it. It should fire in about 1 hour, then +1 day, then +2 days. And it should lead with reassurance, not "you forgot something."
Two problems, not one:
- It never fires. 0 people in 13 days.
- Even the design is late. It waits for Stripe's
checkout.session.expired, which arrives 24+ hours after the fact. By then they've moved on.
- Start a real $9.95 checkout. Walk away. Wait 24h. Check the Stripe webhook log and the sequence count. That tells you if it's a wiring bug or a "Stripe never sends this event for Payment Links" problem.
- Either way, add a faster trigger: when someone clicks a Buy button and doesn't come back within an hour, tag them in Kit and start the emails.
- Rewrite email 1 to handle doubt, not forgetfulness.
Subject: quick thing about the Playbook
You clicked, then stopped. Totally fair — here's the honest version.
It's 10 minutes of reading, not a course.
It's $9.95, not a subscription.
If it doesn't help, reply "refund" and I send the money back. No form, no questions.
The 3 scripts inside are the ones people use in the meeting they're dreading tomorrow.
[ Finish checkout — $9.95 ]
— Bailey
Why this one is first: $839/month is sitting there. That's more than you made all week. MEASURED — the Jun 24 abandon telemetry
P1
R2 · Stop the ad test. It cannot give you a real answer at this budget.
INFERRED from Denney's minimums + your spendDenney"Minimum $30–50/day per ad set for the system to learn." Your test ad sets are $5 and $10/day. Also: good click rate + bad cost-per-sale = offer problem, not creative problem. Making more videos is not the fix here.
Do this:
- Turn OFF the SALES ENGINE copy of the StayPositive video (ad
120248739739550465). It's fighting its own twin. This doesn't kill the test. - Keep one test ad set (the StayPositive picture — the only one with a pulse). Turn the other off.
- Total ad spend goes from ~$35/day back to ~$25/day. You save ~$70/week while the checkout gets fixed.
- When you do test again, fund one ad set at $30/day for a week, not four at $5.
P2
Do this: check that the Playbook onboarding sequence still contains its Templates upsell (email 1 of that sequence was edited on Jul 8 — worth confirming nothing broke). Its whole job is catching the people who skipped the offer at checkout. Re-present it 3 days after they buy, while the receipt is warm.
R3 · Sell something to your email list. Anything.
MEASURED — 3 emails, ~1,000 sends, $0Lifecycle operatorHealthy businesses get 15–25% of revenue from email. You get ~0%. Your open rate is 47% — the attention is there, and you are not asking for anything with it.
Also: 0 Templates ($19.95) sold this week, vs 3 last week. Four out of five buyers took the $9.95 and stopped.Do this: check that the Playbook onboarding sequence still contains its Templates upsell (email 1 of that sequence was edited on Jul 8 — worth confirming nothing broke). Its whole job is catching the people who skipped the offer at checkout. Re-present it 3 days after they buy, while the receipt is warm.
P2
R4 · Fix Bailey's opening question — one tap, not an essay
MEASURED — 8+ people quit on that exact questionWiebe + SethiYou're asking for the hardest thing (a written two-part reflection) at the moment of lowest commitment (a click from an email). Reverse it. Give first, ask second. Make the first ask a single tap.
Bailey: "Let's plan your week in 60 seconds.
First — how did this week actually feel?"
[ Solid ] [ Mixed ] [ Rough ]
→ then:
"Got it. Most 'Mixed' weeks come down to one of these.
Which one is yours?"
[ Someone isn't delivering ]
[ A deadline slipped ]
[ Everything routes through me ]
[ Something else ]
Honest note on priority: this was #1 last night. It's #4 now. Reason: the coach makes $0. Fixing it makes a free thing nicer, which matters — but it does not put money in the bank this month. The checkout does. Target: email-sourced plans go from 1-in-49 to 6-in-49 (matching signup-page visitors). Check Jul 19.
P2
R5 · Mine the coach transcripts for real customer words — you've never done this
OBSERVED — 7 days of saved transcriptsWiebe"The most powerful copy you'll ever write was already written — by your customers. Your job is to find it and put it on the page."
As of last week, Bailey saves what people type. That's voice-of-customer data you are paying nothing for and using nowhere. Real examples from this week alone:
"Getting ahead of the game before i take a management position."
"The whole team." (what slipped this week)
Do this: pull 30 days of transcripts, highlight every phrase you would never have written yourself, and use them word-for-word as ad hooks and email subject lines. This is the cheapest new-idea source you have, and it costs $0.
P3
R6 · Write your changes down
4 changes shipped this week with no change-log entry, and the notes file cites two event IDs that don't exist. Backfill them and renumber the ad-test work toEV-2026-07-08-02/-03.
P3
R7 · Make the trial-reel lane obey the format rule
The plan passes the check; the actual feed doesn't (8 of 13 posts were video), because trial reels post outside the plan. Have the checker count them too.
P3
R8 · Add Meta's new "engage-through" column to reporting
OBSERVED — Denney playbook, March 2026 attribution change Meta changed how it counts in March 2026. "Click-through" now only counts direct link clicks, and a new engage-through number counts conversions that came from video views. We are not pulling that column. Some of the "missing" sales may just be sitting in a column we never look at. Worth adding before we conclude anything about a drop.10 · Honest Checks Before Shipping This
Did I consult the experts first this time? Yes. Denney, Wiebe, Sethi, Hormozi, the lifecycle-email operator and the product-growth CPO playbooks were all read before a single recommendation was written. Two conclusions flipped as a result (the ad test, and the coach's priority). That's what "consult first" is supposed to do — if nothing ever changes, you're not really consulting.
Did I check things were really running, not just assume? Yes, re-checked live today: ad statuses (5 active), Stripe charges (5, $106.80), the abandoned-checkout sequence (still 0 people), the guardrail rule (ENABLED), the comment robot (live, mode=auto), the tokens (good until Sep 20).
Did I flag things that were paused on purpose? Yes. The Planner campaign and the Layoff video were both switched off deliberately. Neither is a problem.
What would prove me wrong? Facebook waits up to 7 days to credit a sale to an ad, and 3 of this week's 5 sales landed on Jul 9–11. This week isn't finished counting. Re-read the numbers on Jul 19 before treating the drop as real. That's why R1 (checkout) is about a leak we've measured for a month — not about this one wobbly week.
Sample size honesty: 5 sales vs 8 sales. That's small. Every conclusion about the drop is tagged as such. The conclusions about the checkout leak and the coach stand on much bigger samples.
Did I check things were really running, not just assume? Yes, re-checked live today: ad statuses (5 active), Stripe charges (5, $106.80), the abandoned-checkout sequence (still 0 people), the guardrail rule (ENABLED), the comment robot (live, mode=auto), the tokens (good until Sep 20).
Did I flag things that were paused on purpose? Yes. The Planner campaign and the Layoff video were both switched off deliberately. Neither is a problem.
What would prove me wrong? Facebook waits up to 7 days to credit a sale to an ad, and 3 of this week's 5 sales landed on Jul 9–11. This week isn't finished counting. Re-read the numbers on Jul 19 before treating the drop as real. That's why R1 (checkout) is about a leak we've measured for a month — not about this one wobbly week.
Sample size honesty: 5 sales vs 8 sales. That's small. Every conclusion about the drop is tagged as such. The conclusions about the checkout leak and the coach stand on much bigger samples.
Report v2, rebuilt July 12, 2026 with expert playbooks loaded first. Sources: Meta Marketing API, Meta Pixel, Stripe, Kit, Microsoft Clarity, the coach's own transcript logs, Buffer, and the comment robot's live diagnostics — all pulled fresh today.