Rank OTD Decline Drivers
When a regional delivery review shows OTD or OTC slipping, the Delivery Operations Manager has to explain the drop before leadership, carrier managers, and dispatch teams start asking for different cuts of the data. This workflow assembles the comparison, isolates the main drivers, and produces a ranked briefing with evidence and next-action guidance.
The manager still decides which interventions to approve, escalate, or defend; the workflow removes the cross-domain querying, baseline calculation, driver assembly, and briefing formatting work. The estimated impact is a 45.9% shift in role time out of execution-layer work.
See the Decision Intelligence Workflow in Action
This workflow handles a compound operations analysis task: compare current delivery performance to baseline, separate true carrier performance deterioration from lane-mix effects, and test whether exception composition changed enough to matter. The hard part is not one calculation; it is assembling a defensible explanation across shipment outcomes, carrier-lane behavior, and exception categories quickly enough for an operating review.
What the Decision Intelligence Workflow Does
This workflow handles a compound operations analysis task: compare current delivery performance to baseline, separate true carrier performance deterioration from lane-mix effects, and test whether exception composition changed enough to matter. The hard part is not one calculation; it is assembling a defensible explanation across shipment outcomes, carrier-lane behavior, and exception categories quickly enough for an operating review.
Click any step below to see the business logic, data query, and sample output for that step of the workflow.
The output is a driver attribution brief with overall OTD/OTC movement, a ranked list of the top drivers, supporting evidence bullets, and recommended next actions by driver type. It can also feed an email, dashboard panel, or scorecard-ready summary.
Data Warehouse Integration
The workflow depends on warehouse-ready operational facts that can be parameterized by region and time window, then compared to a prior baseline period. It pulls delivery outcomes, shipment volumes, carrier-lane performance, and exception categories into one repeatable analysis so the manager is not rebuilding the same joins and calculations for every review.
The cross-domain requirement matters because the explanation can be wrong if shipment outcomes are reviewed without carrier-lane shifts or exception mix changes in the same frame.
CREATE TABLE ops.ops.shipment_delivery_fact (
shipment_id STRING,
order_id STRING,
region_id STRING,
lane_id STRING,
carrier_id STRING,
promised_delivery_date DATE,
actual_delivery_date DATE,
delivered_complete BOOLEAN,
exception_flag BOOLEAN,
ship_date DATE,
delivered_date DATE
);
SELECT
period_label,
COUNT(*) AS shipments,
AVG(CASE WHEN actual_delivery_date <= promised_delivery_date THEN 1.0 ELSE 0.0 END) AS otd_rate,
AVG(CASE WHEN delivered_complete = true AND actual_delivery_date <= promised_delivery_date THEN 1.0 ELSE 0.0 END) AS otc_rate
FROM (
SELECT
string">'analysis' AS period_label,
*
FROM ops.ops.shipment_delivery_fact
WHERE region_id = string">'{region_id}'
AND promised_delivery_date BETWEEN string">'{start_date}' AND string">'{end_date}'
UNION ALL
SELECT
string">'baseline' AS period_label,
*
FROM ops.ops.shipment_delivery_fact
WHERE region_id = string">'{region_id}'
AND promised_delivery_date BETWEEN date_sub(string">'{start_date}', {baseline_window_days})
AND date_sub(string">'{start_date}', 1)
) x
GROUP BY period_label;
CREATE TABLE ops.ops.delivery_exception_fact (
exception_id STRING,
shipment_id STRING,
region_id STRING,
carrier_id STRING,
lane_id STRING,
exception_code STRING,
exception_category STRING,
exception_ts TIMESTAMP,
resolved_flag BOOLEAN
);
WITH scoped_shipments AS (
SELECT
shipment_id,
region_id,
carrier_id,
lane_id,
promised_delivery_date,
CASE
WHEN promised_delivery_date BETWEEN string">'{start_date}' AND string">'{end_date}' THEN string">'analysis'
WHEN promised_delivery_date BETWEEN date_sub(string">'{start_date}', {baseline_window_days}) AND date_sub(string">'{start_date}', 1) THEN string">'baseline'
ELSE NULL
END AS period_label
FROM ops.ops.shipment_delivery_fact
WHERE region_id = string">'{region_id}'
AND (
promised_delivery_date BETWEEN string">'{start_date}' AND string">'{end_date}'
OR promised_delivery_date BETWEEN date_sub(string">'{start_date}', {baseline_window_days}) AND date_sub(string">'{start_date}', 1)
)
), exc AS (
SELECT
s.period_label,
COALESCE(e.exception_category, string">'No Exception') AS exception_category,
COUNT(DISTINCT s.shipment_id) AS shipments
FROM scoped_shipments s
LEFT JOIN ops.ops.delivery_exception_fact e
ON e.shipment_id = s.shipment_id
AND e.region_id = s.region_id
WHERE s.period_label IS NOT NULL
GROUP BY s.period_label, COALESCE(e.exception_category, string">'No Exception')
), totals AS (
SELECT period_label, SUM(shipments) AS total_shipments
FROM exc
GROUP BY period_label
)
SELECT
e.period_label,
e.exception_category,
e.shipments,
(e.shipments / t.total_shipments) AS exception_rate
FROM exc e
JOIN totals t
ON e.period_label = t.period_label
ORDER BY e.period_label, exception_rate DESC;
Where the Work Sits in the Labor Stack
Not all cognitive labor is equally automatable. The KWF analysis breaks the workflow into three layers — execution, judgment, and strategic — and maps each step to the layer it belongs to. Execution-layer work is automatable. Judgment and strategic work stays with the manager.
The Business Case for Automation
Primary Valuation Metric: % of weekly/monthly delivery performance reviews where the briefing is used as the single source of truth and stakeholders accept the top-3 drivers without additional ad-hoc data pulls.
Next Step
Run your own value estimate for delivery operations — or talk to us about your OTD driver attribution workflow.
The Cognitive Labor Value Calculator models team size, fully loaded role cost, and how much execution-layer work this workflow can absorb. It takes under two minutes to generate a planning estimate.