Four of these six areas come from Brendan's three years as the sole analyst on a multi-billion dollar DTC e-commerce account. All six are done with AI tools now and checked before they count. Each one names what you receive and the offering it sits in.
Response and propensity models that turn a rule set by habit into a score per record. The score is built on part of your history, then judged once on the part it never saw, against the rule you use today.
A score beside every record, the line that turns it into a decision, and a memo that shows whether it beats your current rule.
Scored Decisions ›Customer lifetime value, customer segmentation, and response estimates for mail and email. This is the work Brendan did for three years.
Segments your team can name and act on, a value estimate per customer, and the list for the next campaign.
Scored Decisions ›A readout designed backwards from one recurring call, with the few numbers it turns on and the line that triggers action. Brendan presented results like these to senior stakeholders for three years. The build uses Python and SQL, with the page in Tableau or Python.
A decision brief, the numbers defined once and checked against your finance totals, and one page your team reads every week.
Reporting That Ends in a Decision ›A/B tests, test cells and control groups, sized so the result is readable. A selection result shows who answered. A test cell shows what the change caused.
A holdout cell drawn at your next run, the rule for reading it written down in advance, and the read when the results come back.
Inside every Scored Decision, and the Decision Run ›Manual work timed with a stopwatch, priced at the loaded rate of the person doing it, and ranked by what it would return. The top workflows are then built on your accounts. Where a workflow hides a judgment call, it gets a score instead of a fixed rule.
The timed ranking, the chosen workflows running on your accounts beside the manual process until they match, and the same work timed again after go-live.
Operations Automation ›Kinetric builds with Claude Code and the Claude API, and we direct every build. Each output gets checked before it counts: scripts first, independent review second, a human read last. A number with no code behind it doesn't ship.
Faster delivery without the guesswork. Every figure in your memo traces to the code that produced it, and the checks re-run on request.
Free checks you can run yourself ›We'll tell you what your own history can say about it, and which offering fits. If none does, we'll say that too.
Start a conversation