The CRM audit checklist that cuts unused tools and gets cleaner data in 60 days.

A professional services firm with about 60 people was paying $1,400 a month for software no one was using, and the CRM audit checklist in this guide starts where that cleanup started, with the bill. The tools had been added over the years for good reasons, a scheduling app here and a survey tool there, and the list of 17 subscriptions hadn't been reviewed in a while.

I went through it with the firm's team. Six subscriptions with no user were canceled after their data was exported, three jobs moved into the CRM, and every tool that stayed got a named owner and a reminder 30 days before any notice deadline.

Run the audit before the next renewal and next year's budget.

Why a CRM audit checklist starts with the bill and ends with the data.

In Validity's State of CRM Data Management in 2025 (602 CRM users and administrators in the US, UK and Australia), 76% said less than half of their organization's CRM data is accurate and complete, and workers reported spending an average of 13 hours a week hunting for basic information in the CRM. Validity sells data-quality tools. For example, even at half that figure, a team of ten loses more than one full-time person's week to searching.

Tadhg Nagle, Thomas Redman and David Sammon had 75 executives in programs in Ireland pull the last 100 records their department created, mark obvious errors in 10 to 15 critical fields and count the clean ones (Harvard Business Review, 2017). On average, 47% of new records had at least one critical error, half the scores were below 57 out of 100, and only 3% met the "high nineties" bar the managers themselves called acceptable. The 2017 results give the test below a benchmark.

The same HBR article offers a first estimate of what flawed records cost, that a task done with flawed data costs about ten times as much as one done with clean data, and that cost shows up in the team's hours, not on the CRM bill.

In February 2002, three Bain & Company authors reported that the firm's 2001 survey of 451 senior executives ranked CRM in the bottom three of 25 management tools for satisfaction. Their diagnosis was that companies expected the software to manage customer relationships for them. One tour company in the article spread one operator's handwritten thank-you notes companywide before it looked at software.

"Successful CRM depends more on strategy than on the amount you spend on technology." Darrell K. Rigby, Frederick F. Reichheld and Phil Schefter, Harvard Business Review, February 2002

A new tool rarely fixes a setup problem, and the audit tells you which problem you have before you pay for an answer.

Step 1. Audit every subscription before it renews.

1. Pull 12 months of card statements, invoices and expense reports, where expensed tools show up.
2. For each tool, write down who uses it, what job it does, the cost, the renewal date, the notice deadline and where its data lives.
3. Mark the tools that do a job the CRM edition you pay for already covers, like forms, meeting links, email or reporting.
4. Export the data from every tool with no user, name the file with the date, for example "2026-11-survey-tool-export", then cancel.
5. Put every renewal on a shared calendar with a name next to it and a reminder 30 days before the notice deadline.
6. Check paid CRM seats against each user's last active date (in HubSpot, the Seats tab in Users & Teams, and in Salesforce, the Last Login field on each user).

At a larger company, the work is matching each contract to a business owner. Zylo's 2026 SaaS Management Index, built on its customers' spend and license data, puts the average portfolio at 305 applications. Zylo sells SaaS management software. At that scale, pull usage from single sign-on logs.

Step 2. Ask the team what slows them down, and why.

Book 15 minutes each with two salespeople, a marketer and a manager. Ask which screen they avoid, what they keep outside the CRM, and which report they don't trust, then watch one person log a call and count the clicks. At a larger company, send the questions as a survey by role.

Step 3. Keep a required field only when someone uses it.

A required field stays when it feeds a report someone reads or a decision someone makes, like routing, scoring or the forecast.

• HubSpot. Required fields sit under Settings, Data Management, Objects, "Customize the 'Create [object]' form" (Starter and up). Conditional stage properties ask for a field only at a given stage, and property insights in the data quality tools flag properties with no data or no use.
• Salesforce. Review field-level requirements, page layouts and the validation rules that block a save, and retire the ones no report depends on.
• Mailchimp, Constant Contact or a spreadsheet. Cut the signup form and import template to the columns someone reads.

The usual pushback is that required fields get you data, and they get a value, not always a true one. Validity's 2025 report says 37% of staff regularly fabricate data to tell leaders what they want to hear, and a field required before anyone knows the answer invites it. Ask for a field when the person knows the answer, like the amount at "Proposal sent".

Step 4. Merge duplicates, and re-engage before you archive.

Run the duplicate tools first, Manage duplicates in HubSpot or, in Salesforce, a duplicate job (Performance and Unlimited) or manual merges, with duplicate rules on for new records. Then filter contacts with no activity in two years (in HubSpot, Last activity date and Last engagement date before the cutoff or unknown, and Create date before the cutoff), leaving out existing clients and open deals, and send them one short email in small batches, pausing if bounces climb. HubSpot's "Don't send to unengaged contacts" switch is on by default and skips anyone who ignored their last 11 to 16 marketing emails, and those contacts go straight to the export. Two weeks later, export the non-responders, then delete them or set them as non-marketing.

In HubSpot, deleted contacts and companies can be restored for 90 days, except permanent or GDPR deletions. Non-marketing contacts (Marketing Hub Starter and up) don't count toward the contact tier from the next update date, and HubSpot lowers a contact tier only at renewal, which makes the weeks before a renewal the time to do it. The HubSpot marketing contacts guide covers that bill.

Step 5. Rename pipeline stages to steps the prospect takes.

A stage should name something you can check happened, like "Discovery call held", "Proposal sent" or "Contract sent", where "In progress" names nothing. Write one exit rule per stage, like "Proposal sent: the proposal is attached to the deal." If a stage's meaning changes and not only its name, create a new stage and move deals on purpose.

Step 6. Automate the task that takes the most time.

HubSpot's Sales email extension logs emails automatically, and a workflow with the "Create task" action (Professional and Enterprise) can add a follow-up task when a deal reaches "Proposal sent". A task doesn't fill Next activity date, which HubSpot sets from a future call, sales email or meeting, and booking the next meeting from the deal is what clears count 1. In Salesforce, a record-triggered flow does the same job.

Step 7. Set up the weekly CRM check.

Start with five counts, using HubSpot's property names:

1. Open deals with Next activity date unknown.
2. Open deals with Close date before today.
3. Open deals with Amount unknown.
4. Contacts with Contact owner unknown, from the lead stage onward.
5. Contacts with Lifecycle stage unknown.

Open means Is Closed Won and Is closed lost are both false, in every pipeline. In HubSpot Professional or Enterprise, build each in the custom report builder as a KPI chart (or a gauge with colored bands for the target), put the five on one dashboard, and use Share, "Email this dashboard" to send it weekly, Monday at 7am. Recurring dashboard emails go only to HubSpot users. On Starter or the free tools, save each filter as a view and open the five views on Monday.

In Salesforce, count 1 is open opportunities with Next Step blank (a standard field), count 4 is records owned by an inactive user, an integration user or a lead queue, and count 5 is leads still at the default Lead Status after seven days. Put the five reports on one dashboard as metric components, which show one number each, and subscribe the sales and marketing leads to it weekly on Monday. With no CRM, use one COUNTIFS formula per count.

Send team totals, with each person seeing their own list inside the CRM, because a list of names with the most gaps turns a data check into a ranking.

Step 8. Give it 60 days, then decide.

Switching makes sense when the CRM can't connect to the tools you need or the team has outgrown it in permissions, objects or volume. The cleanup still pays off, because clean data moves faster, and the one-page marketing report depends on these counts staying near zero.

A worked example with the five counts.

Example numbers: a company with $1.9M in open pipeline across 140 deals finds 31 deals past their close date worth $420,000, 26 with no next activity, 18 with no amount, 40 contacts with no owner and 55 with no lifecycle stage.

The past-due deals are 22% of the pipeline ($420K ÷ $1.9M). If two-thirds close as lost, the pipeline drops about $280,000 to roughly $1.62M.

The unowned contacts have a price too. LocaliQ's 2026 search advertising benchmarks, from thousands of its customers' Google and Microsoft Ads campaigns, put the average cost per lead at $66.69 across industries and $93.69 for business services, and LocaliQ sells advertising services. If those 40 came from search ads at the business services rate, that's about $3,748 paid for leads with no one assigned to follow up (40 × $93.69), the gap the inbound lead follow-up guide is built to close. Day-60 targets: 5 or fewer past-due deals, 10 or fewer with no next activity, and zero in the other three counts.

Common mistakes, and the pushback a CRM admin would raise.

• Moving close dates without a next step. Any date change comes with a scheduled next activity.
• "Next activity date can be faked with a placeholder meeting." It can, which is why the 100-record score below counts a next activity as clean only when the prospect is on it.
• "Five counts is too simple for an org with 400 users." At a larger company, keep the five counts, filter the dashboard by team, add a data-quality score per object and give it to the revenue operations owner.

A test you can run on your own CRM.

This is the method from Nagle, Redman and Sammon's study, adapted for a CRM.

1. Export the last 100 deals created (or 100 contacts).
2. Pick 10 to 15 fields that matter, like owner, amount, close date, stage, next activity, lifecycle stage and source.
3. Mark each record clean or not, and the clean count is your score out of 100.
4. Run the weekly check until day 60, then score the last 100 records created after day 1, or all of them if there are fewer, as a percentage.

Measure the score, the five counts and the hours a week the team spends looking for information. A result is a day-60 score at least 15 points above the first one, with four of the five counts at target (this guide's threshold, not a research figure). A score that hasn't moved points at the setup, and a higher score with a job the CRM still can't do points at switching.