Saying no
Every yes costs something
Players ask for pets, battles, a level editor, a movie. Every idea is good to someone, and every yes adds work, bugs and months.
Construx warns that when a product is redefined after the plan is made, the cone of uncertainty "will get wider": the estimate gets worse again.
Non-goals are decisions
Construx says defining the product, "including committing to what you will not do", narrows the cone. Malte Ubl's design docs have a non-goals section for the same reason.
A written no also ends the same argument at every meeting: the team points to the doc.
Quick check
Why does writing down non-goals make estimates better?
- Each decision removes an unknown, which narrows the cone of uncertainty
- Shorter documents are always more accurate
- Publishers only read the non-goals
Show the answer
Each decision removes an unknown, which narrows the cone of uncertainty
An estimate is uncertain because the project is. Every firm "we will not" removes a way the project could grow.
Not now is a kind no
"Not at launch" keeps an idea without paying for it today. Kai keeps a later list.
After launch, real numbers decide what comes off it: how many players there are, what they play, and what they ask for most.
How to say no
- Thank themThey cared enough to write.
- Say what you are focused onName the goal the idea does not fit.
- Say what happens to the ideaOn the later list, or not planned, honestly.
Player message
PLEASE add player-versus-player battles, my friends would love it!
Shard Garden's design doc lists battles as a non-goal at launch. What should Kai reply?
- Thanks! At launch we are focused on four friends defending one garden together, so battles are not planned. They are on our list for after launch.
- No.
- Sure, coming next week!
Show the answer
Thanks! At launch we are focused on four friends defending one garden together, so battles are not planned. They are on our list for after launch.
It thanks the player, names the focus and says honestly what happens to the idea. "No" alone feels rude, and a false promise is worse.
Quick check
Saying no to a feature means it can never be added.
- True
- False
Show the answer
False
A no is for now, for this plan. After launch, real player numbers can bring an idea back from the later list.
True story: Among Us 2, cancelled
In 2020, Among Us suddenly had far more players than its small studio, Innersloth, had ever planned for. In August they announced Among Us 2, because the first game "really was not created to be this big".
On 23 September they changed course: "We have decided to cancel Among Us 2 and instead put all our focus into improving Among Us 1." They called it "probably the more difficult choice", since it meant "going deep into the core code of the game". First on the list: servers, colourblind support and accounts.
Source: Innersloth: The Future of Among Us (23 September 2020), innersloth.itch.io
Quick check
What did Innersloth say no to?
- A sequel, so they could make the game players already loved stronger
- Fixing the servers
- Colourblind support
Show the answer
A sequel, so they could make the game players already loved stronger
They cancelled Among Us 2 and moved its plans into Among Us 1, with servers, colourblind support and accounts first.
Shard Garden has a plan
A bet with a kill rule. A design doc, RFCs and decision records. A range estimate and a promise rule. A risk register. A prototype of the hard part, a vertical slice, milestones with written definitions, testable requirements, and a list of noes.
Now the studio can build.
Bolt says: Boss time: one quiz across the whole Plan stage.
Fact-checked 12 October 2026
- Construx: The Cone of Uncertainty construx.com
- Malte Ubl: Design Docs at Google industrialempathy.com
- Innersloth: The Future of Among Us (23 September 2020) innersloth.itch.io