Using Your Sold History to Decide What to Buy Next

Most used-car desks buy the car in front of them. The auction catalog, the trade on the write-up, the wholesale list a friend forwarded. Sold history sits in a DMS report that nobody opens except to argue about last month’s gross. That is backwards. The only evidence you have about what your market will take, at what pace, and at what recon cost is the file of cars you already sold. Ignoring it is how lots repurchase the same aged mistake under a new stock number.

Using your sold history to decide what to buy next is not a data-science project. It is a weekly scoring habit. You tag what left the lot, you notice which combinations of year, segment, price band, and condition actually funded, and you treat that list as the default buy card. Anything outside it needs a reason, not a feeling.

This article is for used-car managers and general managers who can pull a sold report and who already fight with buyers about "great deals" that do not match the store. It covers how to structure the sold file, how to score units, how to turn scores into a buy list, and the traps that make sold history lie.

Pull Sold History You Can Actually Score

If the export is a dump of every accounting field, it will not get read. Build a working file with only the columns that change a buy decision.

Columns that earn their place

• Stock number and VIN
• Year, make, model, trim, body style
• Segment (your mix buckets, not marketing names)
• Purchase source (auction, trade, street, other dealer)
• Purchase date and sold date (days-to-turn)
• Purchase cost, recon cost, pack, front-end gross
• Mileage band
• Title status and any rebuilt / salvage flag
• Retail price at sale
• Whether the unit was aged past your internal line before it sold

Wholesale-only units belong in a separate tab. Mixing them with retail solds will tell you to buy cars you only know how to pack out.

Time windows that do not fool you

Use three windows, not one:

1. Last 30 days: what is moving now.
2. Last 90 days: enough volume to see a pattern.
3. Same 90 days last year: only as a seasonal check, not as a quota.

If 30-day and 90-day disagree, the 30-day window wins for next week’s buys. The 90-day window keeps you from overreacting to a single weekend. Last year keeps you from buying convertibles in October because August was strong.

Exclude noise before you look for patterns

Drop or flag:

• Deals that unwound or were rewritten as different units.
• Employee and dealer trades that were never true retail demand.
• Units sold at a loss to a wholesaler after a failed retail attempt, unless you are studying what not to buy.
• One-off specialty cars that will never be a program.

Consider a lot with 60 units that sold 36 retail cars in 30 days. After you strip unwinders and wholesale, you might have 30 clean retail solds. That is enough to see which segments left and which sat. It is not enough to declare a law of the market. Treat the list as a bias, not a guarantee.

Score Each Sold Unit Before You Shop

A raw list of "we sold a lot of trucks" is how you overbuy trucks that are the wrong trucks. Score at the unit level, then roll up.

A simple scorecard

Give each retail sold a 1–5 on four questions, using your own thresholds, not industry folklore:

Add the four scores. High-scoring units are repurchase candidates in that shape: similar year range, mileage band, equipment, and cost-to-market. Low-scoring units are a warning even if the body style is popular.

Roll scores up to a buy pattern

After scoring, group by segment and a second dimension that actually changes the buy:

• Price band or payment band (especially on BHPH lots)
• Drivetrain (FWD vs AWD, 4x4 vs 2WD) where it matters in your climate
• Mileage band
• Source (trade-in vs auction) if one source consistently scores better

The output is a short list: "compact AWD crossovers in this price and mileage band, preferably trades, recon under this cap." That sentence is more useful than a make/model trophy list. Makes go in and out of the lane. The pattern is what your buyers keep taking.

Watch the failures as hard as the wins

Sold history’s most valuable rows are often the slow, skinny deals. They tell you what to stop buying. Build a second list of units that:

• Crossed your aged line before they sold.
• Required a price cut that erased the recon.
• Came back to the shop after delivery.
• Only sold after you took them to the auction.

Those units should appear on the buyer’s card as "do not replace." Pair this with building a used inventory mix plan by vehicle segment so a high-scoring sedan does not get ignored because the buyer prefers trucks, and a low-scoring truck does not get rebought because the lot "needs iron."

Turn the Scorecard Into Next Week’s Buy List

Scoring is wasted if it does not change who gets in the van.

Write a one-page repurchase card

Before each sale or each week of online buying:

1. List the top repurchase patterns (segment + band + source).
2. List the do-not-replace patterns.
3. Note remaining slots from the mix plan and from your sales-pace acquisition budget.
4. Pre-write walk-away cost notes for the open patterns. Detailed walk-away math belongs in setting auction walk-away numbers before you start bidding.
5. Leave two blank lines for true exceptions, each of which needs a GM initial.

Use trades as the first fill, auction as the second

If sold history says trades in a given pattern outperform auction cars, the appraisal desk is the primary buyer. Coach the desk to stretch for those trades and to wholesale the patterns that scored poorly. Auction time should fill remaining gaps, not redo the whole mix.

Confirm the unit can be exposed, not just owned

A pattern that sold well only after two weeks of photos, recon, and listing work is a slower pattern than the sold date suggests. Check whether those units were actually advertised promptly. A car that sat unlisted for ten days and then sold in four is a posting problem, not a buy signal to add more of them. Review that gap in the same weekly meeting where you look at unposted ready units.

Illustrative walk-through

Consider a lot that scored last month’s solds and found:

• Compact crossovers in a mid price band, under a moderate mileage cap, mostly trades: high speed, acceptable gross.
• Full-size two-wheel-drive trucks bought at auction: slow, recon over cap, sold only after a cut.
• Certified late-model sedans: average speed, strong gross, limited supply.

Next week’s card is then: stretch for crossover trades, skip loose 2WD auction trucks, and take certified sedans only if they land inside cost and recon caps. The example is illustrative. Your file will read differently. The method does not.

Traps That Make Sold History Lie

Bad inputs create confident, wrong buy lists.

Survivor bias

You only see what sold. You do not automatically see the twins that are still on the lot. Always read sold history next to current inventory of the same pattern. If you sold two and still hold seven, you do not have a repurchase signal. You have a surplus.

Price and pack distortion

A unit that "made money" after a pack, a pack-out of a trade, and a backend product load is not proof that the next one will. Score on front-end after recon, then glance at backend separately. Do not let F&I rescue a bad buy on the scorecard.

One manager’s pet units

If the same rep keeps discounting a body style until it sells, the sold file will bless that body style. Look at original list price vs. sold price. Patterns that only move after repeated cuts are not repurchase patterns.

Title, odometer, and history surprises

Units with branded titles, odometer issues, or incomplete history can sell once to a unique buyer and never again. Do not clone them. Operational listing and disclosure questions are a separate discipline; this article is about buy decisions. Confirm title and history with your usual vendors and counsel before you treat a one-off sold as a program.

Season and fuel spikes

A two-week run on a segment during a weather event is a weather event. Keep it in the 30-day window as context, and do not rebuild the 90-day mix around it.

How PostToMarket Automates This

Sold history is only useful if "sold" is a real status, dated, and no longer advertised. PostToMarket does not replace your DMS sold report. It does stop the operational leaks that make the report and the lot disagree.

Keep the sold file honest

• Mark as sold and archive from inventory or the queue so a funded unit leaves Marketplace the same day.
• Inventory statuses so a pending or failed post is not confused with a live, available car.
• Leads inbox so managers can see which units actually produced conversations before anyone claims a pattern "has no demand."

Get the next purchase into inventory on day one

• Capture from auction and wholesale sites when the buy matches a repurchase pattern.
• Bulk CSV and Excel import with a row-level error log when you bring in a pack.
• AI-generated titles and descriptions from that unit’s specs, editable before posting, so the replacement car is not a clone of last month’s template.

Close the loop between what sold and what gets posted

• Per-rep queues and manager assignment so the next units in a winning pattern are owned and listed, not sitting in a shared folder.
• Team analytics by rep and source to see whether the pattern is moving through Marketplace or only through other channels.
• Mobile dashboard so a manager can reorder the queue toward the patterns the sold file just blessed.

The buy decision is still yours. The tool’s job is to make sure the cars you chose to replace are visible to buyers instead of repeating the "we sold it, then took a week to list the next one" cycle.

Let the Sold File Write the Buy Card

Using your sold history to decide what to buy next is how you stop shopping your own taste. Score retail solds for speed, gross, recon, and demand quality. Roll those scores into a short repurchase card. Read the current lot before you treat a sold pattern as a shortage. Then buy only what the card and the mix plan still have room for.

PostToMarket keeps the operational side of that loop tight: archive sold units, capture the replacements, and queue them for paced posting with unique copy. The Starter plan runs $59 per month for a solo seller or independent lot (one login). The Dealership plan runs $299 per month for up to 13 logins.

Start your 10-day free trial via secure Stripe checkout. Get this week’s solds off the lot in the system, queue the replacements, and cancel anytime.