Arjun Patel Advisory memo Decision Call · USD 220
Cut four of the eleven, and the one you will not want to cut
60 minute call, 2 October. Reviewed and signed off by Arjun Patel on 4 October.
Eleven initiatives, nine engineers, two quarters. You asked which four to cut. Three of them are easy and everyone in your team would name the same three. The fourth is the integrations platform, which is your CEO's initiative and the one thing on the list with no evidence behind it at all.
On the evidence you supplied, seven initiatives have a stated problem, a user signal and a measure. Four do not. Three of those four are uncontroversial. The fourth is the integrations platform: 14 engineer-weeks committed, requested by two customers, one of whom has since churned, and no one has asked the other.
The question
What was asked, and what was reviewed
As you put it: "We have eleven things and capacity for about seven. Which four go?"
| Reviewed | Source |
|---|---|
| Roadmap, eleven initiatives with estimates | Supplied |
| Supporting evidence for each, where it existed | Supplied. Four had none |
| Product usage data | Supplied |
| Support ticket themes, 12 months | Supplied |
| Sales loss reasons, 24 deals | Supplied |
| Customer interview notes, 9 interviews | Supplied |
| The integrations platform's origin | Established on the call, see clause 3 |
Scored
The eleven, on one basis
Four questions, asked identically of each: is there a stated problem, is there a signal that real users have it, is there a measure of success, and is the estimate grounded in anything.
| Initiative | Problem | Signal | Measure | Verdict |
|---|---|---|---|---|
| Bulk import | Yes | 41 tickets | Yes | Keep |
| Permissions rework | Yes | 9 lost deals | Yes | Keep |
| Reporting v2 | Yes | Usage and interviews | Yes | Keep |
| Mobile web fixes | Yes | 23 percent of sessions | Yes | Keep |
| Onboarding rebuild | Yes | Activation data | Yes | Keep |
| Audit log | Yes | 6 enterprise asks | Yes | Keep |
| Search improvements | Yes | Usage | Weak | Keep, define the measure |
| Dark mode | No | 3 tweets | No | Cut |
| Workspace themes | No | None | No | Cut |
| AI summary panel | No | None | No | Cut |
| Integrations platform | Stated | 2 customers, 1 churned | No | Cut |
Seven of eleven pass cleanly, which is a healthier roadmap than most. The four that fail, fail on the same question: nobody can point to evidence that a user has the problem.
The hard one
The integrations platform
| Claim | What the evidence shows |
|---|---|
| Customers are asking for integrations | Two asked, 14 months ago. One churned in June |
| It unblocks enterprise deals | Not in the loss reasons for any of the 24 deals reviewed |
| It is a platform play | True, and platform plays need a partner pipeline. There is none |
| It is 14 engineer-weeks | Estimated before any design. The team's own view is that it is closer to 22 |
| The CEO is committed to it | True. This is the actual reason it is on the list |
Next
What to do
- Apply the four questions in clause 2 to everything, permanently. It is the cheapest prioritisation method that works and you can run it yourself.
- Cut dark mode, workspace themes and the AI summary panel this week. Nobody will object.
- Take the whole scored table to your CEO. Do not lead with the integrations platform.
- Ask the one remaining customer who requested integrations whether they still want it. One email, and it settles the evidence question either way.
- Whatever is decided, write it down with the date and the reason, so that in six months nobody relitigates it from memory.