THE WORK AS IT DEVELOPS

What’s new.

New ideas to examine. Better ways to use them.

Explore the latest published additions to Polaris Shift. Each note explains what changed and where to try it. Dates mark site releases, not the dates of the events or research being examined.

Choose a next step.

· TWO-WEEK REVIEW

Easier to explore. Still learning what helps.

Over this build cycle, I’ve made the original white paper available to download, refreshed the evidence reviews, and improved the path between a review and its Lab rehearsal. Two fictional films offer another way into the question of better execution and customer value.

Release checks and public walkthroughs show that these paths work. I still need verified usage reporting and completed reader walkthroughs to understand how people use them. Traffic, downloads, returning use, and reader impact remain unmeasured in this review.

The next three priorities are to observe a reader taking a question through the Lab, establish a dated baseline from available reporting, and revisit a provisional signal for evidence that could change the interpretation.

If you try it, could you explain why you chose to Stay, Adjust, or Shift, and name one piece of evidence that would change your mind? Bring a question or correction →

Published additions.

  1. Lab improvement

    Take the challenge into the meeting.

    The North Star Brief’s meeting summary now keeps the evidence that would change your mind in its preview, copied notes, and download. To try it, write your challenge evidence in the brief, then open “Prepare a meeting summary”. Blank answers remain explicit. This fixes an omitted field; it does not establish better decisions.

    Build a brief for your next conversation →
  2. Evidence review

    When measured gains support staying the course.

    The speed-versus-value review now includes a retail field-experiment counterpoint, with separate sales results, postpurchase guardrails, and limits on what one platform can tell us. The original self-report survey remains distinct.

    Examine the measured counterpoint →
  3. Signal review

    Delegating payment does not settle product fit.

    The discovery review now includes OpenAI’s limited Wallet rollout, distinguishes agent payments from discovery referrals, and follows the question through purchase completion and customer fit. Announced capability remains separate from outcome evidence.

    Review discovery, payment, and product fit →
  4. Source clarity

    Kept trying. Recovered. Finished?

    The agentic-work review now separates continued tool use, later recovery with the same tool, and whole-task completion using Microsoft’s November paper. It asks when a deliberate human handoff is the better outcome.

    Examine persistence, recovery, and completion →
  5. Source clarity

    After the cart transfers, what happens?

    The commerce review now explains where Google’s cart-transfer reporting ends and what merchant evidence could connect the handoff to a suitable completed purchase.

    Examine the handoff and outcome evidence →
  6. Film

    Every Metric Was Green. So Why Were Customers Leaving?

    A fictional reporting-software story explores the gap between operational improvement and customer value, with three questions for a strategy review.

    Watch Every Metric Was Green →
  7. Film

    The Execution Trap

    In this fictional story, a paper map company keeps making better maps while live navigation changes what one traveler needs. The film asks what evidence would make you Stay, Adjust, or Shift.

    Watch The Execution Trap →
  8. Source clarity

    Universal Cart moves from preview to rollout.

    The commerce review now includes Google’s October 15 report of a U.S. rollout in Search and Gemini. How widely shoppers can access it, whether they use it again, and what it does for their purchases remain open questions.

    Read the rollout evidence and its limits →
  9. Signal review

    Did the agent finish the work?

    The agentic-work review distinguishes recovering from a failed tool call from completing the job. It asks what a team should measure across the whole workflow, including human review and rework.

    Read the agentic-work evidence review →
  10. Signal review

    Discovery and checkout can move differently.

    A review of OpenAI’s March announcement separates product discovery from merchant checkout and asks what would show that a shopper found a better fit. Evidence limits and reasons to reconsider are explicit.

    Review discovery and checkout →
  11. Signal review

    An announcement is a starting point.

    The agentic-commerce entry now separates Google’s May announcements from evidence of repeat use and customer outcomes, with a dated review and clearer reasons to reconsider the interpretation.

    Review the agentic-commerce evidence →
  12. Lab improvement

    Make revisions in the right place.

    When revised starter answers differ from an existing brief, the Lab explains which text will be downloaded and links directly to the relevant editing fields. Existing writing stays protected.

    Try the starter and revision flow →
  13. Source clarity

    What does the survey mean by value?

    The speed-versus-value review now explains whose judgment the survey captures, what its multiplier range represents, and why customer outcomes need separate evidence.

    Read the clarified evidence review →
  14. Strategy argument

    Staying can be a decision, too.

    An argument with a fictional example asks what we are preserving, what would make Stay wrong, and when to review the choice. Carry the reasoning into the existing North Star Brief.

    Examine the case for Stay →
  15. Evidence review

    When the work changes, does the measure still hold?

    A companion review examines METR’s February methodology update and asks whether a changing task mix makes familiar comparisons less useful.

    Examine the companion review →
  16. Lab improvement

    Keep your starting point.

    Answer three starter questions, then download your current brief immediately or continue with evidence. Unanswered fields remain explicit, and your existing writing is preserved.

    Start a working brief →
  17. Practical guide

    Five questions before changing direction.

    Clarify purpose, North Stars, and metrics; compare the three responses; and decide what evidence, ownership, and follow-through your next conversation needs.

    Explore the five questions →
  18. Decision History 01

    Adobe: when progress makes the dashboard look worse.

    Examine the 2012–2013 subscription transition through dated evidence, discussion alternatives, and a clearly separated Polaris interpretation.

    Read the decision history →
  19. Fictional worked example

    The dashboard improved. Did the customer’s life?

    Follow a missing-refund scenario through Signal, Shift, Star, and System, then consider a bounded test of durable resolution. This teaching example claims no business results.

    Read the worked example →
  20. Evidence review

    Faster work. More value?

    A dated review of METR’s May 2026 research separates speed from value, explains the evidence limits, and offers a question to test with your team.

    Examine the evidence review →
  21. Reading and sharing

    Keep the original white paper.

    Download the original branded When to Move Your North Star PDF directly from the readable edition, without signing in. Browser printing remains available.

    Read or download the paper →

HELP SHARPEN THE WORK

What would make this more useful?

A question you could not answer, evidence that challenges an interpretation, or a specific obstacle in using a tool can guide the next improvement.

Bring a question or correction →