13/13 Gate✓ IQ Certified10/10?

How do I get reps to log accurate next steps in CRM?

📖 3,677 words🗓️ Published Jul 18, 2026

Direct Answer

You get reps to log accurate next steps in CRM by making a specific next step structurally required to advance a deal, small enough to log in under a minute, and reviewed often enough that reps know someone reads it. Concretely: define what a "next step" must contain — an action, a named human on the buyer side, and a real calendar date — then require those three fields together at your mid-funnel stages, template the inputs so logging takes seconds, review them inside your existing weekly deal-review cadence, and coach on quality rather than policing completion. A reader who stops here has the whole answer: structure the field so vagueness is impossible, make logging frictionless, and inspect what you expect on a predictable rhythm. Everything below is the how — the exact validation logic, the 14-day rollout, the coaching moves, and the honest places this fails.

The core mistake teams make is treating this as a discipline problem ("reps are lazy") when it is almost always a design problem ("the field lets them be vague, and nobody looks"). A blank text box labeled Next Step will fill with "follow up," "circle back," and "TBD" forever, because those are the lowest-effort strings that satisfy the field. Change the design of the field and the review, and behavior follows within a few weeks — no motivational speech required.

One precondition: this only works if your calls actually *produce* a real next step. If reps leave discovery without a concrete mutual commitment from the buyer, no validation rule will save you — you'll just force them to type fiction. Fix call quality first, then turn on enforcement.

flowchart TD A[Call or meeting ends] --> B{Did the buyer commit to a concrete next action?} B -->|No| C["Fix discovery: get a real commitment before advancing"] B -->|Yes| D[Log within 2 hours using template] D --> E{Action + named person + real date all present?} E -->|No| F[Cannot save at Stage 3+] E -->|Yes| G[Deal advances] G --> H[Reviewed in Monday deal review] H --> I{Still fresh and specific?} I -->|Yes| J[Forecast trust] I -->|No| K[Flag as stalled, coach the rep]

Why Reps Log Vague Next Steps in the First Place

Before you build a single rule, understand the four root causes, because each one has a different fix and stacking the wrong fix on the wrong cause is why so many "CRM hygiene initiatives" quietly die.

Cause 1: The call produced no real next step. This is the big one. A rep gets off a call where the buyer said "let me think about it" or "reach out after the holidays," and there is genuinely no concrete action. The rep types "follow up next week" because that is the honest summary of a vague call. The vague CRM entry is a *symptom* of a vague call. If you enforce specificity without fixing call quality, reps invent fake specifics — which is worse than honest vagueness, because now your pipeline looks healthy and isn't.

Cause 2: Logging is slow and interrupts the rep's flow. A rep finishes a call, has three more calls stacked, and the CRM makes them click through four screens to update one field. So they batch it for "later," and later never comes, or comes at end of week when memory has decayed. The fix here is pure friction reduction: templates, keyboard shortcuts, mobile logging, or a Gong/Chorus-style tool that drafts the next step from the call transcript for the rep to confirm.

Cause 3: Nobody reads it, so it feels pointless. If a rep has logged detailed next steps for six months and no manager has ever referenced one, they correctly conclude it is busywork and downgrade their effort to the minimum that passes. Reps optimize for what gets inspected. The fix is visibility: pull next steps up in the weekly forecast call so the rep sees that the field feeds a real conversation.

Cause 4: The definition of "good" was never made explicit. Most teams never told reps what a next step is *supposed* to contain. "Log a next step" is not a spec. "A next step is a specific action, tied to a named buyer-side person, with a date on the calendar" is a spec. You cannot hold people to a standard you never wrote down.

Diagnose which of these dominates your team before you touch Setup. If it's Cause 1, you have a discovery problem wearing a CRM costume, and validation rules will make it worse.

The Three-Field Standard That Kills Vague Next Steps

The single highest-leverage change is replacing one free-text Next Step box with three linked fields that are useless when vague:

Next Step Action — what will happen. Free text, but with a length floor and a banned-phrase guard so "follow up" alone cannot save.
Next Step Owner (buyer-side) — the actual human on the customer's side the action involves. Not "the team," not the rep themselves. A named person forces the rep to have an actual relationship and an actual thread.
Next Step Date — a real day on the calendar, and one that must be in the future and typically within ~14 days. "Next week" is not a date; 2026-07-24 is.

The reason the trio works where a single field fails is that vagueness in one field can hide, but vagueness across three linked fields cannot. A rep can fake an action string, but pairing it with a named person and a near-term date makes the fake visible and, more importantly, makes the *real* next step easier to just type than to invent. When all three are present and consistent — "Walk Priya (VP Ops) through the security questionnaire, Thursday 24th, 10am" — you have something a manager can actually inspect and a rep can actually be held to.

A useful mental model: a good next step answers who does what, with whom, by when. Strip any one of those and it degrades into a wish. Your field design should make it structurally hard to omit any of them.

Two design cautions. First, do not require this at every stage — early-stage exploratory conversations legitimately don't always have a hard next step, and forcing one there manufactures noise. Require it from your first *qualified* stage onward (commonly Stage 3 / "Solution" / post-discovery). Second, keep the required set to exactly three. Every additional required field is a tax the rep pays on every save, and taxes get evaded. Three is enough to be useful and few enough to be tolerable.

Enforcement Mechanics: Salesforce and HubSpot

Here is the concrete configuration. Test everything in a sandbox first, because a badly-scoped validation rule can lock reps out of saving legitimately and generate a revolt by lunchtime.

Salesforce — the validation rule pattern. Create a validation rule on the Opportunity object that fires only at your target stages. A robust condition looks like this in spirit:

Fire when StageName is at or past your qualified stage, AND
any of: NextStep is blank, LEN(NextStep) < 15, the linked buyer contact field is blank, or NextStepDate is blank / in the past.

You can add a banned-phrase guard with REGEX — matching case-insensitive variants of "follow up," "TBD," "check in," "circle back," "touch base." Keep the error message *actionable*: not "Invalid next step" but "Next steps at this stage need an action, a named buyer contact, and a future date within two weeks — e.g. 'Send Priya the ROI model, Thu 7/24.'" The error message is a teaching surface; use it.

Salesforce's own documentation covers the mechanics of validation rules and required fields; follow their guidance on formula limits and testing so you don't hit governor issues on high-volume orgs.

HubSpot — the equivalent. HubSpot doesn't have Salesforce-style validation rules across all tiers, so you assemble the same outcome from: making the next-step properties required to move to a stage (deal stage property requirements), a workflow that flags or re-assigns deals where the next step is stale or blank, and conditional property logic so the fields surface at the right stage. HubSpot's knowledge base documents required properties per deal stage and workflow triggers — build the "blank or stale next step" trigger to notify the rep and, on repeat, the manager.

The arms-race warning. Reps learn what your rule checks. Set a 15-character floor and you'll get "follow up with him later this week" (35 characters, still useless). Ban "follow up" and you'll get "reach out." You cannot win the regex war, and pouring engineering hours into ever-cleverer banned-phrase lists is a losing game. The regex is a speed bump that catches the laziest 60–70% of garbage; the remaining garbage is caught by *human review of quality*, not by more regex. Set a reasonable floor, ban the four or five worst offenders, and stop — then invest the saved energy in the weekly sample review below.

The 14-Day Rollout Plan

Do not flip validation rules on for the whole org on a Monday morning. Stage it so reps experience the change as help, not ambush.

Days 1–3 — Audit and baseline. Pull every open opportunity above a meaningful threshold (say $10k, or whatever is material for you). For each, mark whether it has a next step written in the last 7 days, a specific future date, and a named buyer-side person. Compute the percentage that pass all three. This is your baseline — you'll need it to prove the initiative worked, and you'll need it to calibrate expectations with leadership. In most unmanaged pipelines a large share of "active" deals fail at least one of the three checks; seeing the real number is usually the moment leadership stops arguing about whether this is needed.

Days 4–6 — Define and socialize the standard. Write the one-paragraph definition of a good next step (who / what / with whom / by when) and share it with three real before-and-after examples from *your own* pipeline. Reps accept standards illustrated with their own deals far more than abstract policy. Announce that validation goes live in one week, and that the first two weeks are "learning mode" — flags and reminders, no hard blocks.

Days 7–10 — Soft launch (warnings only). Turn on the fields and the reminder workflow, but not the hard save-block. Reps get nudged when a next step is missing or stale, and see the new fields, but nothing stops them saving. Watch the pass-rate climb and note which reps and which stages lag. This is where you catch legitimate edge cases (a stage where a hard next step genuinely doesn't apply) before they become support tickets.

Days 11–14 — Hard enforcement + first review. Turn on the save-block at your target stages. Run the first real weekly review (below). Expect grumbling for about a week; expect it to fade as reps realize the logging genuinely takes seconds with the template and that the review is coaching, not a tribunal.

Behavior change across a team of ten or more reps generally takes on the order of four to six weeks to fully stick, with partial habits visible by week two and regression likely if reinforcement stops. Budget for that. The mistake is declaring victory at week two when the pass rate first spikes, dropping the reviews, and watching it decay by week six.

Coaching, Cadence, and Comp — Making It Stick

Tooling gets you to "the field is filled." Cadence and coaching get you to "the field is *accurate*," which is the actual goal. These are the reinforcement layer without which the rules decay.

The weekly quality sample — your most important ritual. Once a week, ideally the morning of your pipeline call, pull "deals updated in the last 7 days" and read five next steps per rep. Rate each 1–3 on specificity: 3 = who/what/with-whom/by-when all present and plausible; 2 = mostly there, one element weak; 1 = vague or fake. Track the *human-rated score per rep over time*, not the regex pass rate. This is the single move that beats Goodhart's Law, because you are inspecting the thing you actually care about (quality) rather than a proxy (completion) that reps can game. Fifteen minutes a week for a ten-rep team.

Surface stale steps in the forecast call — let peer visibility do the work. The cleanest, least authoritarian enforcement is making a bad next step visible to the room. Build a forecast-call view that auto-flags any commit-stage deal whose next step is older than 7 days or fails the specificity bar. When a rep has to explain in front of peers why a $40k "commit" deal has a next step reading "follow up" from 12 days ago, the behavior corrects itself without the manager ever playing cop. The dashboard flags it; social proof enforces it.

Coaching, not punishment, for the first cycle. For the first quarter, treat the specificity score as a *coaching* metric. Pair your best rep on next-step discipline with the laggards for a short weekly huddle. When a next step is weak, the coaching question isn't "why didn't you fill this in" — it's "what did the buyer actually commit to?" That question routes straight back to Cause 1 (call quality), which is usually the real problem.

Comp — small, positive, and only at scale. If and only if you're at a scale where forecast accuracy is a real dollar problem (typically larger teams and higher ARR), you can tie a *small* slice of variable pay to next-step SLA — think a fraction of a percent of OTE, or a "clean pipeline" bonus. The rules of thumb: make it a bonus for hitting the standard, not a penalty for missing it (penalties breed gaming), keep it small (people respond to any comp signal, so you don't need much), and introduce it only after the behavior is already mostly established through coaching. Comp is the last lever, not the first — leading with it before the standard is clear just teaches reps to game an unclear standard.

Where This Breaks — The Honest Bear Case

Most CRM-hygiene playbooks stop at "here's how to do it." Skipping the failure modes is a major reason these programs get abandoned. Four ways this quietly fails, and how to hedge each.

1. The Goodhart trap. The moment "next-step freshness" becomes a tracked KPI, it stops measuring what it used to. Reps optimize the number, not the behavior. The tell: your dashboard shows 90%+ freshness but win rate and forecast accuracy haven't moved after a quarter. Hedge: never track freshness alone. Pair it with the human-rated quality score and audit *that*. Completion is gameable; a manager reading five entries a week is not.

2. The correlation-causation error. It is true that diligent reps both update next steps and close faster. It does not follow that *forcing* a lazy rep to type a next step turns them into a diligent rep or speeds their deals. The next step is often a marker of underlying diligence, not a cause of it. Hedge: do not promise leadership "we'll close X% faster after rollout." Promise better *forecast accuracy* — tighter variance between called and closed numbers — and measure that. Forecast variance will genuinely improve because you can now see which deals are real; close speed may not budge, and promising it sets you up to look like you failed when you didn't.

3. The small-team paradox. On a team of roughly seven reps or fewer, heavy validation plus formal audits can actively *hurt*. The manager already knows every deal personally, so the data-quality gain is small while the administrative drag on both manager and reps is real, and early productivity can dip. Hedge: under ~8 reps, run the review verbally and skip the hard validation — the manager's memory is the system of record. Fully systematize above ~15 reps. The 8–15 zone is genuinely ambiguous; pilot for one quarter and measure whether forecast variance actually tightened before committing.

4. The fiction risk. Enforce specificity on top of bad discovery and you don't get accurate next steps — you get *precise-looking* fake ones, which are more dangerous than honest vagueness because they make a broken pipeline look healthy. Hedge: always fix call quality first. If reps consistently can't produce a real next step, the validation rule is treating the wrong disease.

Honest verdict: this works well at scale, is questionable-to-harmful at small scale, and only delivers real value if you measure human-rated *quality* rather than machine-checked *completion*, and if your calls actually generate real commitments. If your team is small and your forecast variance is already tight, don't build this — coach instead.

A short list of red flags to kill on sight, once you're running:

"Follow up next week" on a real deal is a hope, not a next step. Replace it with a who/what/when, or move the deal to a stalled-deal queue.
A next step older than ~14 days usually means the deal is stuck. Route it to a stalled-deal review rather than letting it rot in "active" pipeline.
The identical next step logged twice in a row is a copy-paste tell. Flag it in the weekly sample.
A next step whose date keeps sliding forward every week ("push to next Friday" x4) is a deal dying in slow motion. Name it.

FAQ

Why do reps log vague next steps like "follow up" instead of specific actions?

Usually because the call itself produced no concrete commitment, so "follow up" is the honest summary of a vague conversation — the CRM entry is a symptom of the call. Secondary causes are friction (logging is slow, so they type the shortest string that saves) and pointlessness (no one ever reads it, so minimum effort is rational). Fix the dominant cause: for vague calls, improve discovery so reps leave with a real mutual commitment; for friction, add templates and let reps log within two hours while memory is fresh; for pointlessness, pull next steps into the weekly forecast call so reps see the field feeds a real decision.

What if reps resist the new validation rules?

Resistance almost always means the change feels like surveillance rather than help. Blunt it by (1) rolling out in "learning mode" first — warnings and reminders for a week or two before any hard save-block — so reps aren't ambushed; (2) writing error messages that teach ("needs an action, a named contact, and a date — e.g. …") instead of scold; and (3) framing the "why" around the rep's own interest: accurate next steps catch slipping deals early, which protects their commission and reduces end-of-quarter fire drills. Start with one team as a pilot, show the reduced surprises, then expand with that team as advocates.

How do I handle deals where the next step is genuinely unknown?

Log the *action that resolves the unknown*, not a placeholder. Instead of "TBD," the rep writes "Confirm budget authority with Priya by Fri 7/24" or "Ask champion who the economic buyer is, this week." A genuine unknown is itself a concrete next action — go find out. If a rep is repeatedly logging unknowns, that's a discovery-quality signal, not a CRM signal: they're leaving calls without the information a qualified deal requires, and the fix is upstream in how they run the conversation.

Can automation and AI tools help enforce accurate next steps?

Yes, as a safety net and a friction reducer, not as a root-cause fix. Validation rules and workflows flag missing or stale entries and remind reps to log within a couple of hours. Conversation-intelligence tools (the category Gong, Chorus and similar occupy) can draft a suggested next step from the call transcript for the rep to confirm, which cuts logging friction dramatically. But automation can't manufacture a real commitment that the call didn't produce, and it can't judge specificity as well as a manager reading five entries a week. Use automation for the mechanical layer; keep a human on quality.

What's the right cadence for reviewing next-step accuracy?

Weekly and short. A 15-minute review the morning of your pipeline call, sampling five next steps per rep in Stage 3+ that are stale (>7 days) or non-specific, rated 1–3 on quality. Track the per-rep quality score over weeks so you can see trends and target coaching. Layer a monthly deeper look for systemic patterns — for example, if next steps consistently rot at one specific stage, that stage's exit criteria are probably unclear. Avoid daily policing (it reads as micromanagement and burns manager time) and avoid quarterly-only reviews (too slow to change behavior before it hardens).

How do I tie next-step accuracy to compensation without breeding resentment?

Make it a small, positive bonus rather than a penalty, and add it last. Reward reps who sustain the standard over a quarter — for instance a modest "clean pipeline" bonus tied to forecast accuracy — rather than docking pay for lapses, because penalties reliably produce gaming (precise-looking fake entries) instead of honesty. Keep the dollar amount small; comp is a signal, and people respond to even a fraction of a percent of OTE. Most importantly, only introduce comp after the behavior is already mostly established through field design and coaching — leading with comp on an unclear standard just teaches reps to game the ambiguity.

Sources

Salesforce, *State of Sales* research report — https://www.salesforce.com/resources/research-reports/state-of-sales/
Salesforce Help, validation rules and required fields documentation — https://help.salesforce.com/s/articleView?id=sf.fields_about_validation_rules.htm
HubSpot Knowledge Base, set required properties and deal stage settings — https://knowledge.hubspot.com/deals/set-up-and-customize-deal-stages
Gong Labs, sales research and revenue-intelligence findings — https://www.gong.io/labs/
Harvard Business Review, sales management and behavior-change coverage — https://hbr.org/topic/subject/sales
Gartner, sales force effectiveness and CRM adoption research — https://www.gartner.com/en/sales

flowchart LR A[Weekly review cycle] --> B[Pull deals updated last 7 days] B --> C[Read 5 next steps per rep] C --> D[Rate each 1 to 3 on specificity] D --> E{Rep average trending up?} E -->|Yes| F[Reinforce, light touch] E -->|No| G[Pair coaching huddle] G --> H{Root cause?} H -->|Vague calls| I[Fix discovery quality] H -->|Slow logging| J[Reduce friction, templates] H -->|Feels pointless| K[Surface in forecast call] F --> A I --> A J --> A K --> A

Related on PULSE

[How do you run a weekly forecast call that's accurate — and not just status theater?](/knowledge/q10883)
[How do we know if Clari forecasting is actually more accurate, or just more confident?](/knowledge/q402)
[How Do I Get My Reps to Log Competitive Intel?](/knowledge/q16068)
[How Do I Get My Showroom Reps to Log Every Up?](/knowledge/q16040)
[How do you log deal intelligence in Salesforce when Foundry is the system of record for account planning?](/knowledge/q10508)

Download:

## Direct Answer

You get reps to log accurate next steps in CRM by making a specific next step **structurally required to advance a deal, small enough to log in under a minute, and reviewed often enough that reps know someone reads it.** Concretely: define what a "next step" must contain — an action, a named human on the buyer side, and a real calendar date — then require those three fields together at your mid-funnel stages, template the inputs so logging takes seconds, review them inside your existing weekly deal-review cadence, and coach on quality rather than policing completion. A reader who stops here has the whole answer: **structure the field so vagueness is impossible, make logging frictionless, and inspect what you expect on a predictable rhythm.** Everything below is the how — the exact validation logic, the 14-day rollout, the coaching moves, and the honest places this fails.

The core mistake teams make is treating this as a discipline problem ("reps are lazy") when it is almost always a **design problem** ("the field lets them be vague, and nobody looks"). A blank text box labeled `Next Step` will fill with "follow up," "circle back," and "TBD" forever, because those are the lowest-effort strings that satisfy the field. Change the design of the field and the review, and behavior follows within a few weeks — no motivational speech required.

One precondition: this only works if your calls actually *produce* a real next step. If reps leave discovery without a concrete mutual commitment from the buyer, no validation rule will save you — you'll just force them to type fiction. Fix call quality first, then turn on enforcement.

```mermaid
flowchart TD
    A[Call or meeting ends] --> B{Did the buyer commit to a concrete next action?}
    B -->|No| C[Fix discovery: get a real commitment before advancing]
    B -->|Yes| D[Log within 2 hours using template]
    D --> E{Action + named person + real date all present?}
    E -->|No| F[Cannot save at Stage 3+]
    E -->|Yes| G[Deal advances]
    G --> H[Reviewed in Monday deal review]
    H --> I{Still fresh and specific?}
    I -->|Yes| J[Forecast trust]
    I -->|No| K[Flag as stalled, coach the rep]
```

## Why Reps Log Vague Next Steps in the First Place

Before you build a single rule, understand the four root causes, because each one has a different fix and stacking the wrong fix on the wrong cause is why so many "CRM hygiene initiatives" quietly die.

**Cause 1: The call produced no real next step.** This is the big one. A rep gets off a call where the buyer said "let me think about it" or "reach out after the holidays," and there is genuinely no concrete action. The rep types "follow up next week" because that is the honest summary of a vague call. The vague CRM entry is a *symptom* of a vague call. If you enforce specificity without fixing call quality, reps invent fake specifics — which is worse than honest vagueness, because now your pipeline looks healthy and isn't.

**Cause 2: Logging is slow and interrupts the rep's flow.** A rep finishes a call, has three more calls stacked, and the CRM makes them click through four screens to update one field. So they batch it for "later," and later never comes, or comes at end of week when memory has decayed. The fix here is pure friction reduction: templates, keyboard shortcuts, mobile logging, or a Gong/Chorus-style tool that drafts the next step from the call transcript for the rep to confirm.

**Cause 3: Nobody reads it, so it feels pointless.** If a rep has logged detailed next steps for six months and no manager has ever referenced one, they correctly conclude it is busywork and downgrade their effort to the minimum that passes. Reps optimize for what gets inspected. The fix is visibility: pull next steps up in the weekly forecast call so the rep sees that the field feeds a real conversation.

**Cause 4: The definition of "good" was never made explicit.** Most teams never told reps what a next step is *supposed* to contain. "Log a next step" is not a spec. "A next step is a specific action, tied to a named buyer-side person, with a date on the calendar" is a spec. You cannot hold people to a standard you never wrote down.

Diagnose which of these dominates your team before you touch Setup. If it's Cause 1, you have a discovery problem wearing a CRM costume, and validation rules will make it worse.

## The Three-Field Standard That Kills Vague Next Steps

The single highest-leverage change is replacing one free-text `Next Step` box with **three linked fields that are useless when vague**:

1. **Next Step Action** — what will happen. Free text, but with a length floor and a banned-phrase guard so "follow up" alone cannot save.
2. **Next Step Owner (buyer-side)** — the actual human on the customer's side the action involves. Not "the team," not the rep themselves. A named person forces the rep to have an actual relationship and an actual thread.
3. **Next Step Date** — a real day on the calendar, and one that must be in the future and typically within ~14 days. "Next week" is not a date; `2026-07-24` is.

The reason the trio works where a single field fails is that vagueness in one field can hide, but vagueness across three linked fields cannot. A rep can fake an action string, but pairing it with a named person and a near-term date makes the fake visible and, more importantly, makes the *real* next step easier to just type than to invent. When all three are present and consistent — "Walk Priya (VP Ops) through the security questionnaire, Thursday 24th, 10am" — you have something a manager can actually inspect and a rep can actually be held to.

A useful mental model: a good next step answers **who does what, with whom, by when.** Strip any one of those and it degrades into a wish. Your field design should make it structurally hard to omit any of them.

Two design cautions. First, do not require this at every stage — early-stage exploratory conversations legitimately don't always have a hard next step, and forcing one there manufactures noise. Require it from your first *qualified* stage onward (commonly Stage 3 / "Solution" / post-discovery). Second, keep the required set to exactly three. Every additional required field is a tax the rep pays on every save, and taxes get evaded. Three is enough to be useful and few enough to be tolerable.

## Enforcement Mechanics: Salesforce and HubSpot

Here is the concrete configuration. Test everything in a sandbox first, because a badly-scoped validation rule can lock reps out of saving legitimately and generate a revolt by lunchtime.

**Salesforce — the validation rule pattern.** Create a validation rule on the Opportunity object that fires only at your target stages. A robust condition looks like this in spirit:

- Fire when `StageName` is at or past your qualified stage, AND
- any of: `NextStep` is blank, `LEN(NextStep) < 15`, the linked buyer contact field is blank, or `NextStepDate` is blank / in the past.

You can add a banned-phrase guard with `REGEX` — matching case-insensitive variants of "follow up," "TBD," "check in," "circle back," "touch base." Keep the error message *actionable*: not "Invalid next step" but "Next steps at this stage need an action, a named buyer contact, and a future date within two weeks — e.g. 'Send Priya the ROI model, Thu 7/24.'" The error message is a teaching surface; use it.

Salesforce's own documentation covers the mechanics of validation rules and required fields; follow their guidance on formula limits and testing so you don't hit governor issues on high-volume orgs.

**HubSpot — the equivalent.** HubSpot doesn't have Salesforce-style validation rules across all tiers, so you assemble the same outcome from: making the next-step properties **required to move to a stage** (deal stage property requirements), a **workflow** that flags or re-assigns deals where the next step is stale or blank, and **conditional property logic** so the fields surface at the right stage. HubSpot's knowledge base documents required properties per deal stage and workflow triggers — build the "blank or stale next step" trigger to notify the rep and, on repeat, the manager.

**The arms-race warning.** Reps learn what your rule checks. Set a 15-character floor and you'll get "follow up with him later this week" (35 characters, still useless). Ban "follow up" and you'll get "reach out." **You cannot win the regex war**, and pouring engineering hours into ever-cleverer banned-phrase lists is a losing game. The regex is a speed bump that catches the laziest 60–70% of garbage; the remaining garbage is caught by *human review of quality*, not by more regex. Set a reasonable floor, ban the four or five worst offenders, and stop — then invest the saved energy in the weekly sample review below.

## The 14-Day Rollout Plan

Do not flip validation rules on for the whole org on a Monday morning. Stage it so reps experience the change as help, not ambush.

**Days 1–3 — Audit and baseline.** Pull every open opportunity above a meaningful threshold (say $10k, or whatever is material for you). For each, mark whether it has a next step written in the last 7 days, a specific future date, and a named buyer-side person. Compute the percentage that pass all three. This is your baseline — you'll need it to prove the initiative worked, and you'll need it to calibrate expectations with leadership. In most unmanaged pipelines a large share of "active" deals fail at least one of the three checks; seeing the real number is usually the moment leadership stops arguing about whether this is needed.

**Days 4–6 — Define and socialize the standard.** Write the one-paragraph definition of a good next step (who / what / with whom / by when) and share it with three real before-and-after examples from *your own* pipeline. Reps accept standards illustrated with their own deals far more than abstract policy. Announce that validation goes live in one week, and that the first two weeks are "learning mode" — flags and reminders, no hard blocks.

**Days 7–10 — Soft launch (warnings only).** Turn on the fields and the reminder workflow, but not the hard save-block. Reps get nudged when a next step is missing or stale, and see the new fields, but nothing stops them saving. Watch the pass-rate climb and note which reps and which stages lag. This is where you catch legitimate edge cases (a stage where a hard next step genuinely doesn't apply) before they become support tickets.

**Days 11–14 — Hard enforcement + first review.** Turn on the save-block at your target stages. Run the first real weekly review (below). Expect grumbling for about a week; expect it to fade as reps realize the logging genuinely takes seconds with the template and that the review is coaching, not a tribunal.

Behavior change across a team of ten or more reps generally takes on the order of four to six weeks to fully stick, with partial habits visible by week two and regression likely if reinforcement stops. Budget for that. The mistake is declaring victory at week two when the pass rate first spikes, dropping the reviews, and watching it decay by week six.

## Coaching, Cadence, and Comp — Making It Stick

Tooling gets you to "the field is filled." Cadence and coaching get you to "the field is *accurate*," which is the actual goal. These are the reinforcement layer without which the rules decay.

**The weekly quality sample — your most important ritual.** Once a week, ideally the morning of your pipeline call, pull "deals updated in the last 7 days" and read **five next steps per rep**. Rate each 1–3 on specificity: 3 = who/what/with-whom/by-when all present and plausible; 2 = mostly there, one element weak; 1 = vague or fake. Track the *human-rated score per rep over time*, not the regex pass rate. This is the single move that beats Goodhart's Law, because you are inspecting the thing you actually care about (quality) rather than a proxy (completion) that reps can game. Fifteen minutes a week for a ten-rep team.

**Surface stale steps in the forecast call — let peer visibility do the work.** The cleanest, least authoritarian enforcement is making a bad next step visible to the room. Build a forecast-call view that auto-flags any commit-stage deal whose next step is older than 7 days or fails the specificity bar. When a rep has to explain in front of peers why a $40k "commit" deal has a next step reading "follow up" from 12 days ago, the behavior corrects itself without the manager ever playing cop. The dashboard flags it; social proof enforces it.

**Coaching, not punishment, for the first cycle.** For the first quarter, treat the specificity score as a *coaching* metric. Pair your best rep on next-step discipline with the laggards for a short weekly huddle. When a next step is weak, the coaching question isn't "why didn't you fill this in" — it's "what did the buyer actually commit to?" That question routes straight back to Cause 1 (call quality), which is usually the real problem.

**Comp — small, positive, and only at scale.** If and only if you're at a scale where forecast accuracy is a real dollar problem (typically larger teams and higher ARR), you can tie a *small* slice of variable pay to next-step SLA — think a fraction of a percent of OTE, or a "clean pipeline" bonus. The rules of thumb: make it a **bonus for hitting the standard, not a penalty for missing it** (penalties breed gaming), keep it small (people respond to any comp signal, so you don't need much), and introduce it only after the behavior is already mostly established through coaching. Comp is the last lever, not the first — leading with it before the standard is clear just teaches reps to game an unclear standard.

```mermaid
flowchart LR
    A[Weekly review cycle] --> B[Pull deals updated last 7 days]
    B --> C[Read 5 next steps per rep]
    C --> D[Rate each 1 to 3 on specificity]
    D --> E{Rep average trending up?}
    E -->|Yes| F[Reinforce, light touch]
    E -->|No| G[Pair coaching huddle]
    G --> H{Root cause?}
    H -->|Vague calls| I[Fix discovery quality]
    H -->|Slow logging| J[Reduce friction, templates]
    H -->|Feels pointless| K[Surface in forecast call]
    F --> A
    I --> A
    J --> A
    K --> A
```

## Where This Breaks — The Honest Bear Case

Most CRM-hygiene playbooks stop at "here's how to do it." Skipping the failure modes is a major reason these programs get abandoned. Four ways this quietly fails, and how to hedge each.

**1. The Goodhart trap.** The moment "next-step freshness" becomes a tracked KPI, it stops measuring what it used to. Reps optimize the number, not the behavior. The tell: your dashboard shows 90%+ freshness but win rate and forecast accuracy haven't moved after a quarter. **Hedge:** never track freshness alone. Pair it with the human-rated quality score and audit *that*. Completion is gameable; a manager reading five entries a week is not.

**2. The correlation-causation error.** It is true that diligent reps both update next steps and close faster. It does not follow that *forcing* a lazy rep to type a next step turns them into a diligent rep or speeds their deals. The next step is often a marker of underlying diligence, not a cause of it. **Hedge:** do not promise leadership "we'll close X% faster after rollout." Promise better *forecast accuracy* — tighter variance between called and closed numbers — and measure that. Forecast variance will genuinely improve because you can now see which deals are real; close speed may not budge, and promising it sets you up to look like you failed when you didn't.

**3. The small-team paradox.** On a team of roughly seven reps or fewer, heavy validation plus formal audits can actively *hurt*. The manager already knows every deal personally, so the data-quality gain is small while the administrative drag on both manager and reps is real, and early productivity can dip. **Hedge:** under ~8 reps, run the review verbally and skip the hard validation — the manager's memory is the system of record. Fully systematize above ~15 reps. The 8–15 zone is genuinely ambiguous; pilot for one quarter and measure whether forecast variance actually tightened before committing.

**4. The fiction risk.** Enforce specificity on top of bad discovery and you don't get accurate next steps — you get *precise-looking* fake ones, which are more dangerous than honest vagueness because they make a broken pipeline look healthy. **Hedge:** always fix call quality first. If reps consistently can't produce a real next step, the validation rule is treating the wrong disease.

**Honest verdict:** this works well at scale, is questionable-to-harmful at small scale, and only delivers real value if you measure human-rated *quality* rather than machine-checked *completion*, and if your calls actually generate real commitments. If your team is small and your forecast variance is already tight, don't build this — coach instead.

**A short list of red flags to kill on sight, once you're running:**
- "Follow up next week" on a real deal is a hope, not a next step. Replace it with a who/what/when, or move the deal to a stalled-deal queue.
- A next step older than ~14 days usually means the deal is stuck. Route it to a stalled-deal review rather than letting it rot in "active" pipeline.
- The identical next step logged twice in a row is a copy-paste tell. Flag it in the weekly sample.
- A next step whose date keeps sliding forward every week ("push to next Friday" x4) is a deal dying in slow motion. Name it.

## FAQ

### Why do reps log vague next steps like "follow up" instead of specific actions?

Usually because the call itself produced no concrete commitment, so "follow up" is the honest summary of a vague conversation — the CRM entry is a symptom of the call. Secondary causes are friction (logging is slow, so they type the shortest string that saves) and pointlessness (no one ever reads it, so minimum effort is rational). Fix the dominant cause: for vague calls, improve discovery so reps leave with a real mutual commitment; for friction, add templates and let reps log within two hours while memory is fresh; for pointlessness, pull next steps into the weekly forecast call so reps see the field feeds a real decision.

### What if reps resist the new validation rules?

Resistance almost always means the change feels like surveillance rather than help. Blunt it by (1) rolling out in "learning mode" first — warnings and reminders for a week or two before any hard save-block — so reps aren't ambushed; (2) writing error messages that teach ("needs an action, a named contact, and a date — e.g. …") instead of scold; and (3) framing the "why" around the rep's own interest: accurate next steps catch slipping deals early, which protects their commission and reduces end-of-quarter fire drills. Start with one team as a pilot, show the reduced surprises, then expand with that team as advocates.

### How do I handle deals where the next step is genuinely unknown?

Log the *action that resolves the unknown*, not a placeholder. Instead of "TBD," the rep writes "Confirm budget authority with Priya by Fri 7/24" or "Ask champion who the economic buyer is, this week." A genuine unknown is itself a concrete next action — go find out. If a rep is repeatedly logging unknowns, that's a discovery-quality signal, not a CRM signal: they're leaving calls without the information a qualified deal requires, and the fix is upstream in how they run the conversation.

### Can automation and AI tools help enforce accurate next steps?

Yes, as a safety net and a friction reducer, not as a root-cause fix. Validation rules and workflows flag missing or stale entries and remind reps to log within a couple of hours. Conversation-intelligence tools (the category Gong, Chorus and similar occupy) can draft a suggested next step from the call transcript for the rep to confirm, which cuts logging friction dramatically. But automation can't manufacture a real commitment that the call didn't produce, and it can't judge specificity as well as a manager reading five entries a week. Use automation for the mechanical layer; keep a human on quality.

### What's the right cadence for reviewing next-step accuracy?

Weekly and short. A 15-minute review the morning of your pipeline call, sampling five next steps per rep in Stage 3+ that are stale (>7 days) or non-specific, rated 1–3 on quality. Track the per-rep quality score over weeks so you can see trends and target coaching. Layer a monthly deeper look for systemic patterns — for example, if next steps consistently rot at one specific stage, that stage's exit criteria are probably unclear. Avoid daily policing (it reads as micromanagement and burns manager time) and avoid quarterly-only reviews (too slow to change behavior before it hardens).

### How do I tie next-step accuracy to compensation without breeding resentment?

Make it a small, positive bonus rather than a penalty, and add it last. Reward reps who sustain the standard over a quarter — for instance a modest "clean pipeline" bonus tied to forecast accuracy — rather than docking pay for lapses, because penalties reliably produce gaming (precise-looking fake entries) instead of honesty. Keep the dollar amount small; comp is a signal, and people respond to even a fraction of a percent of OTE. Most importantly, only introduce comp after the behavior is already mostly established through field design and coaching — leading with comp on an unclear standard just teaches reps to game the ambiguity.

## Sources

- Salesforce, *State of Sales* research report — https://www.salesforce.com/resources/research-reports/state-of-sales/
- Salesforce Help, validation rules and required fields documentation — https://help.salesforce.com/s/articleView?id=sf.fields_about_validation_rules.htm
- HubSpot Knowledge Base, set required properties and deal stage settings — https://knowledge.hubspot.com/deals/set-up-and-customize-deal-stages
- Gong Labs, sales research and revenue-intelligence findings — https://www.gong.io/labs/
- Harvard Business Review, sales management and behavior-change coverage — https://hbr.org/topic/subject/sales
- Gartner, sales force effectiveness and CRM adoption research — https://www.gartner.com/en/sales

## Related on PULSE

- [How do you run a weekly forecast call that's accurate — and not just status theater?](/knowledge/q10883)
- [How do we know if Clari forecasting is actually more accurate, or just more confident?](/knowledge/q402)
- [How Do I Get My Reps to Log Competitive Intel?](/knowledge/q16068)
- [How Do I Get My Showroom Reps to Log Every Up?](/knowledge/q16040)
- [How do you log deal intelligence in Salesforce when Foundry is the system of record for account planning?](/knowledge/q10508)

Was this helpful?

Sources cited

clari.comhttps://www.clari.com/gartner.comhttps://www.gartner.com/en/documents/sales-forecasting joinpavilion.comhttps://www.joinpavilion.com/compensation-report bridgegroupinc.comhttps://www.bridgegroupinc.com/blog/sales-development-report gong.iohttps://www.gong.io/forcemanagement.comhttps://forcemanagement.com/

⌬ Apply this in PULSE

Free CRM · Revenue IntelligenceAudit pipeline, score reps, ship the fix

Kory White