How to forecast B2B sales accurately (and the categories that matter)

How do you forecast B2B sales accurately?

Short answer: by separating deals into three categories — Commit, Best Case, Pipeline — based on discovery completeness, not gut feel. By running a weekly forecast call where reps defend their categorisations against exit criteria. And by measuring forecast accuracy as a KPI separate from quota attainment, so reps cannot game one by manipulating the other.

Accurate forecasting is one of the single hardest skills in B2B sales leadership, and one of the most under-invested. Most companies treat the forecast as a vibes summary rolled up from the CRM. The teams that forecast within 10% of actual revenue every quarter run it as an operational discipline.

TL;DR — the forecast structure

A clean forecast number = Closed Won + Commit. Best Case is reported separately as upside.

Why most forecasts are wrong

The three failure modes:

Failure mode 1: optimism bias. Reps want to look strong in the forecast call. They categorise hopeful deals as Commit and avoid downgrading deals that have stalled. Result: forecast is consistently 15–30% over actual.

Failure mode 2: sandbagging. The reverse — reps under-forecast to make sure they overshoot the number. Looks rosy on the surface ("we beat the forecast!") but produces unreliable planning data for the business.

Failure mode 3: deal-stage proxy. Many teams forecast by deal stage (e.g., "all deals in Proposal stage = 70% commit"). This is mechanically simple but ignores discovery quality. A deal can be in Proposal with no Economic Buyer engaged and no Decision Process documented — it is not 70% likely to close. It is 20%.

The fix for all three is the same: a structured, defended forecast call where each Commit and Best Case deal must satisfy explicit exit criteria.

Forecast accuracy benchmarks

Forecast accuracy is measured as: |Forecast - Actual| ÷ Actual, expressed as a percentage. Lower is better.

Most B2B companies under £10M ARR operate in the 15–25% accuracy band. Companies that install structured forecasting move into the 8–15% band within a quarter and the 5–8% band within 2–3 quarters.

The exit criteria for each category

The cleanest forecasting system uses MEDDPICC (or SPICED) field completeness as the exit criteria for category placement.

Commit (90–100% probability)

A deal in Commit must have:

• Verbal yes from the Economic Buyer (not just the champion).
• All MEDDPICC fields complete and scored at 7+ / 8.
• A mutual close plan documented — a written timeline both sides have agreed.
• Commercial agreement on price, even if the paper is not yet signed.
• No identified blockers that could realistically extend the timeline past forecast period.

If any one of these is missing, the deal is not Commit. It is Best Case.

Best Case (50–70% probability)

A deal in Best Case must have:

• Identified Economic Buyer (rep has met them or has a clear path).
• MEDDPICC score 5+ / 8.
• Champion in place (not just a friendly contact).
• Decision Process documented with realistic dates.
• One or two outstanding risks that are tractable.

Below this threshold, the deal is Pipeline.

Pipeline (not forecast-eligible)

Open opportunities that do not yet meet Best Case criteria. They are real deals, but they are not predictable enough to commit to in the current period.

Reps often resist this — every deal feels real to the rep working it. The discipline is that "real" and "forecastable" are different things.

The weekly forecast call

The forecast is built and defended in a single recurring meeting. The structure:

The meeting is short (60 minutes for a team of 4–6 reps), repeats weekly, and produces one output: a number the sales leader commits to leadership.

The rep's job: defend each Commit deal in 60–90 seconds against exit criteria. "This deal is Commit because [Economic Buyer], [MEDDPICC score], [mutual close plan], [no blockers]."

The manager's job: challenge any unclear answer. If the rep cannot articulate the Economic Buyer's name, the deal is not Commit. Downgrade in real time.

The math behind a forecast number

For a £400K quarterly target:

Forecast number reported to leadership: £130K + £220K = £350K Commit, with £108K upside potential to £458K Best Case.

Notice what this is not: it is not a probability-weighted sum of every deal in the pipeline. That math (often called "weighted forecast") is mathematically convenient and operationally useless — it produces a single number that is wrong by definition (no deal will close at 60% — it will close at 100% or 0%).

The Commit / Best Case / Pipeline structure produces a forecast number the sales leader can stand behind, plus a clearly demarcated upside.

Forecast accuracy as a KPI

Forecast accuracy should be tracked per rep, weekly, with a rolling average over the last 8–12 weeks.

Tracking forecast accuracy as a separate KPI from quota attainment prevents two common dysfunctions:

• A rep can hit quota while being a terrible forecaster (lucky upside on a few deals). Tracking only quota misses this.
• A rep can be an excellent forecaster while missing quota (the system is the problem, not the rep). Tracking only quota also misses this.

Both pieces of information are needed.

What kills forecast accuracy fast

Deal pull-ins. A rep pulls a Best Case deal into Commit to make the number look stronger this week. The deal slips, and the forecast misses. Pull-ins should require manager sign-off and explicit justification — not be a rep-only decision.

Stale deals never being demoted. A deal sits in Best Case for 4 months. The rep keeps believing in it. The exit criteria say it should have been moved to Pipeline or Closed Lost 6 weeks ago. Weekly review must include explicit demotions, not just promotions.

Late-quarter heroics. Reps grind to close a deal in the last week of the quarter that was never realistically in this quarter's forecast. Hits the number, looks good — and trains everyone that the forecast is decorative. The fix: hold the forecast number as set at the start of the last week. Late closes are next-quarter pull-ins.

Founder-driven deals not in the rep forecast. The founder is working two large deals personally that nobody else is forecasting. They close. The forecast misses by 30%. Founder-driven deals should be in the forecast under a "Founder" rep allocation, not invisible.

Mid-quarter strategy changes. Leadership decides mid-quarter to chase a new segment or change pricing. The deals already in the pipeline are no longer representative. The forecast becomes unreliable. Strategy changes should be timed for quarter starts when possible.

The forecast call versus the pipeline call

These are often confused. They are different meetings.

A well-run sales team runs both, on different days. The Pipeline call works the long-term hygiene; the Forecast call sets the short-term commitment.

Companies that try to combine them produce muddy outcomes — either the forecast slips because the meeting got bogged down in early-stage deal hygiene, or pipeline rot accelerates because the meeting got consumed by Commit defence.

Forecasting for new segments and new products

A standard caveat: forecasting works when you have historical data. For new segments or new products without baseline data, the forecast is closer to a guess.

Tactics for forecasting new segments:

1. Use bottom-up activity math instead of opportunity-based forecast. If you know your meeting-to-opportunity and opportunity-to-close rates from a similar segment, project from activity volume.
2. Set Commit deliberately low for the first 2–3 quarters. Better to under-promise and learn than to commit blindly.
3. Track lead indicators more carefully — pipeline coverage ratio, sales cycle length, win rate — and adjust the model as data arrives.
4. Run the forecast call at the same cadence regardless. Discipline matters even when the data is thin.

By quarter 4 of a new segment launch, you should have enough data to forecast it with normal accuracy.

Pipeline coverage as the leading indicator

A forecast is a lagging signal. The leading signal is pipeline coverage ratio — the ratio of open pipeline to remaining quota target.

If forecast accuracy is the leadership metric for "are we hitting the number?", coverage ratio is the leadership metric for "are we going to in 90 days?"

The two together — current quarter forecast + next quarter coverage — give the sales leader (and the CFO, and the board) a clear forward view.

(Covered in more depth in our post on pipeline coverage ratio.)

Tooling — what actually helps

The serious B2B forecasting tools:

For companies under £5M ARR, the built-in forecast in HubSpot or Salesforce is enough. For mid-market and enterprise, Clari or Boostup add genuine value — predictive flags ("this deal is more likely to slip than the rep thinks") that catch errors humans miss.

The trap: buying a £40K/year forecast tool to fix what is actually a forecast discipline problem. The tool does not make the rep more honest. The weekly call does.

For UAE & KSA teams

Forecasting in the GCC has structural deviations from Western benchmarks.

Build seasonality into the forecast model explicitly. Q2 (April–June, encompassing Ramadan + Eid + early Hajj) reliably underperforms Q1 and Q4 in MENA sales. A flat 1/12-per-month forecast model will miss Q2 by 30–50% every year. Bake the seasonal adjustment into the model — typically a 30% downward adjustment on April + May Commit numbers, with the deferred revenue showing up in June and July.

Cycles are longer. UAE enterprise deals close in 100–140 days; KSA enterprise in 150–220. Commit category eligibility in the GCC should require a longer "verbal yes to actual signature" timeline than Western frameworks suggest. A deal that has Economic Buyer verbal yes in the UK closes in 2–3 weeks; in KSA enterprise, it closes in 6–10 weeks because of formal procurement. Forecast accordingly.

Paper Process slippage is normal. Procurement and legal in KSA government-adjacent entities adds 4–10 weeks. Many GCC deals that look Commit at the end of one quarter actually close in the first week of the next quarter. Forecast Commit deals with a buffer week or two, or be explicit about which Commits are "this quarter" vs. "early next quarter."

Currency mix. Multi-currency forecasts (GBP + AED + SAR) need a consistent FX assumption set at quarter start. Changing the FX assumption mid-quarter changes the forecast number without any actual change in deal status — a major source of confusion. Lock the FX rates at quarter start; report any FX impact as a separate line in the variance analysis.

Relationship-status as a forecast input. GCC deals frequently have a relationship variable that does not fit cleanly into MEDDPICC. A deal where the founder has met the Economic Buyer in person three times is more likely to close than one where the founder has only emailed. Adding a "relationship strength" field (cold / introduced / met / dined / multi-meeting) to the CRM and weighting Commit eligibility by it improves GCC forecast accuracy materially.

Forecast call language. In bilingual UAE/KSA teams, run the forecast call in English (for tool compatibility and consistency) but allow Arabic clarifications on individual deals. Forcing rapid English-only forecasting on a non-native team produces worse data than allowing language flexibility on individual deal commentary.

What MAVEN does about it

Forecast structure and the weekly forecast call cadence are part of every Sales Process Program and a core deliverable of the Fractional VP Retainer. We install the categories, train the team, and personally run the first 4–6 forecast calls until the leader can take it over.

The Sales OS Blueprint covers the forecasting architecture in more depth — including the CRM field structure and the dashboards needed to support it.

If your forecast is off by more than 15% in either direction and you cannot tell why, book a virtual coffee. We will look at the last 4 quarters of your forecast vs. actual and identify the specific pattern.

Frequently asked

How often should I forecast?

Weekly for the current quarter; monthly for next quarter; quarterly for the next 12 months. Anything more frequent than weekly is noise. Anything less frequent than weekly misses too much.

Should the founder sit on the forecast call?

Up to about £3M ARR, yes — the founder still has visibility on most deals and adds judgement. Above that, the sales leader runs it and reports to the founder.

Is a weighted-pipeline forecast useful at all?

Marginally, as a sanity check. The weighted sum should be roughly within 10–20% of the Commit + (50% × Best Case) number. Big divergences suggest something is off with stage probabilities or categorisation discipline.

What about deals that close ahead of schedule?

Track them — they are signal that your sales cycle assumptions might be too conservative, or that your pull-in discipline is too loose. Either way, log them and review at quarter end.

Should every rep set their own forecast or should the manager set it?

Reps propose. Manager challenges. Manager and rep agree. The forecast is jointly committed but the manager has the final call — and the accountability for the team number.

What is the role of AI in forecasting?

AI flagging (Clari, Boostup, Gong Forecast) is genuinely useful — it catches patterns humans miss. AI-generated forecasts without human discipline produce worse outcomes than human forecasts with discipline. Use AI to augment, not replace, the weekly call.

---

Post 15 of our outbound + sales OS series.

Related reading

• What is a sales operating system?
• What is MEDDPICC?
• How to forecast B2B sales accurately
• How to build a B2B sales pipeline from scratch
• Healthy pipeline coverage ratio
• Sales discovery questions that work