Month-end MBR, generated from check-ins:
"Build my October monthly business review – performance against our OKRs, what drove the results, and the decisions we need to make."
→ headline that takes a position
→ status split: 3 on track / 2 at risk / 1 off track, 46% avg progress
→ worked / stalled / decision needed
→ source list, every claim traceable to a check-in
The citations are the feature.
The OKR spreadsheet bug nobody notices:
Most templates keep one 'current progress' cell per KR. Week 6, it says 55%. Expected pace is 46%. Looks green.
Now plot the weekly history: 40, 52, 55, 55, 55, 55. It's been flat for a month.
One data point gives you a false positive. Trends tell the real story. Your tooling should show you the line, not the dot.
Zero-click dashboards.
36 widgets across 6 categories. Instead of learning them, you prompt:
"Executive dashboard. 4 headline KPI cards, plan progress + confidence, at-risk KRs paired with blocked initiatives, short agenda from the latest check-ins. Show me the layout before creating it."
AI Mode picks the widgets, sizes and layout. You approve.
Prompt we run before every leadership meeting:
"What's at risk this week? Show me the key results that need intervention, why, and what I should do about each."
It reads every KR, check-in and initiative, ranks the gaps vs target, and attaches a next step to each.
Sample output: contribution margin at -4% vs a 0% target. Suggested move: decide on a $5 free-shipping minimum.
5 ranked gaps > 13 amber dots.
Wire OKR scores into bonuses and you've built a system that optimizes for the wrong variable.
Comfortable target: ~80% odds, full bonus.
Bold target: ~40% odds, same full bonus.
Expected payout says pick the safe one. Every rational person does. The score stops measuring ambition and starts measuring negotiation skill.
We run them as two separate loops on two separate cadences.
Retro mechanic we stole from dev teams and now use everywhere:
The brainstorm is debate-free. Disagree with a sticky? Hold it.
Then everyone gets 5 votes. 50-60 suggestions collapse to the top 3-5 in minutes, and those are the only ones you discuss.
Skip the votes and you'll spend the whole hour arguing about item #4.
Did the math on OKR overhead for a 100-person org, per quarter:
building reports: 270 hrs
answering ad-hoc questions: 60
finding risks: 45
chasing updates: 45
writing check-ins: 40
tracing dependencies: 15
alignment reviews: 15
490 hrs. ~$31,850 at $65/hr. None of it moves a goal.
Our SEO agent writes its own status report after every run:
"About 1,450 commercial non-branded visits a month, up from a baseline of 950. Wrote one new article and saved it as a draft. It's in review, so no traffic impact yet. Next: watch rankings once the drafts are published."
Number, change vs baseline, what was done, what's next. That's the whole template. Humans can use it too.
Shipped: KPIs next to OKRs, with way less admin.
- date range presets (30/90d, YTD, 1-5y, all)
- actuals vs target history on one chart
- edit any cell inline, Enter to save
- CSV import, export, or full replace
- AI Mode builds the KPIs + a shared scorecard from one prompt
No more hundreds of individual forms to backfill a year of data.
We published a blog post whose real audience is LLMs.
Paste the URL into Claude/ChatGPT/Grok and type 'Run this playbook'. The bottom half is instructions for the agent: one question at a time, read the site and connected analytics, never invent a baseline, draft a 90-day plan as markdown, push it into Tability via MCP, record a first real check-in, then schedule a quiet progress check and a weekly review.
Three words in, Q1 plan out.