AuraQu
← BACK
auraqu>
Data Pipeline
Auto Dealer Availability & Discounting

End-to-end workflow: define a universe, design the questions, call at scale, extract structured fields, aggregate into signals.

1Target Universe24 endpoints

The exact set of businesses an analyst would otherwise call manually — scraped, verified, and normalized into a call-ready schema with phone number, region, and segment.

Northeast: 6Southeast: 6Midwest: 6West: 6
IDNameRegionSegment
AD-001Liberty Ford AlbanyNortheastFord
AD-002Empire Chevrolet SyracuseNortheastChevrolet
AD-003Bay State Toyota BostonNortheastToyota
AD-004Garden State Honda NewarkNortheastHonda
AD-005Colonial Nissan HartfordNortheastNissan
19 more
2Call Script4 questions · v2.4

Each question is designed to extract one structured field. The script is iterated across collection cycles based on completion rates and extraction quality.

Q1How many new vehicles do you currently have on the lot, and which models are most available?
extracts → availability_statustype: in_stock / limited / out_of_stock
"We have about 15 F-150s, down from 25 last month. Allocation has been tight."
Q2If a customer ordered a new vehicle today, what would the typical wait time be?
extracts → wait_time_daystype: numeric (days)
"RAV4 is about 3 weeks right now. Hybrids are backed up more, maybe 6 weeks."
Q3Are you currently running any incentives, rebates, or discounts on new inventory?
extracts → discounting_leveltype: none / small / meaningful
"Yeah, $2,000 off MSRP on most Silverados. Trying to move inventory before quarter-end."
Q4Compared to last month, would you say customer demand is up, flat, or down?
extracts → demand_trendtype: up / flat / down
"Trucks are still strong. Selling everything that comes in within a week."
versions: v2.1 → v2.2 → v2.3 → v2.4
3Call Engine291 calls over 8 weeks

Automated voice calls to every endpoint in the universe, with retries on failed attempts. Each call runs the script above and records the full interaction.

answered
164(56%)
no answer
46(16%)
voicemail
38(13%)
refused
21(7%)
callback scheduled
17(6%)
wrong number
5(2%)
avg duration (answered): 137sconnect rate: 40.9% → 58.1%
4Field Extractionavg 81% fields filled

Each answered call is transcribed and parsed. Every question maps to a structured field — turning a conversation into a data row.

Example: Liberty Ford Albany · 2025-12-22
Q1availability_status=limited
Q2wait_time_days=14 d
Q3discounting_level=small
Q4demand_trend=flat
Dataset output (192 records)
EntityWeekAvailabilityWait (days)DiscountingDemandConf.
Liberty Ford Albany2025-12-22limited14 dsmallflat0.79
Liberty Ford Albany2025-12-29limited14 dsmallflat0.85
Liberty Ford Albany2026-01-05limited11 dsmallflat0.93
Liberty Ford Albany2026-01-12limited12 dsmallflat0.78
188 more records
5Signal Aggregation2 indices

Individual records are aggregated into regional indices — the recurring time-series that analysts and funds consume. Not transcripts, signals.

Availability Index 1.7
57.8
Discounting Index 2.2
42
6Anomaly Detection3 alerts

Automated flagging of statistically significant week-over-week shifts — the actionable output that triggers research.

7.5%Availability Index · Southeast · 2026-01-19

Southeast auto dealer availability dropped 7.5% WoW — the steepest weekly decline across all monitored regions. 4 of 6 dealers reported "limited" or "out of stock" on popular truck trims.

8.3%Discounting Index · Midwest · 2026-02-02

Midwest discounting intensity reached 52 — the highest level in the 8-week panel. Dealers are offering meaningful incentives on both trucks and sedans to clear aged inventory.

Delivery Format

Output delivered as structured CSV / Parquet with full audit trail. Each record includes source call ID, timestamp, extraction confidence, and transcript snippet. Refreshed on weekly cadence.

CSVParquetAPIDashboard
Auto Dealer Availability & Discounting · 24 endpoints · 8 weeks