Work · Case study
A decision tree for short-form video onboarding
Turning ambiguous sales and onboarding conversations into a reproducible workflow. A choose-your-own-adventure assessment that walked Customer Success through whether a prospective customer's content was usable on the platform, and routed every outcome to the right internal team.
Overview
Every prospective customer arriving at Firework brought a different content situation. Some had hundreds of clips already in vertical short-form. Some had long-form videos that needed to be re-edited. Some had no content at all and wanted help producing it. Each scenario routed to a different internal team and a different commercial conversation, but the routing was happening case by case, in DMs and meetings, and it slowed everyone down.
I built a decision tree that turned those conversations into a reproducible workflow. Customer Success could run a prospect through the questions, hit an outcome, and know which team to loop in next.
The problem
"Can we use your platform" looked like one question and was actually a stack of questions:
- Do you have content already?
- If yes, is it in the right format (short, vertical)?
- If not, do you want our Creative Services team to produce or re-edit it?
- If you have content in spec, how do you want to get it onto the platform (manual upload, mRSS feed, batch import)?
- Once it's on the platform, who manages playlists and tagging?
- Do you need the content syndicated, and if so, with what restrictions?
Every branch had different staffing implications. A SaaS-only customer was lightweight. A customer that needed Creative Services involvement was heavier. A customer that wanted Content Operations to manage their content on an ongoing basis was a different commercial conversation entirely. Customer Success didn't always know which branch they were on until weeks in, and by then they'd accidentally promised the wrong thing.
Context
Multiple teams felt the consequences:
- Customer Success ran the relationship and the onboarding conversations.
- Sales needed to scope correctly so the deal made sense commercially.
- Creative Services got pulled in when content needed to be produced or re-edited.
- Content Operations (my team) handled ingestion, tagging, moderation, and playlist management for customers who needed it.
- Leadership needed predictability on which deals turned into which kind of workload.
My role
I built the assessment, drove its adoption, and iterated on it as we learned where it broke:
- Mapped every onboarding scenario I could think of and the actual operational consequences of each
- Translated the scenarios into a sequence of questions a Customer Success person could ask in a normal conversation
- Defined the routing outcomes: SaaS-only, Creative Services, Content Operations, or Blocker
- Built the visual decision tree the team could share with prospects and use internally
- Trained Customer Success on the assessment and gathered feedback as they used it
- Refined the tree based on the cases it didn't handle well
What the assessment covered
Do you have content?
First branch. No content with no interest in producing it was a blocker. Content with mismatched format routed to Creative Services. Content in spec moved forward.
Is it in spec?
Short, vertical format. If not, did the customer want Creative Services to handle re-editing, or did they want to handle it themselves?
How does it get uploaded?
Self-serve manual upload. mRSS feed from their CMS. Batch upload that needs Content Operations to handle. Each path had its own onboarding sequence.
Who manages playlists?
If the customer wanted to manage their own playlists, the SaaS deal made sense. If they wanted Content Operations to curate on their behalf, that was a different commercial conversation.
Syndication?
Did the content need to be syndicated to external surfaces, and if so, with what regional or content-type restrictions?
Outcome
Every path ended in one of four outcomes: SaaS-only, Loop in Creative Services, Loop in Content Operations, or Blocker.
Product decisions and trade-offs
- Outcomes over advice. The tree didn't try to give Customer Success the right thing to say. It told them which internal team to loop in. Smaller surface area, faster adoption.
- Blocker as an outcome. Most decision trees pretend every prospect can be onboarded. Naming "Blocker" as a real outcome let us politely close conversations that were going to waste everyone's time.
- Visual, not procedural. The tree was a single flowchart Customer Success could screenshot and share with a prospect. That made it usable in customer-facing conversations, not just internal ones.
- Plain-language questions. The questions were phrased like a salesperson would actually ask them, not like an internal taxonomy. That made the assessment feel like a conversation, not a screening.
Outcomes
Customer Success had a tool they could run in real time during prospect conversations. The right internal team got involved earlier. Commercial scoping got more accurate because the routing was clearer. Blockers got identified before they became weeks of effort.
More broadly: the operation moved from case-by-case judgment to a documented pattern. The next Customer Success hire could be onboarded into the workflow instead of being trained on the institutional memory of every weird scenario.
What I learned
Process work is product work. Internal teams use these decision trees the same way customers use product UI. Designing them deserves the same care.
Naming the blocker is a feature. The willingness to say "this doesn't fit" was the most valuable part of the tree. Most processes don't have that outcome available.
The job is to remove judgment calls, not arguments. A good decision tree turns repeated judgment calls into structure. It doesn't try to remove every disagreement, it just makes sure the disagreements happen earlier and at a higher level.