JJ Signal
What we have actually measured
Research, points of view, field notes and benchmarks drawn from our own delivery work. Every number states its sample and its method, because a benchmark that hides those is an advertisement.
Everything we have published
Newest first.
- BenchmarkWhat demand forecasting is actually worth to a 50-person retailerAcross our retail engagements the recoverable number was working capital, not revenue — and it showed up in the first quarter.7 min read
- ResearchWhere a clinic's hours actually goWe timed reception work across four clinics. The consulting room was not the constraint, and neither was the software.8 min read
- ResearchThe billable hour leak is a retrieval problemSenior time in a professional practice goes on reconstructing what the firm already knows.9 min read
- Field noteThe listing bottleneck moves every fortnightEveryone at the agency named a different stage. The timing data said they were all right, in turn.5 min read
- ResearchWhat an exception actually costs to clearThirty-eight minutes per exception, of which twenty-nine were gathering context rather than deciding anything.7 min read
- Point of viewGuardrails belong in the schema, not in the promptAn advice system that cannot select a product is safer than one instructed not to.6 min read
- BenchmarkTenant isolation is where most RAG implementations leakFiltering by tenant after retrieval is not isolation. It is a filter, and filters get forgotten.10 min read
- BenchmarkWhat an agent actually costs to runThe model bill is rarely the expensive part, and the part that is expensive is rarely budgeted.8 min read
Read something you disagree with?
Good — that is what a thesis is for. Tell us where your business breaks the pattern and we will look at your numbers rather than ours.
Scope it with us