Week 1: CRM Hygiene -- Stop the Data Decay

A VP of Sales at a Series B SaaS company called me last year, frustrated. They had just spent forty thousand dollars on an AI-powered forecasting tool. The vendor promised ninety percent forecast accuracy within sixty days. After three months, the tool's predictions were worse than the sales manager's gut feel. The VP wanted to know what went wrong with the AI. Nothing went wrong with the AI. Everything was wrong with the data feeding it.

When we audited their CRM, we found that thirty-eight percent of open opportunities were missing close dates. Twenty-two percent had amounts of zero or one dollar — placeholder values reps entered to bypass required field validation. Fourteen percent of contacts associated with active deals had bounced email addresses. The AI model was training on garbage and producing garbage. No algorithm, no matter how sophisticated, can compensate for data that is fundamentally broken.

This is the dirty secret of AI in sales operations: the limiting factor is almost never the AI. It is the data. Every AI tool you deploy in the next four weeks will only be as good as the data it consumes. That is why we start with hygiene, and that is why hygiene is not a one-time project but a permanent operating discipline.

Your CRM is decaying right now. Every day your reps skip a field, enter a company name differently than the last person, or leave a call disposition blank, your data gets a little worse. Most teams do not notice until a board meeting when the pipeline number does not match reality, or until an expensive AI tool produces nonsensical output.

Data decay follows a predictable pattern that I call the Decay Compounding Curve. In the first month after a CRM cleanup, data quality holds relatively steady — maybe declining one to two percent. By month three, the decline accelerates because early errors create downstream errors. A misspelled company name leads to a duplicate record, which leads to split activity history, which leads to inaccurate lead scoring, which leads to misrouted leads. By month six, you have lost twenty to thirty percent of the quality gains from the original cleanup. By month twelve, you are back where you started.

The reason most data quality initiatives fail is that they treat the problem as episodic rather than continuous. You cannot solve a daily problem with a quarterly project. It is like trying to stay fit by going to the gym once every three months for eight hours. The math does not work. You need a daily habit — small, consistent, and integrated into your existing workflow.

Before you start fixing anything, you need to know where you stand. Run this assessment today. It takes thirty minutes and gives you the baseline every future measurement will reference.

Step 1: Field Completeness Rate. Export all active opportunities (created or modified in the last ninety days). Count the total number of required fields across all records. Divide the number that are actually populated with real data (not placeholders like "TBD" or "$1") by the total. This is your completeness rate. Most teams score between fifty-five and seventy percent. Write this number down.

Step 2: Contact Coverage. For every active opportunity, check whether at least one valid contact is associated. Then check whether the primary decision-maker (economic buyer, champion, or technical evaluator) has a contact record with a working email and phone number. Most teams have eighty percent opportunity-level coverage but only forty to fifty percent decision-maker coverage.

Step 3: Duplicate Rate. Run your CRM's built-in duplicate detection (or export accounts and do a fuzzy match on company name). Calculate the percentage of records that have at least one likely duplicate. The industry average for B2B CRMs with more than five thousand accounts is twelve to eighteen percent.

Step 4: Data Freshness. Pull all contacts associated with active opportunities. Check the last activity date on each contact. What percentage have had no logged activity in the last thirty days? These are "stale contacts" — they may have changed roles, left the company, or simply disengaged. On most teams, twenty to thirty percent of opportunity contacts are stale.

This is not a one-time cleanup. One-time cleanups are like crash diets — they work for a week and then everything goes back to normal. Instead, you are going to build a daily habit that takes thirty minutes and prevents decay from compounding. I call this the Three-Block Routine because it splits into three ten-minute blocks that align with natural breaks in your workday.

Block 1: The Morning Scan (10 minutes). Before your first meeting, open your CRM dashboard and check three things: (1) any new leads from overnight that are missing required fields — name these explicitly in your process, such as company name, source, and lead score inputs, (2) any deals that moved stages yesterday without updated close dates or amounts, and (3) any contacts that bounced from last night's email sends. Fix what you can in ten minutes. Flag anything that needs the rep's input with a specific task assigned to them with a due date of today.

Block 2: The Midday Audit (10 minutes). After lunch, pull a quick report showing all opportunities modified today. Check that stage changes have corresponding notes explaining why the deal moved. Check that any new contacts added today have company associations — orphaned contacts are one of the largest sources of duplicate records. Check that any meetings logged today have outcomes recorded, not just "meeting held."

Block 3: The End-of-Day Review (10 minutes). Before you log off, run your duplicate detection. Most CRMs have built-in duplicate matching — turn it on if it is off. Review the top five suggested merges. Merge the obvious ones, skip the ambiguous ones. This alone prevents eighty percent of the duplicate problem. Then check your data completeness dashboard — you built one in the baseline assessment — and note whether today's number went up or down.

To make hygiene sustainable, you need a visible, shared metric. I use a framework called the Data Hygiene Scorecard, which tracks four dimensions weekly and produces a single composite score from zero to one hundred.

Completeness (25 points): Percentage of required fields populated across active opportunities, converted to a 25-point scale. Eighty-five percent completeness equals roughly 21 points.

Accuracy (25 points): Percentage of records that pass validation checks — valid email formats, phone numbers with correct digit counts, amounts greater than zero, close dates in the future for open deals. This catches the "technically populated but actually garbage" problem.

Freshness (25 points): Percentage of active opportunity contacts with logged activity in the last thirty days. This measures whether your data reflects current reality or historical fiction.

Uniqueness (25 points): Inverse of your duplicate rate. If twelve percent of records have duplicates, your uniqueness score is eighty-eight percent, which converts to twenty-two points.

Add the four dimensions. A score above eighty means your data is AI-ready. Between sixty and eighty means you have work to do but can start deploying basic AI tools. Below sixty means AI tools will produce unreliable output and you need to focus exclusively on hygiene before investing in anything else.

1. 1.Treating hygiene as the admin's job. Data quality is everyone's responsibility. If only one person is fixing data while twenty people are creating bad data, the math never works. Reps need to own the accuracy of their records. Managers need to enforce it in pipeline reviews. The admin coordinates — they should not be the sole contributor.
2. 2.Building validation rules that are too strict. Over-zealous required fields and validation rules do not improve data quality. They train reps to enter garbage that passes validation. A required "Competitor" field results in hundreds of records with "N/A" or "None" — which is worse than a blank field because it looks like real data in reports. Only require fields that reps genuinely know the answer to at that stage of the deal.
3. 3.Cleaning up without fixing the source. If reps are entering bad data because the page layout is confusing, the picklist values are outdated, or the field is on the wrong object, cleaning the existing data without fixing the input mechanism just creates a recurring cleanup project. Fix the inputs first.
4. 4.Measuring activity instead of outcome. Tracking "number of records cleaned this week" is an activity metric. It tells you someone is busy but not whether the system is getting healthier. Track the Data Hygiene Scorecard — it measures the actual state of your data, not the effort being applied.
5. 5.Ignoring the enrichment opportunity. Manual hygiene is necessary but insufficient at scale. As your data volume grows, you need enrichment tools (Clearbit, ZoomInfo, Clay) that automatically fill in firmographic data, validate contact information, and flag stale records. The daily routine catches the gaps that enrichment misses — they are complementary, not alternatives.

Day 1: Run the Data Quality Baseline Assessment. Record your scores for completeness, contact coverage, duplicate rate, and data freshness. Share the results with your sales manager — they need to see the starting point to appreciate the improvement.

Day 2: Build your hygiene dashboard. Create three saved reports in your CRM: (1) Opportunities missing required fields, filtered to active pipeline, (2) Contacts with no activity in thirty-plus days on active opportunities, (3) Duplicate account candidates. Pin these to a dashboard you will open every morning.

Day 3: Start the Three-Block Routine. Morning scan, midday audit, end-of-day review. Time yourself — if any block takes more than twelve minutes, you are going too deep. The goal is prevention, not perfection.

Day 4: Enlist your team. Share the Data Hygiene Scorecard with your reps. Explain that you are tracking four dimensions and that the team score will be visible. Do not punish low scores — create positive accountability by recognizing reps whose records are consistently clean.

Day 5: Measure your first week. Re-run the baseline assessment. Compare to Day 1. You should see completeness improve by ten to fifteen percentage points and duplicate rate drop by three to five points. If you do not see improvement, the routine is not being followed consistently — diagnose which block is being skipped and why.

Days 6-7: Refine and repeat. Adjust your saved reports based on what you learned this week. Add any fields to your tracking that you discovered were consistently problematic. Document any systemic issues — a field that everyone fills incorrectly, a picklist missing a common value — and fix those root causes so next week's routine is faster.

By the end of week one, your Data Hygiene Scorecard should be above seventy (up from the typical starting point of fifty to sixty). Your completeness rate should be at eighty percent or above. Your duplicate rate should be trending down. Most importantly, your team should understand that data hygiene is a daily discipline, not a quarterly project.

A team with good hygiene habits looks like this: reps update records the same day an interaction happens, not during a Friday afternoon blitz. Managers reference the CRM in pipeline reviews instead of asking reps to "walk me through your deals." New leads are routed and enriched within hours, not days. And when you plug an AI tool into this data in the coming weeks, it actually works — because the foundation is solid.

Clean data is necessary but not sufficient. Now that your CRM is producing reliable signals, you need a system that knows which signals matter most. In Week 2, you will build a lead scoring model that uses your freshly cleaned data to prioritize your pipeline — so your reps stop guessing which leads to call first and start following the math.