Build pipeline
Weekly Growth review · ISO week undefined · undefined undefined · undefined · Generated undefined
Are customers getting through the builder — how many start, where do they stall, and how much actually lands on the marketplace?
Fact · Cause · Action
AI-generated — stub. The builder is losing half of everyone who starts — 40 begin, 20 submit — and the drop concentrates at one step. This is a product problem, not a demand one: the customers are here and willing, but the build asks too much before it lets them finish. The table is the whole story; validate each fact below.
| Section | Fact | Cause | Proposed action |
|---|---|---|---|
| Into the builder vs out of it | Builds started rose 26 → 40 since April while submissions held flat at ~20 — the gap is abandonment, and it's widening; completion has fallen 69% → 50%, below the 65% target since late May. | Acquisition is filling the top faster than the builder converts — the loss is at the product end, not a shortfall of demand (both would fall together if it were). | Hold the 65% completion target and treat the sites-and-connections step below as the lever; re-check the started-vs-submitted gap weekly. |
| Build funnel | Of 40 starts, 8 drop at Products → Sites & connections — a 24% fall, worse than its 22% average and twice its 12% floor; the single biggest leak, and the only step still worsening. | The sites-and-connections step asks for detail customers don't have to hand (site list, connection counts) with no save-and-return — the same friction we already fixed one step down. | Copy the save-and-return fix that's pulling Sites → Requirements set below its average; apply it here, pre-fill from earlier answers, and let a partial brief through for follow-up. |
| Time to submit | Median build now takes 3.6 days, up from 2.4 — customers leave mid-build and come back, or don't. | The build spans more than one sitting; nothing pulls a stalled builder back. | Trigger a build-resume nudge for anyone idle 48 hours with a part-built requirement. |
Into the builder vs out of it
This week. The two bars are fanning apart — builds started have climbed from 26 to 40 since April while submissions have held flat around 20. We're pedalling harder at the top and losing the extra at the product end: the widening gap is abandonment, and completion has dropped to 50% against a 65% target. If this were a demand problem both bars would fall together; instead starts rise and out-the-bottom doesn't move — which points squarely at the builder, not the funnel above it. · AI stub
Catalog Error: Table with name cust_requirements_kpi does not exist!
Did you mean "main.sqlite_temp_master"?
LINE 6: from demo.cust_requirements_kpi order by week_start
^Catalog Error: Table with name cust_requirements_kpi does not exist!
Did you mean "main.sqlite_temp_master"?
LINE 6: from demo.cust_requirements_kpi order by week_start
^Build funnel — where they stall
This week. 40 opened the builder, 20 submitted. The shape is fine until Products → Sites & connections, where 8 of 34 fall away — the heaviest step and the one to fix first. · AI stub
Catalog Error: Table with name cust_build_funnel does not exist!
Did you mean "information_schema.constraint_column_usage"?
LINE 2: ... SELECT * FROM (select step, count, ln(count) as size from demo.cust_build_funnel order by step_order
^Step drop-off — this week vs its trend
This week. The builder bleeds 40 starts down to 20 submitted — a 50% total loss, and the cumulative column shows where it goes. Two steps carry it. Products → Sites & connections is the priority — 24% drop, worse than its 22% average and over floor. By contrast Sites → Requirements set has improved to 15% from a 20% average — which we'd read as the save-and-return we added there doing its job, so the effort now belongs on the step above. Everything else is stable and in line with trend. · AI stub
Catalog Error: Table with name cust_build_steps does not exist!
Did you mean "information_schema.constraint_column_usage"?
LINE 8: from demo.cust_build_steps order by step_order
^