Feature · Sales Forecasting Explained

Sales forecasting, explained the way a VP would actually run it.

A sales forecast is a commitment: a number the revenue team will book in a period. The explanation that actually helps a new VP is not a textbook definition; it is the four forecasting methods real teams run, the math behind each one, where each one breaks, and what software replaces the Monday spreadsheet ritual. This page walks the whole picture, then shows what Strkr forecasting ships for teams who have outgrown the spreadsheet but have not yet built a data science function.

Start free trial See pricing
What a sales forecast actually is

The commitment, not the prediction.

The textbook definition of a sales forecast is "a projection of expected sales for a future period." The textbook definition is also incomplete in a way that trips up new revenue leaders. A sales forecast in practice is not a prediction; it is a commitment. The number the VP of Sales submits for Q3 is the number the board will measure them against. That framing changes everything downstream: which deals get called, which signals matter, who approves the number, and what counts as a forecast miss. The six items below are what a working forecast actually comprises, which no textbook definition captures cleanly.

The number

A dollar commit for a period.

A sales forecast is not a range; it is a point estimate for a period (month, quarter, year). The VP commits $14M for Q3; the board measures against $14M. Ranges are useful internally for scenario planning, but the forecast that reaches the board is one number.

The method

How the number was arrived at.

Weighted pipeline, commit categories, historical run-rate, or a mix. Each method has different mechanics, different inputs, and different failure modes. The method also determines who can argue with the number, which matters for how the forecast call runs. See the four methods in the next section.

The commit tiers

Best case, forecast, commit, worst case.

Mature forecasts carry three to four tiers. Worst case is the floor; commit is the number the rep stakes their job on; forecast is the realistic read; best case is the stretch. The VP forecast is usually the sum of commit plus a portion of best case, calibrated against historical accuracy.

The accuracy history

How close last period's number landed.

A forecast with no history is a wish. A forecast with accuracy history is a calibrated instrument. Teams track forecast accuracy per rep, per team, per segment, with the pattern of over-call versus under-call visible. The accuracy history is what makes next quarter's number credible.

The submission ritual

Reps, managers, VP, locked.

A forecast moves up the hierarchy weekly: reps submit by Thursday, managers roll up by Friday, VP commits by Monday. The ritual is as important as the math. Teams who skip the ritual get a VP forecast disconnected from the reps who have to hit it; teams who skip the math get a ritual that produces unreliable numbers.

The adjustment

VP judgment layered on the methodology.

The final forecast is the methodology output plus VP judgment. The VP sees the weighted-pipeline number, the commit roll-up, and the historical run-rate, then adjusts for signals the system does not see (a reference check that is going slowly, a competitor move, a exec call going well). The judgment is explicit, logged, and auditable.

The four methods revenue teams actually use

Weighted pipeline, commit categories, run-rate, and AI.

Any sales forecasting explainer that lists ten methods is a textbook; real revenue teams run one of four methods, or a blend. The four below are the methods a VP will encounter in a mid-market B2B SaaS career. The right method depends on cycle length, deal size, team size, and data maturity. Teams usually start with weighted pipeline because it is the simplest to compute, graduate to commit categories as the team grows, add historical run-rate for sanity-checking, and add AI-assisted signals once the data volume supports it.

Weighted pipeline

Deal amount times stage probability.

Every deal in the pipeline carries a stage, and every stage carries a historical win probability (Qualified = 10%, Discovery = 20%, Proposal = 50%, Negotiation = 80%). Multiply each deal's amount by its stage probability, sum across all open deals, and the result is the weighted pipeline. Simple to compute; the main failure mode is stage inflation.

Commit categories

Rep calls each deal: commit, best case, pipeline.

Every open deal gets a category: Commit (confident this quarter), Best Case (possible stretch), Pipeline (next quarter or beyond). The rep calls the category weekly. VP forecast is Sum of Commit plus some portion of Best Case based on historical realization. Better than weighted pipeline for mature teams; requires rep discipline.

Historical run-rate

Last four quarters average, adjusted for seasonality.

Trailing four quarters of bookings, averaged, adjusted for seasonality (Q4 up for most B2B, Q1 down post-holiday). The sanity-check method. If weighted pipeline says $18M and run-rate says $12M, something in the pipeline math is wrong. Teams use this to triangulate, rarely as the primary method.

AI-assisted

Model on deal signals: engagement, stage velocity, size.

An AI model trained on historical closed-won versus closed-lost deals scores each open deal on close probability, factoring engagement signals (email opens, meetings scheduled, decision-maker involvement), stage velocity, deal size, and seasonality. Strkr AI assist weighs these signals; the forecast becomes the sum of signal-scored deal probabilities.

Blended method

Commit roll-up plus AI sanity-check plus VP judgment.

Most mature teams run a blend. Reps submit commit categories (the primary method), the AI model scores the same deals (the sanity check), run-rate renders in a sidebar (the baseline), and the VP applies judgment to produce the final number. The blended approach is what most Strkr customers settle on.

When to switch methods

Scale, cycle length, data maturity.

Teams under 20 reps run weighted pipeline cleanly. 20 to 100 reps need commit categories to prevent stage inflation at scale. Above 100 reps, AI assist becomes a credible sanity check because the data volume supports a model. The right method evolves as the team grows; the forecast ritual stays the same.

Where forecasts actually break

The six failure modes every VP eventually catalogs.

The forecast methodology handles the math; the failure modes are operational. The six items below are the reasons a forecast misses that have nothing to do with the formulas. Any new VP will run into every one of them in the first few quarters, and the mature teams have procedures for each. Strkr builds the handling for most of these into the forecast module directly, which is the operational value proposition versus a spreadsheet.

Stage inflation

Deals sit in Proposal that should be in Discovery.

Reps mark deals as further along than they are, which inflates the weighted pipeline number. The forecast says $20M, the reality is $14M, the miss shows up at quarter-end. Strkr flags deals where the stage does not match the activity signals (no decision-maker meeting, no proposal document attached) and surfaces the inflation before the forecast call.

Happy ears

The rep thinks the deal will close; it will not.

Reps are optimistic by nature; some deals in Commit should have been in Best Case. Teams track per-rep commit accuracy; reps whose commit historically closes at 60 percent get adjusted downward. Strkr renders the per-rep calibration automatically.

Stuck deals

Deals aging in a stage without activity.

A deal in Negotiation for 45 days with no activity in the last 14 days is probably dead. The forecast says it will close; the quarter says it did not. Strkr auto-flags stuck deals; the forecast surface renders a prompt to re-category or exclude.

Pull-forward exhaustion

Last quarter's pull-ins drain this quarter's pipeline.

A VP pulls forward every late-stage deal to close out a tough quarter. The next quarter starts with less pipeline than expected and the forecast miss compounds. Strkr tracks pull-forward volume per quarter and renders the drain on next quarter's coverage.

Submission skew

Managers submit higher than reps rolled up.

Managers under pressure submit a bigger number than the rep roll-up supports. VP calls get surprised at quarter-end. Strkr compares each level's submission to the next-level roll-up and surfaces the delta on the manager's dashboard.

External shocks

Deals the system cannot predict.

Customer bankruptcy, macro shift, competitor acquisition. Signals the forecast cannot anticipate. The response is to leave room for shocks (worst-case scenario math) and to adjust the forecast promptly when a shock lands. Strkr lets the VP log a judgment adjustment with reason; the adjustment is auditable.

What forecasting software ships

The nine components a working forecast module needs.

A forecasting module is not one feature; it is nine components that have to compose. The reporting surface is the visible piece; the heavy lifting is in the submission ritual, the accuracy tracking, the commit category discipline, and the AI assist. The nine items below are what Strkr ships, and are the right checklist against any forecasting software evaluation. Spreadsheet-based forecasting typically covers two or three of these; dedicated forecasting software should cover all nine.

Weighted pipeline view

Stage probabilities calibrated per segment.

Weighted pipeline rendered with per-segment stage probabilities (Enterprise Qualified = 15%, SMB Qualified = 8%, because the segments convert differently). Breakdown by segment, by team, by period, with the deal list behind every number.

Commit category submission

Weekly ritual, locked by Thursday.

Reps categorize each deal (Commit, Best Case, Pipeline) by Thursday 5 PM. Managers roll up Friday. The submission is a form, not a spreadsheet; the deal list is pre-filtered to open deals in the current period; the rep clicks the category chip for each deal.

Forecast accuracy

Per rep, per team, per manager, trailing 4 quarters.

Accuracy chart renders per level: this rep's commit historically closes at 73%, this manager's roll-up historically lands within 8% of actual. The chart is the single most important sanity-check against submitted numbers and the input to VP judgment at the top of the chain.

Historical run-rate

Trailing 4 quarters, seasonally adjusted.

Trailing four quarters of bookings with seasonal adjustment, rendered as a sidebar on the forecast surface. When the submitted number diverges from run-rate, the delta is visible immediately. Not the primary input, but the sanity check that catches aggressive submissions.

AI assist

Signal-scored deal probabilities.

The AI model scores each open deal on close probability using engagement signals, stage velocity, deal size, and historical patterns. The score renders next to the rep's category; disagreements between rep category and model score flag for review. Not the final answer; the sanity check.

Stuck-deal detection

Flag deals aging without activity.

Deals in a stage for N days without activity (configurable per stage) flag on the forecast surface. The rep re-categorizes, excludes, or re-engages. The flag runs before the forecast submission so the submission reflects the current state.

Pull-forward tracking

What came early, what pushed out.

The forecast module tracks deals that pulled in from a future quarter and deals that pushed out from the current quarter. The pull-forward volume feeds the next-quarter coverage math; the push-out volume feeds the stuck-deal queue. The pattern that drains next quarter is visible before next quarter starts.

Submission lock

Window closes, numbers snapshot.

The weekly submission window closes with a snapshot. Numbers are preserved for accuracy tracking; downstream calculations run against the locked state; late edits are logged separately. The ritual has a defined moment of truth, which is the input to the forecast accuracy chart.

VP judgment layer

Explicit adjustment with reason, logged.

The VP applies judgment to the methodology output: adjust up for a reference check going well, down for a competitive signal landing against. The adjustment is explicit, logged with the reason, and visible to the CRO. No shadow forecast lives in a VP spreadsheet.

Software versus spreadsheet

When the ritual outgrows the spreadsheet.

Spreadsheet works at 5 reps

Everyone fits in one tab.

A five-rep team runs a forecast in a spreadsheet without pain. The data volume is small enough to review manually; the submission ritual is a Slack message; the accuracy tracking is in the VP's head. Nothing broken; no software needed.

Spreadsheet breaks at 20 reps

Three tabs disagree, nobody knows which is current.

At 20 reps across three teams, the forecast spreadsheet has three tabs, each maintained by a different manager, each with slightly different formulas. The VP review is 30 minutes of reconciling tabs before the actual conversation starts. The ritual breaks.

Software earns the slot at 20 to 50 reps

The ritual needs a system of record.

Between 20 and 50 reps, the forecast ritual needs a system of record. Submissions should feed a central roll-up, accuracy tracking should run on real data, stuck-deal flagging should fire automatically. The software is table stakes; the choice depends on cycle length and data maturity.

AI assist meaningful at 50+ reps

Data volume supports a signal model.

Above 50 reps and roughly 500 open deals in a cycle, there is enough data for a model to add value beyond rep judgment. Deal-signal scoring becomes a real sanity check on commit categories. Below that scale, the AI assist is directional but not statistically meaningful.

Enterprise forecasting

Multi-geo, multi-product, hierarchical commits.

A 500-rep enterprise motion with multi-geo and multi-product forecasting needs hierarchical commits (region to global, product to total), currency normalization, and cross-functional roll-ups (customer success renewals, services bookings). Software is required; the spreadsheet is operationally untenable.

Strkr positions mid-market

20 to 500 reps, built for the ritual.

Strkr forecasting is designed for the 20-to-500-rep mid-market B2B motion where the ritual outgrew the spreadsheet but a Clari or Gong Forecast is overkill. Preconfigured commit categories, AI assist on the base tier, accuracy tracking, judgment layer with audit, submission lock. One module, one invoice line.

From Monday spreadsheet to a real forecast ritual.

Weighted pipeline, commit categories, AI assist, accuracy tracking, stuck-deal detection, submission lock. The forecast module that replaces the spreadsheet, included at the base seat price.

Start free trial See forecasting features
Common questions

What buyers ask about this feature.

What is a sales forecast in simple terms?

A sales forecast is a dollar commitment for a future period. A VP of Sales submits a forecast of $14M for Q3, and the board measures the team against that number. The forecast is not a prediction; it is a commitment, which changes how it is produced (methodology plus judgment), who reviews it (reps roll up to managers roll up to VP), and what counts as a miss (actual bookings versus the committed number). The number is usually broken into tiers (commit, best case, pipeline) with the VP forecast landing between commit and best case based on historical accuracy.

Which forecasting method is the right one to start with?

For a team under 20 reps, weighted pipeline (deal amount times stage probability, summed across open deals) is the simplest working method. It is easy to compute, easy to explain, and surfaces pipeline-coverage math directly. The main failure mode is stage inflation: deals get moved to later stages than they should be, which inflates the number. Teams under 20 reps can review the stages manually; above 20 reps the method breaks and commit categories become the primary approach, with weighted pipeline retained as a sanity check.

How is AI-assisted forecasting different from weighted pipeline?

Weighted pipeline uses a single stage probability per deal. AI-assisted forecasting scores each deal on close probability using many signals (email engagement, meeting cadence, decision-maker involvement, stage velocity, deal size, product mix, seasonality) and the model learns from historical closed-won versus closed-lost outcomes. The output is a per-deal probability calibrated to that team's actual close rate, not an industry-average stage probability. AI assist becomes statistically meaningful at roughly 50 reps and 500 open deals per cycle; below that scale, it is directional.

What is forecast accuracy and how is it measured?

Forecast accuracy is the ratio of actual bookings to the committed forecast at the start of the period. A team that commits $10M and books $9.4M has 94% forecast accuracy for the period. The metric is tracked per rep, per team, per manager, trailing four to eight quarters, with the pattern of over-call versus under-call visible. Teams above 90% accuracy are operationally reliable; below 75% is a signal that either the methodology or the discipline is broken. The accuracy history is the primary input to VP judgment at the top of the submission chain.

When does a team outgrow spreadsheet forecasting?

The spreadsheet works at five reps. It breaks at twenty, when multiple managers each maintain a tab with different formulas and the VP review becomes 30 minutes of reconciling tabs before the actual conversation starts. Between 20 and 50 reps, software becomes table stakes; above 50 reps and 500 open deals per cycle, AI assist becomes statistically meaningful; above 500 reps with multi-geo and multi-product scope, dedicated enterprise forecasting is required. Strkr is designed for the 20-to-500-rep range where the ritual has outgrown the spreadsheet but a Clari-sized tool is overkill.

Does Strkr replace Clari, Gong Forecast, and InsightSquared?

For the 20-to-500-rep mid-market B2B motion, yes. The nine components of a working forecast module (weighted pipeline, commit submission, accuracy tracking, run-rate baseline, AI assist, stuck-deal detection, pull-forward tracking, submission lock, VP judgment layer) all ship on the Strkr base tier. For enterprise motions above 500 reps with multi-geo, multi-currency, and cross-functional roll-ups (renewal, services, hardware), Clari and InsightSquared have maturity in those edge cases that Strkr is still building; the two coexist and Strkr forecast tiles are available via REST for the enterprise analytics stack.

Related features

Pair this with...

AI Sales Forecasting Sales Analytics KPI Dashboard Sales Dashboard

Try it free. Bring your team next week.

No sales call, no migration consultant, no four-month implementation. Enter your card, get 14 days of the full Pro tier, cancel any time before day 14 with zero charge. Spin up a workspace, import your CSV, and have something useful before lunch.

Start free trial See pricing