The weekly decision rhythm: continuous capture, weekly calls, ship, measure.

Quarterly planning is a story product teams tell themselves about control. The customer doesn’t care about your quarter. The market doesn’t care about your OKR cycle. What both of them respond to is whether you can ship something this week that moves their experience meaningfully forward.

Here is how the loop actually runs: signal is captured continuously, and the decision gets made on a weekly rhythm. This post lays out that loop, the artifacts that hold it together, and where it breaks down in practice.

Why a weekly decision, not a quarterly one

Quarterly cadence assumes the world holds still long enough to plan against it. For an early-stage SaaS company, it doesn’t. The customer who churned in week three of the quarter changes the math on weeks four through thirteen. By the time the quarterly plan is reviewed, half its premises are stale.

A weekly decision cadence assumes the opposite: that the most recent two weeks of signal are more reliable than any planning artifact older than that. John Cutler at Amplitude has written extensively about this shift, and Basecamp’s Shape Up is another argument for short, opinionated cycles over long planning horizons.

The four phases

We organize the week into four phases. Each one has a single artifact attached, and the phases are time-bounded so the loop fits inside a normal week without becoming the week.

Phase 1: Listen (always-on)

Signal flows in continuously from every customer-facing surface: Slack, Intercom, sales calls, support, analytics. The operator’s job in this phase is to make sure nothing is missing, not to read every message. If the signal store doesn’t have a feed of your Tuesday sales call notes by Wednesday morning, fix that before anything else.

Artifact: a unified signal feed, browsable by source, time, customer, and topic.

Phase 2: Synthesize (Friday morning, 60 min)

Once a week, cluster the signal into named opportunities. Rank them by evidence weight, $ at risk, and implementation cost. The output is an ordered list with two or three contenders at the top.

Artifact: a ranked opportunity list with linked source evidence.

Phase 3: Decide and draft (Friday afternoon, 60–90 min)

Pick one. Write the call as a one-paragraph statement with confidence score, linked evidence, and what you’ll defer to make space. Then (and this matters) draft the thin spec and open the engineering tickets the same afternoon. Don’t wait for Monday.

Artifact: a written call, a spec, and tickets, all linked together and back to the source signal.

Phase 4: Ship and measure (next two weeks)

Build it, ship it, watch the metrics that should have moved if the call was right. Write a one-paragraph post-release note: what moved, what didn’t, what surprised you. File it back into the signal store so it becomes input for future loops.

Artifact: a post-release note. Boring discipline, compounding returns.

What holds the loop together

The thing that fails first in this cadence is not the planning. It’s the connective tissue. Three artifacts hold the loop together over time.

1. The signal store. Every piece of customer input, searchable, with author and timestamp. Without this, every weekly synthesis starts from zero.
2. The decision log.Every call you’ve made and the confidence score you gave it. Three months in, this becomes your most valuable artifact: it tells you which kinds of calls you’re good at and which you’re consistently wrong on.
3. The impact record. What moved after each release. This is the feedback loop on the operator itself.

These three artifacts are why running this loop in a spreadsheet breaks down past month two. They’re also why we built Cleo: the signal store, decision log, and impact record are exactly the connective tissue the product carries for you.

Where the loop fails

Three failure modes show up consistently.

• Synthesis takes 90 minutes instead of 60. Always because the signal store is incomplete or stale. Fix the listening phase, not the synthesis phase.
• The Friday call slips to Monday. Once or twice is fine. Three weeks in a row means the operator is overloaded; the loop needs to get cheaper, not the operator more disciplined.
• The post-release note never gets written.The single most common failure. Without it, the loop has no feedback and the operator can’t learn from week to week. Defend this phase ruthlessly.

How quarterly planning fits

Weekly loops don’t replace longer planning. They replace the weekly prioritization debate. We still run a light quarterly exercise: it sets thematic direction (which customer segment matters most this quarter, which area of the product gets the focus) and budget. But the actual call about what to ship this week comes out of the weekly loop, every time.

This is consistent with what SVPGcalls “product strategy” versus “product discovery.” Strategy is quarterly and rare; discovery and shipping happen continuously.

Starting next Monday

If you want to try this loop next week, do four things:

1. Make sure every customer-feedback channel actually feeds into one place you can read. If it doesn’t, fix this first.
2. Block 90 minutes on Friday for synthesize + decide + draft. Put it on the calendar.
3. Make the weekly call in writing, in one paragraph, with linked evidence. Share it with your team on Monday.
4. Write the post-release note four weeks later. Honestly.

That’s the loop. Cleo is what makes it sustainable past month three. Book a walkthroughif you’d like to see it running.