Getting data out of a dealer management system is the unglamorous half of every dealership analytics project, and the half that decides whether the other half is trusted. We build the extracts, normalize the account structures across rooftops, reconcile to the financial statement, and keep it running.
Every load ties to the statement before it is published
One chart of accounts across rooftops and brands
Nightly loads that pick up restated prior periods, not just new rows
A feed that fails raises an alert instead of showing yesterday as today
Dealership analytics projects rarely fail at the visualization layer. They fail because the used-vehicle gross on the dashboard does not match the used-vehicle gross on the statement, nobody can say why, and within a month the general manager is back in the spreadsheet.
So we treat extraction, normalization, and reconciliation as the substance of the engagement rather than as plumbing to get past. A dashboard that reconciles is used. One that does not is a demo.
CDK Global, Reynolds and Reynolds, Dealertrack, Tekion, Auto/Mate and others. Deal detail, repair orders, parts transactions, technician time, general ledger, and the financial statement itself.
VinSolutions, DealerSocket, Elead and comparable systems. Lead source and cost, appointment and show rates, sold and unsold traffic, and the salesperson activity that sits underneath a closing ratio.
vAuto, HomeNet, Dealer Specialties and the listing syndication feeds. Stocking, pricing history, cost-to-market, photo and description completeness, and days on lot from the date that actually matters.
Valuation and market pricing sources for cost-to-market and price-to-market, plus the auction and wholesale data behind an acquisition decision.
Credit application decisions, funding status, contracts in transit, menu presentation records, product administration, and reinsurance reporting.
Warranty claims and status, incentive and stair-step programs, allocation, and the manufacturer reporting that has to be produced whether or not it is convenient.
Appointment volume and capacity, multi-point inspection results, and declined recommendations — the data behind the fixed-ops upside nobody can quantify without it.
Pay plans and commission detail, accounting exports for groups running a separate general ledger, and telematics or fleet maintenance data where a commercial book is involved.
DMS vendors differ enormously in how open they are, and the right method depends on the platform, the contract you have with it, and what the data is for. We settle this during assessment rather than assuming it:
Access stays yours. Integrations run under credentials your organization owns and can revoke, scoped to the data the reporting needs. We do not ask for broader access than the work requires, and every extract is documented so you can see exactly what leaves each system.
A two-rooftop group with two brands has two charts of accounts, and they will not agree. One store books internal labor somewhere the other does not. One reports a lot-porter cost in variable expense, the other in fixed. Neither is wrong; they simply were not set up together.
Comparing those stores without reconciling that first produces a number that is technically correct and operationally worthless. So the mapping layer is explicit: each store's accounts map to one common set of line-item definitions, each definition is written down, and the mapping is data you can inspect and change rather than logic buried in a query.
| Stage | What happens |
|---|---|
| Extract | Scheduled pull from each source into a landing area, with the raw payload retained so any published figure can be traced back to what the source actually said. |
| Map | Store accounts mapped to common line definitions. Unmapped accounts are flagged rather than silently dropped — a new account appearing in a store is a notification, not a slow leak. |
| Reconcile | Each store's mapped totals are compared to its financial statement. A variance outside tolerance stops publication and raises an alert. |
| Consolidate | Stores roll up to group with intercompany handled explicitly, so a group total is not a sum of things that were counted twice. |
| Publish | Reconciled data is made available to dashboards, reports, and exports. One dataset, so the dashboard and the board package cannot disagree. |
| Monitor | Freshness, row counts, and reconciliation variance are checked every run. A stale feed shows as stale rather than as a quietly flat trend. |
Dealership data is not append-only. Deals get unwound, repair orders reopen, month-end adjustments restate a period that already closed, and a warranty claim resubmits under a different amount. An integration that only loads new rows will drift away from the statement and nobody will notice until quarter end.
So loads are incremental but re-checking: prior periods within a defined window are re-read every night and compared, changes are applied, and a restatement large enough to matter is reported rather than absorbed. It costs more to build and it is the difference between a warehouse that stays true and one that has to be rebuilt every year.
Integration engagements typically run six to twelve weeks for a single rooftop and longer for multi-brand groups, where the mapping work rather than the extraction work sets the pace.
It is a real cost and we surface it during assessment rather than after. Where a vendor's integration program is expensive, scheduled exports or a reporting replica often satisfy the same requirement for considerably less, and we will tell you when that is the case. What we will not do is route around a restriction your agreement prohibits.
They should not. Loads run overnight during the platform's quiet window, read incrementally, and use a reporting replica where one is available. If a source can only be queried in a way that would affect production performance, we say so before building rather than after someone complains about a slow lookup on a Saturday.
Yes. The warehouse is the deliverable and it is open — Power BI, Tableau, Looker, or a spreadsheet connection all work against it. Several clients run our dashboards for daily operations and keep an existing BI tool for finance's own analysis, off the same reconciled data.
The monitoring catches it, usually on the first run: a column disappears, row counts move, or reconciliation goes out of tolerance, and the feed alerts instead of publishing. Adapting the extract is routine maintenance under a support arrangement. This is the main reason we recommend one.
It is common and it is handled. Each platform gets its own extraction approach, and everything meets at the mapping layer where accounts are translated to common definitions. Mixed-platform groups take longer to map, but the consolidated result is no less reliable.
You do. The warehouse can live in your cloud account, credentials are yours, and the pipeline code is delivered to you. If you decide to take the work in house or move to another firm, nothing about that is obstructed.
Tell us which systems you run and how your reporting is produced today. We will come back with an integration approach and what it takes to reconcile.
Schedule a Consultation
K Auto Tech