Week three is where a lot of good side projects quietly die.
Not because the idea was nonsense. Not because you “lost passion.” Usually because the honeymoon fog lifts: the repo still has open tabs, the job ate two evenings, your definition of done ballooned into a platform, and nobody outside your head has reacted yet. Tuesday feels heavier than Saturday felt in week one. You start negotiating with yourself instead of shipping.
This is a nights-and-weekends playbook for that exact dip. Diagnose what kind of dip you are in, shrink the next ship to something finishable in one evening, protect two blocks on the calendar, get one outside signal, then run a short 14-day reset—or decide the dip is actually a pause/kill call.
If your build is already too fat for a normal week, fix that first with how to scope a side project MVP you can actually ship on weekends. Motivation tactics cannot rescue a scope that needs a team.
Week three feels like the project died (it usually did not)
Early energy is cheap. Novelty, a new domain, a clean repo, maybe a friend who said “cool idea.” By week three the novelty is gone and the unpaid invoice arrives: incomplete auth, a half-built settings page, a landing copy draft that still says “TODO,” and a vague promise to yourself that you would “ship something.”
That heaviness is information, not a verdict. Treat it like a dashboard light:
Scope fog — you cannot name what “done for this week” means in one sentence.
Life load — job, family, health, travel. The project is fine; the calendar is not.
No users / no signal — you are building in a vacuum, so every session feels abstract.
Perfectionism — you will not show it until it is “ready,” so you never get the feedback that would make the next evening easier.
Most dips are a mix. Name the dominant one before you invent a productivity system.
If you cannot finish a useful slice in one evening, you do not have a motivation problem. You have a chunking problem.
Diagnose the dip in ten minutes
Open a note. Answer these without rewriting your life story.
1. Scope check. What is the single job a stranger should complete in under five minutes? If you list three jobs, that is the fog.
2. Calendar check. How many real evening hours did you get in the last 14 days? Be honest. Two heroic Saturdays do not cancel twelve empty weeknights.
3. Signal check. Has anyone outside your household used it, replied to a demo, or said they would pay? Silence is data. Polite “nice” is not pull.
4. Energy check. After the last three sessions, did you feel useful or resentful? Resentment often means you are polishing theater instead of finishing a slice.
5. Avoidance check. What task do you keep postponing? Auth, pricing, outreach, and “show a stranger” are the usual suspects. The postponed task is usually the real next ship.
Write one line: This dip is mostly ___. Then pick the matching move below. Do not do all four “systems” at once.
Shrink the next ship to one evening
Pick a finishable unit. Not “MVP.” Not “v0.1.” One evening.
Good one-evening ships look like:
A landing page with one clear promise and a working waitlist/email form
One core flow end-to-end for a fake user (even if ugly)
A Loom demo of the painful job + a Stripe payment link or manual invoice path
Five personal outreach messages with a specific ask (try / reply / call)
Delete or hide two half-built features so the happy path is obvious
Bad one-evening ships look like: refactoring, redesigning the logo, building a settings page, researching another framework, or “cleaning up before I start.”
Definition of done template (copy this):
Tonight I will ship: ___.
A stranger can: ___.
I will know it is done when: ___ (link, screenshot, or message sent).
I will explicitly not touch: ___.
If you cannot fill that in two minutes, the ship is still too big. Cut again. The goal is a closed loop before sleep—not a prettier backlog.
Protect two evenings like they are meetings
Motivation does not survive an open calendar. Cadence does.
For the next two weeks, protect two recurring blocks—same days if you can. Ninety minutes is enough. Put them on the calendar with a boring title like “Ship: ProjectName.” Treat cancellations the way you would treat canceling on a customer.
Before each block (five minutes):
Phone in another room or Do Not Disturb with only critical contacts.
Open the one-evening definition of done—nothing else.
Close Slack, X, and the “research” tabs. Research is tomorrow’s excuse.
During the block: ship the slice. If you finish early, stop or do a five-minute polish pass. Do not invent a new epic.
After the block (two minutes): write what shipped, what blocked you, and the next evening’s one-line ship. Future-you should not need archaeology.
If life load is the dominant diagnosis, two evenings may be temporary maximum capacity. That is fine. Shrink scope to match capacity. Do not shame yourself into a third block you will skip and then use as proof you “cannot finish things.”
Reintroduce outside signal (one stranger is enough)
Week-three silence makes every bug feel existential. You need one honest reaction more than you need another feature.
Pick one:
Show a 3-minute Loom to someone in the target audience and ask what they would ignore.
DM three people who already feel the pain. Ask for fifteen minutes, not “feedback.”
Post a sharp problem statement (not a launch) in a niche community and watch what people argue about.
Ask a maker peer for a blunt pass—use how to get brutal feedback from makers before you launch publicly so you ask for useful criticism, not compliments.
Write down what you heard in three bullets: what they cared about, what they skipped, what they would pay for (or not). That note becomes the next two evenings’ scope. Outside signal turns motivation into a to-do list.
If you are still pre-users and heading toward a first quiet launch later, keep the bar practical—the realistic first-week launch checklist for indie hackers pairs well with a reset sprint: ship a thin wedge, then talk to people, then decide.
14-day reset sprint checklist
Use this when the idea still feels worth a fair test and life load is manageable. Two weeks. Two evenings per week minimum. No identity speeches.
Days 1–2 — Name the wedge.
One sentence: who it is for + painful job + what “done” looks like for them.
Cut or hide everything that is not that job.
Write the one-evening definition of done for evening #1.
Days 3–4 — Ship evening #1.
Finish the slice. Deploy or send. Screenshot it.
List three people who could react this week.
Days 5–7 — Get one reaction.
Show it to at least one stranger or maker. Capture three bullets.
Adjust the next ship based on what they ignored, not what they politely praised.
Days 8–10 — Ship evening #2.
Only changes that make the core job clearer or faster.
Optional: price or waitlist ask in plain language.
Days 11–14 — Review, do not romanticize.
Did two evenings actually happen?
Did anyone use it, reply, or ask a follow-up question?
Is the next 14 days obvious, or are you back in fog?
End-of-sprint decision (pick one sentence and put a date on it):
Continue — protect the same two evenings for another 14 days with a sharper wedge.
Pause — calendar a review date; maintenance only; no secret coding.
Kill — archive, thank anyone who tried it, free the evenings.
If you keep “continuing” without protected evenings or outside signal, you are not resetting—you are lingering.
When the dip means pause or kill instead
Not every dip deserves a sprint. Sometimes week three is the project telling you the truth early.
Run a reset if scope fog or perfectionism was the main issue and you still care about the problem. Pause or kill if:
Life load will stay brutal for 30+ days and you keep breaking promises to yourself
You cannot get a single honest conversation after real outreach
Energy stays resentful even when evenings are protected and ships are small
You are keeping it alive for sunk cost, a domain name, or identity
When you hit that fork, do not invent a new motivation hack. Use a clear rubric: how to decide whether to kill, pause, or double down on a side project. The point of the week-three dip is not to suffer longer. It is to choose deliberately while the cost of changing course is still low.
What good looks like after the dip
You are not looking for permanent inspiration. You are looking for a boring loop:
Two protected evenings on the calendar
A one-evening definition of done
Something a stranger can try
A short note of what they said
A dated continue / pause / kill line every two weeks
Side projects die from fog and open calendars more often than from bad ideas. Name the dip, shrink the ship, protect the evenings, ask one honest person, then decide with a date—not a mood. Week three is not the end of the project. It is the first time the project asks if you are serious.