Dealership CRM Audit Checklist

Before you replace the system, find out whether the system is the problem. Seven areas, one afternoon, a score you can act on.

Almost every CRM replacement starts with a specific grievance and ends with a signature eight weeks later. What rarely happens in between is an honest look at whether the current system is failing or whether the store is failing to use it.

This is not a defense of vendors. Plenty of automotive CRMs deserve to be replaced, and stores stay on bad ones far too long out of switching fear. But a store that changes systems without diagnosing its own process carries every one of its habits into the new system, spends four to six weeks in a productivity dip, and arrives at the same complaints with a different login screen. We have watched that cycle repeat more than once, and the second switch is always harder to justify to ownership than the first.

So run the audit first. It takes an afternoon with a laptop and honest access to your own reports. At the end you will have a list split into two columns: things a different vendor could fix, and things only management can fix. If the second column is longer, do not go shopping yet. If the first column is longer, you now have a specific requirements list to take into a demo instead of a vague sense of dissatisfaction, and the requirements checklist turns it into something a vendor has to answer.

Start where leads enter, because everything downstream inherits whatever is broken here.

List every source that can create a lead in your store: website forms, third party providers, chat, phone calls, walk ins, seller or acquisition leads, service referrals, salesperson personal contacts. For each one, answer three questions. Does it create a record automatically or does a human retype it? Is the source labeled distinctly, or does everything land as website? What happens if it stops arriving, and who would notice?

That last question is the one that catches stores out. A form breaks, submissions stop, and nobody notices for eleven days because there is no alarm for the absence of leads. Test it yourself. Submit your own form, start your own chat, call your own main number after hours, and time how long it takes for a human being to contact you. Do it on a Saturday too, because Saturday is a different store.

Pass looks like: every source creates a record automatically, every source is distinctly labeled, and someone is alerted when a source goes quiet. Fail looks like: retyping, a single generic source label, and no monitoring at all.

Ask your managers what your average first response time is. Write down the answer. Then pull the actual number from the system. In most stores the gap between those two figures is the single most useful discovery of the whole audit.

Measure the median rather than the mean, because one lead answered four days later ruins an average and hides an otherwise healthy pattern. Then break it down by hour of day, by day of week, and by source. The interesting failures are always at the edges: leads arriving at 7pm, leads arriving Saturday afternoon during a busy floor, leads from the source nobody respects.

Check the difference between the first automated response and the first human response. An instant auto reply is worth having and is not the same thing as contact. Some stores discover their reported response time is entirely automated and their human median is six hours.

Pass looks like: a median measured in minutes for human contact during business hours, with a defined and staffed after hours path. Fail looks like: nobody can produce the number at all, which is itself the finding.

Take fifty leads that were closed lost in the last quarter and count the actual touches on each: calls attempted, texts sent, emails sent, before the record was closed.

Ask your team first what they think the average is. You will usually hear five or six. The measured number is usually two, occasionally three. That gap is where a large share of your lost gross lives, and it is not a software problem in most stores. It is what happens when follow up depends on human memory during a busy week.

Then check three related things. Do leads get closed with a reason attached, or just closed? Is there a defined cadence that applies automatically when a lead is created, or does each salesperson invent their own? And how many leads are sitting in an ambiguous state where no one is sure whether they are being worked?

Pass looks like: a written cadence applied automatically, an average depth above six touches, and every closure carrying a reason. Fail looks like: depth of two, no reasons, and a large pool of leads in limbo. If this is your finding, follow up processes, task automation and drip campaigns fix it faster than a new vendor will, and the cadence guide gives you a starting shape.

Dirty data quietly makes everything else worse. It inflates your lead counts, wastes your team's hours and damages your ability to send anything at scale.

Check four things. Duplicates: search a common last name and see how many times the same person appears under different spellings and numbers. Dead contact data: what share of your phone numbers are actually reachable and what share of your email addresses bounce. Completeness: what proportion of records have a mobile, an email and a vehicle of interest, because a record missing all three is not a lead. And format consistency, particularly phone numbers, because a database with five different phone formats cannot be reliably matched, filtered or dialed.

You do not have to guess at any of this. An email validator and a phone validator are built into the platform rather than sold as a separate service, so you can measure the decay in your own list rather than estimating it. Most stores find between fifteen and thirty percent of an aged list is unreachable, and finding out before a campaign rather than during one protects your sending reputation.

Pass looks like: a deduplication routine that runs, validation before any bulk send, and a required field standard at capture. Fail looks like: nobody has ever measured it. The data cleanup guide is the fix in order.

This section is short and the stakes are the highest on the page, because the failures here are not merely expensive in gross.

Check that consent is captured and recorded rather than assumed, and that you can produce the record of it for a specific customer in under a minute. Check that an opt out applies across the entire platform and every department rather than to one campaign, because a customer who unsubscribes from sales and then receives a service blast has both a legal complaint and a personal one. Check that your blacklist is real and imported rather than a note in someone's head. Check the hours your automated messages actually send, including whether a drip can fire at 6am on a Sunday because nobody set a window.

Then check the shadow channel. If salespeople are texting customers from personal cell phones, none of the above applies to those conversations, you have no record of them, and the customer relationship walks out the door when the salesperson does. That single finding is often the strongest argument for change in the entire audit. Compliance for dealership texting covers the detail.

Pull your user list and read it properly. Two questions usually produce an uncomfortable silence: how many of these people still work here, and how many have access to more than their job requires.

Every store accumulates active logins for departed employees. Every store has at least one account shared between people, which destroys accountability and any hope of meaningful activity reporting. And most stores have permissions set to whatever was easiest on the day, which usually means either everyone sees everything or nobody sees anything outside their own records.

Check whether roles reflect the actual organization: a BDC agent, a salesperson, a sales manager, a used car director and a general manager need genuinely different views. Check whether login activity is logged, because the week a salesperson leaves for a competitor is the week somebody will ask what they had access to and when. Check the offboarding process exists in writing rather than depending on someone remembering to mention it.

Pass looks like: one login per human, roles that match the org chart, logged access and a written offboarding step. Fail looks like: shared logins and former employees still active.

The final test is the simplest. Ask three managers to produce the same number independently, for example appointments set last month. If you get three different answers, your reporting is decorative.

Then check whether the underlying data can even be right. If a meaningful share of communication happens outside the system, on personal phones or in personal email, then every activity report is a report about the subset of work people chose to record. That is worse than no report, because it looks authoritative.

Decide which numbers actually drive decisions in your store and confirm each one is available without a spreadsheet: leads by source, median first response time, follow up depth, appointments set and shown, closing ratio by source, and cost per sold unit by source. Three reporting layers are what you want here, one for individual activity, one for store performance and one for the roll up, so a manager and an owner are looking at consistent numbers rather than two versions of the truth.

Pass looks like: one number, one source, produced the same way by anybody who asks.

Give each of the seven areas a simple grade: working, weak or broken. Then sort every weakness into the two columns from the start of this page.

Things only management can fix: follow up discipline, capture completeness, shared logins, offboarding, deciding which numbers matter, and stopping the personal cell phone habit. Changing vendors will not solve any of these, and a new system will simply make them visible sooner.

Things a vendor change can fix: leads that cannot be routed automatically, communication that lives outside the system, cadences that cannot be automated, no built in validation, reporting that requires a spreadsheet, permissions that cannot express your org chart, and a contract that punishes you for leaving.

If the second column is where your problems sit, the requirements are now specific enough to shop with. Ask every vendor, including us, what a full month costs with the modules you actually need, what the renewal price is, what you can export if you leave and in what format, and how long implementation takes. Then pilot rather than switch: one team, three or four weeks, a defined slice of lead flow, measured against the same period on your current system. Month to month pricing is what makes that possible, and it is why we sell that way. If you want the audit findings turned into a plan, contact us and bring your list.

How long does this audit take?

An afternoon if you have access to your own reports, longer if you have to request exports. The slowest parts are counting touches on fifty closed lost leads and testing your own intake paths, and both are worth doing by hand at least once.

Who should run it?

Someone who will not be defending the results. A general manager, a dealer principal or an outside pair of eyes. If the person who configured the CRM grades their own configuration, the audit tends to come back healthier than the store is.

What is the most common finding?

Follow up depth. Managers estimate five or six touches before a lead is closed lost and the measured number is usually two. The second most common is that nobody can produce a median first response time at all.

Does a bad audit mean we should switch CRM?

Not automatically. Sort every weakness into what only management can fix and what a vendor could fix. If most of it is discipline, capture and offboarding, a new system will expose those problems rather than solve them.

How do we test our own lead intake?

Submit your own website form, start your own chat, and call your main number after hours and on a Saturday. Time how long until a human contacts you. Then check whether anyone would be alerted if a source stopped producing leads entirely.

What if our salespeople text from personal phones?

Treat it as a serious finding. Those conversations are unrecorded, unmanaged, outside your consent and opt out handling, and they leave with the salesperson. Fixing it requires messaging tools good enough that using the CRM is faster than picking up a personal phone.

If the fix is process, we will say so. If it is the software, we will show you ours running on real lead flow. Month to month, no long term contract.