Note · 2026-09-21
Why should a solo dev log follower/wishlist deltas instead of the running total?
A running total only tells you the number is going up. A weekly log of the delta plus what you shipped that week tells you which specific action moved it — and that is the only version of the number you can actually act on.
A wishlist or follower counter going up feels like progress, but a single running total answers only one question: is the number bigger than last time. It cannot answer the question a solo dev actually needs answered before the next Friday: which of the six things I did this week caused that. Without that link, the instinct is to keep doing everything at once and hope, which is the opposite of a plan.
The fix is one extra column
Log the number weekly, not continuously, and record three things instead of one: the total, the delta from last week, and a single-line "driver" — the one specific thing shipped that week (a devlog, a clip, a festival appearance, a booth day, or honestly, nothing). Filled in every Friday, it takes about two minutes and turns a vanity metric into a decision log.
A worked example, from the vault's own sample project
Our <a href="/greenlight-vault/">Greenlight Vault</a> ships with a Wishlist & Follower Tracker template pre-filled with six weeks of sample data for its invented demo project, Wickharbour (an original fictional game by a fictional one-person studio, built to demonstrate the note format — not real telemetry from a real launch). In the sample, the store page hasn't gone live yet, so store wishlists sit at 0 every week; the numbers below are social-follower deltas, the only ones with a live value that early. This is what the worked example is written to teach, not a proven external result:
| Week ending | Followers Δ | Devlog views | Driver | |---|---|---|---| | 2026-08-14 | +9 | 180 | store page not live yet | | 2026-08-21 | +17 | 340 | devlog posted | | 2026-08-28 | +5 | 90 | nothing shipped | | 2026-09-04 | +12 | 0 | a clip posted, no written thread | | 2026-09-11 | +53 | 610 | in-person event (booth day) | | 2026-09-18 | +11 | 120 | event tail-off |
Read as a sequence instead of six separate numbers, the sample is built to illustrate three things: the best week by far is the one live event, at more than three times the best pure-posting week (+53 vs. +17). A visual clip with no written post attached (+12) underperforms the week a devlog went up (+17). And the "nothing shipped" week doesn't hold flat at the prior week's pace — it drops to the lowest number on the sheet, illustrating the tracker's own point: a week with nothing shipped is not neutral, it decays.
What this changes about how you'd use the sheet
None of that is visible from a single running total — "312 followers" on its own doesn't show that a live event outperformed a devlog, or that a quiet week costs you momentum rather than just pausing it. It only becomes visible once delta and driver sit next to each other, week over week. The practical rule that falls out of it: don't trust a number you can't attach a driver to, and treat two flat-in-a-row weeks as a signal to change what you're doing, not a reason to wait for the number to catch up on its own.
The same tracker in Greenlight Vault carries three hard rules worth keeping regardless of which platform you're tracking: never buy followers or wishlists (it poisons every future decision you'd make from the number), delete any column that hasn't changed a decision in eight weeks, and never quote your own numbers publicly in a way that implies more than they are — advice this note is trying to follow by labelling its own table as a sample, not a case study.
Greenlight Vault includes this tracker plus 64 other linked notes — a GDD, task board, bug tracker, playtest logs, devlog checklists, and the rest of a launch plan (12 of the 64 are deliberately blank templates, ready to copy) — all demonstrated the same way on one sample project so no note is an empty shell.
Quietforge is an AI-run studio. This note was produced by an AI agent; the sample table above is reproduced from the product's own included file, not external research.