Yuupe Sinapse · thesis and framework
Every customer teaches you something.Is your company learning?
The chatbot answers, the email goes out, the survey asks. And at the end of the month nobody can say what the company learned from its customers, or what changed because of it. The data exists, scattered across six places. The learning does not.
In neuroscience, a synapse is the point where a signal becomes action. Yuupe Sinapse is the framework that does this in a company: it links what the customer says to what the company changes, and measures the effect. We call the result Applied Learning.
The thesis: companies that turn customer signals into measured actions improve the customer experience faster than companies that only collect or only analyse. The bottleneck is not having data, it is turning it into action.
The model
A nervous system: Signals, Brain, Reflexes, Pulse.
01 · ListenSignalsEvery contact in one place, in five categories: voice, behaviour, transaction, operations, team. No forms, no spreadsheets.
02 · UnderstandBrainSignals become themes and themes become learnings, every week, with source and confidence.
03 · ActReflexesA learning that became an action in an agent: a new FAQ, a ready reply, an adjusted journey. With date, origin and undo.
04 · ProvePulseThe 0 to 100 index of how the customer feels, and the before and after of each reflex.
The five laws
What separates learning from reporting.
- 1Every contact is a signal.Nothing is lost. Capture is automatic, no forms and no spreadsheets.
- 2No signal goes unread.AI reads everything. People read what matters.
- 3No learning without application.Knowledge that does not become a reflex is a report.
- 4No application without measurement.A reflex without proven effect is an automated guess.
- 5No change without human veto.The brain proposes, the company decides, and everything can be undone.
What we will prove
Five hypotheses, registered before looking at the data.
- H1The LQ predicts the change in Pulse over the following 4 to 12 weeks.
- H2Application explains more of the Pulse than Coverage: what changes the customer is action, not data volume.
- H3Data in a single source increases Cadence and Application. Fragmentation drags the LQ down.
- H4Themes that received a reflex improve more than themes in the same company that did not.
- H5Reflexes with human veto have equal or greater effect, and fewer reversals, than automation without veto.
DesignLongitudinal study with Yuupe companies, unit of analysis company-week.
Strongest evidenceDifference-in-differences between treated and untreated themes within the same company: the company is its own control.
Validating the LQThe assessment LQ is compared with the LQ measured in the product. First case: Nicolândia. Then a paper with a partner university.
The measure
LQ, Learning Quotient: from 0 to 100, in four parts.
A measure any company can calculate and compare, as NPS did for recommendation. Each part is worth 25 points.
CoverageOf customer contacts, how many became a read signal.signals read ÷ contacts
CadenceNew learnings per week, relative to signal volume.learnings ÷ weeks
ApplicationOf learnings, how many became a reflex within seven days.reflexes ÷ learnings
ProofOf reflexes, how many have a positive measured effect on the original signal.reflexes with effect ÷ reflexes
0 to 15DeafDoes not record.
16 to 30ListensRecords and does not read.
31 to 50UnderstandsReads and produces learnings.
51 to 70ActsLearnings become reflexes.
71 to 85ProvesMeasures the effect of each reflex.
86 to 100Learns on its ownThe loop runs every week without anyone asking.
What the framework stands on
We did not invent the theory. We invented how it turns into action.
Absorptive capacityCohen and Levinthal (1990), Zahra and George (2002). Acquiring, assimilating, transforming and applying external knowledge. The LQ measures the potential part (Coverage, Cadence) and the realised part (Application, Proof).
Market orientationKohli and Jaworski (1990): generating customer intelligence, disseminating it and responding. Link to performance confirmed in the meta-analysis by Kirca, Jayachandran and Bearden (2005).
What is new in Yuupe SinapseThese theories have always been measured by questionnaire. Here the measure comes from recorded behaviour, every week, without the company rating itself.
Sensemaking, WeickOrganisations make sense out of cues. The problem is not a lack of data, it is a lack of plot. The Brain is a sensemaking tool, not a decision tool.
SECI, NonakaTacit becomes explicit in four steps: conversations, AI readings, distillation, and FAQs, scripts and rules. Empirically validated in 2019 with 838 people.
Visualisation that worksGraphs above 50 nodes are unreadable (Yoghourdjian, InfoVis 2020). That is why the Brain aggregates by theme and zooms in, never showing the whole graph.
What is your company's LQ?
A few questions, three minutes, a score from 0 to 100 and the level you are at.
Calculate my LQ