Matchback analysis connects the vehicles you sold, and the repair orders you wrote, to the marketing that reached those buyers. You take the sold report from your DMS, compare it with who saw or clicked your campaigns, and see which channels sit behind real transactions instead of form fills. Done well, it settles the argument between vendor reports that each claim credit for the same sale. Done badly, it hands you a confident number that proves nothing.
This guide explains how a dealership matchback works, which data you need, which metrics matter, and where the method breaks down.
TL;DR
- Matchback starts from the sale and works backward to the ad exposure. It does not start from the lead.
- You need a clean DMS sold report, campaign and spend data by channel, and a defined attribution window.
- Judge results on matched sales, match rate, cost per sale and influence sales, not on leads or clicks alone.
- A high match rate is not an accurate match rate. Validate before moving budget.
- Matchback shows association, not causation. Use it to ask better questions of vendors, then test.
Table of Contents
- What matchback analysis means in a dealership
- How the matching works, step by step
- Choosing an attribution window and rule
- Metrics that show whether marketing contributed
- Where matchback misleads
- Turning findings into budget decisions
- Our perspective
- Getting an independent read on your own data
- FAQ
- Sources
What matchback analysis means in a dealership
In most industries a matchback compares a list of marketing recipients with a list of purchasers. Direct mail made it famous: mail a file, then match orders back to it. Automotive adopted the same logic and widened it to digital. You begin with confirmed sales, then identify which of those buyers were exposed to your advertising before they bought.
The output is not a list of named shoppers. In a typical setup the dealership uploads a sold report, customer fields are hashed, and an ad platform compares the hashed values against its own users. What comes back is aggregate: how many sales matched, and which campaigns those buyers saw or clicked. Google describes a similar process for store sales measurement, where hashed in-store transaction data is matched to signed-in Google users. Reporting only appears once match and attribution thresholds are met, and the feature is limited to eligible advertisers in supported markets.
The same logic works on the service drive. A repair order is a transaction like a vehicle sale, so it can be matched to a campaign too. We cover that in more detail in our piece on fixing service appointment attribution.
Why it matters more than lead reports
A lead report counts form fills, calls and chats. A matchback counts sold units. The gap between the two is where most wasted spend hides. A channel can produce cheap leads that rarely buy, and another can produce few leads but a steady flow of buyers who saw its ads and then walked in or called.
Cox Automotive's Transparency study of dealership transactions found that on average 58% of sales and service transactions involved consumers who had visited Autotrader and Kelley Blue Book within four months of purchase. The study dates from 2016, so treat the percentage as a historical reference. The lesson holds: a large share of buyers touch media you never see in a lead report.
How the matching works, step by step
The mechanics are similar whether you run them in-house, through a vendor tool or through an independent review.
- Export the sold report. Pull deals from the DMS for a defined period. Include sale date, deal type (new, used, service), and the customer fields you intend to match on, such as name, address, email and phone.
- Clean the file. Remove duplicates, fleet and wholesale deals, test records and obvious typos. Standardize addresses and phone formats. Bad inputs lower the match rate and raise false matches.
- Hash and upload. Customer fields are hashed before they leave your control, and the platform compares the hashed values with its user records.
- Apply an attribution window. The window is the period after exposure during which a purchase must occur to count toward a campaign. Choose it before you look at results.
- Read the aggregate report. You receive matched sales by campaign, with spend alongside. Divide spend by matched sales to get cost per sale.
- Reconcile against your own totals. Compare matched units with total units sold. This is the match rate, and it frames everything else.
Data you need before you start
| Input | Where it comes from | Common problem |
|---|---|---|
| Sold deals and repair orders | DMS | Missing email or phone, duplicate households |
| Campaign and spend data | Ad platforms, vendor invoices | Spend that does not match invoices |
| Lead and appointment records | CRM | Sources mislabeled or left as "other" |
| Site and event data | Analytics | Conversion events firing twice |
| Attribution rules | Written down by leadership | Different windows used by different vendors |
If your analytics events and CRM imports are unreliable, fix them first. Our 90-minute GA4 audit for dealerships walks through the checks.
Choosing an attribution window and rule
The window you choose changes the answer. A one-day view window credits only campaigns seen immediately before a sale. A 28-day window credits far more, and a longer one credits more still. None of these is correct in the abstract. A shopper deciding on a used compact in a week behaves differently from a family spending two months on a three-row SUV.
Two practices keep this honest:
- Fix the window before the data arrives. Changing it after seeing results until a vendor looks good is how reports end up contradicting each other.
- Report more than one rule. Google's dealer guidebook notes that 90% of shoppers switch between screens and recommends testing models such as linear, position-based or time decay rather than relying only on last click. The guidebook is from 2017, but the advice on testing multiple models still applies. Our guide to dealership attribution modeling shows how position-based rules play out in practice.
Also separate two kinds of credit. A buyer who clicked an ad and then purchased is direct. A buyer who was reached by your ads but purchased outside the window is only influenced. Reports that merge the two overstate what a channel did inside the window.
Metrics that show whether marketing contributed
Use a small set and define each one in writing.
- Matched sales: sold units tied to at least one campaign exposure.
- Match rate: matched records divided by uploaded records. It tells you how much of your sales base the analysis can see.
- Cost per sale: channel spend divided by matched sales. This is the number to hold next to cost per lead.
- Influence sales: buyers who engaged with ads but bought outside the attribution window.
- Share of sales by deal type: new, used and service matched separately, since the mix of channels differs.
Cost per lead flatters some channels and punishes others. A channel at a low cost per lead and a high cost per sale is not cheap. For a worked comparison on paid social, see our Facebook ads benchmarks for dealers, which sets cost per lead against cost per sale.
A worked example with invented numbers
The figures below are illustrative, not client data.
A store sold 300 units in a quarter and uploaded all 300 records. The platform matched 120, so the match rate is 40%. Paid social spend was $12,000 and matched to 24 sales, a cost per sale of $500. Paid search spend was $15,000 and matched to 30 sales, a cost per sale of $500. On leads alone, social looked far cheaper. On matched sales the two are equal.
But 180 sales were not matched. Some of those buyers may have seen the ads under different emails or without being signed in. So the right read is not "social and search perform the same." It is "within the visible 40%, they cost the same per matched sale, and we cannot see the rest." That wording survives a challenge from a vendor or an owner.
Where matchback misleads
Match rate is not accuracy
A 2023 CIMM study of television data matching makes a point that carries over to dealership data: match rate measures how many records two datasets share, not whether the matches are right. Outdated identifiers can produce incorrect matches, and independent validation is needed to judge accuracy. Its practical advice is to weigh coverage and bias together. A larger match rate helps with scale, but identifier validity and the makeup of the matched sample decide whether you can trust the result.
For a dealer, that means asking who is missing. People who use a different email for shopping than for the paperwork, buyers on a spouse's account, and cash buyers with thin records all fall out. If the missing group skews toward a particular channel's audience, the matched result is biased even when the match rate looks healthy.
Exposure is not cause
A matched buyer saw your ad and bought. That does not prove the ad caused the purchase. Someone already committed to your brand may have seen a branded search ad on the way to your site. Retargeting is especially prone to taking credit for sales that were coming anyway. Matchback tells you where buyers were reached. Holdout tests, geo splits or paused-spend experiments tell you what the spend added.
Walk-in and phone sales
A buyer who calls from a vehicle detail page, then comes in, may be recorded in the CRM under a different source than the one a platform sees. If your phone tracking is weak, matchback will understate the channels that drive calls.
Channels without a click
Billboards, radio, local sponsorships and branded vehicles produce no clickstream, so platform matchbacks cannot see them. Direct mail can be matched on name and address, and some offline media can be assessed with market-level comparisons. Leaders sometimes look at how other businesses approach vehicle wrap ROI tracking when thinking about measuring non-digital exposure. Whatever the channel, state in the report what it could not measure.
Turning findings into budget decisions
The point of the exercise is a decision. A workable sequence:
- Rank channels by cost per matched sale and by influence sales. Look at both, because a channel strong on influence may be doing top-of-funnel work.
- Flag the gaps. Where a vendor's claimed sales exceed matched sales by a wide margin, ask for their method in writing.
- Move a small share first. Shift a modest portion of spend from the weakest channel to the strongest and watch the next cycle. Do not rebuild the budget on one quarter.
- Test what matchback cannot prove. Pair the shift with a holdout or a pause in one market segment.
- Repeat on a schedule. Quarterly is common, as it gives enough sold units for patterns to show.
The industry direction supports this discipline. When Cox Automotive announced its Fullpath acquisition, President Steve Rowley described a shift from manual, campaign-based marketing to always-on, data-driven retail strategies. More automation raises the value of someone who can check what the dashboards claim.
For a full 90-day approach built on sold-unit data, read our guide to vendor-agnostic sales match analysis. If paid search is where you suspect leakage, the paid search audit shows how to prove DMS sales before changing bids.
Our perspective
Most dealerships we speak with are not short of reports. They have too many, and each is produced by a party with something to sell. A vendor measuring its own campaigns will tend to find that its campaigns worked. That is not dishonesty. It is the result of choosing the window, the match logic and the credit rule inside the same company that invoices you.
So we treat matchback as a tool for accountability rather than a verdict. We do not sell advertising, manage campaigns or take commissions from vendors, which means we have no channel to defend. We start from your sold report, apply one attribution rule to every vendor, and show where the numbers disagree and why. Then we recommend what to keep, what to question and what to test.
The discipline matters more than the software. Define the window first. Report match rate beside every result. Separate direct credit from influence. Never move a large budget on a single matched quarter. A number you can defend in front of an owner is worth more than a larger number you cannot.
— AutoROIQ
Getting an independent read on your own data
If your vendors' reports do not add up to your sold units, an outside review is the fastest way to find out why. Our Marketing Intelligence Review compares vendor claims with your DMS results and delivers executive-level recommendations. Ongoing strategic advisory is available if you want the analysis repeated each quarter. See the Marketing Intelligence Review and advisory details or start from the AutoROIQ home page.
FAQ
What is matchback analysis in automotive?
It is a method that starts with vehicles sold, or service repair orders written, and works backward to the advertising those customers were exposed to. The result shows matched sales by campaign and, with spend, a cost per sale. It reports aggregate associations, not the identities of individual shoppers.
What match rate should a dealership expect?
There is no universal figure, because it depends on how complete your customer records are and which platform does the matching. Track your own rate over time and compare it with total units sold. A low rate does not make the result useless, but it limits how much of your sales base you can claim to understand.
Can matchback prove that an ad caused a sale?
No. It shows that a buyer was exposed to a campaign before purchasing. To estimate what the spend added, run a holdout, a geographic split or a planned pause, and compare sales across the groups.
Does matchback work for service?
Yes. Repair orders can be matched to campaigns in the same way as vehicle sales, provided the DMS export includes the customer fields needed for matching. Report service results separately from sales, since the channels that bring in service customers are often different.
Sources
- About store sales measurement, Google Ads Help
- The Dealer Guidebook, Google
- Transparency study white paper, Cox Automotive
- Television Data Match Rate and Match Bias Study, CIMM
- Cox Automotive to acquire Fullpath, NIADA
- Vehicle wrap ROI