Defining Forecast Accuracy

Strive to call the final (Day 90) gross new bookings¹ amount for the quarter within +/-5% as of Day 15. The reason for Day 15 is that most organizations conduct quarterly business reviews during the first two weeks of the quarter to clean up the pipeline.

How to Achieve an Accurate Forecast

While many revenue leaders commit to a number by triangulating across multiple approaches, we have found the Weighted Pipeline + Create & Close approach to be the most accurate.

Weighted Pipeline + Create & Close Approach (Best)

Accurate forecasting starts with strong pipeline hygiene. Every opportunity should be reviewed by reps and 1st line managers in advance of each forecast roll up. For most organizations, this means weekly or every other week. For forecasting purposes, the key fields are amount, stage and/or forecast category, and close date.

Probabilities should be updated periodically (we recommend quarterly) by looking at win rates over the prior 2 to 4 quarters. To do this, pull the start-of-quarter open pipeline with close dates in-quarter including key fields that distinguish different groups of opportunities (type, segment, etc.).

For highest forecast accuracy, probabilities should be based on conversion rates [ ($ won by end of period of opps open at start of period) / ($ of opps open at start of period) ] rather than win rates [ ($ won) / ($ won + $lost) ]. Dollar-based rates are similar to count-based rates if the ASP of won deals is similar to lost or pushed deals; otherwise, dollar-based rates are more accurate.

To estimate the % of bookings created and closed (C&C) in the quarter range we will start with a simple method and layer on additional complexity.

C&C Method 1A

The most basic method for determining C&C is to look back at the prior 2 to 4 quarters and calculate the percentage of bookings during the quarter for opportunities that did not exist on Day 1 of the quarter. Let’s say that averages 38% overall. Then, just apply that rate to the bookings target for the current quarter adjusting, if needed, for quarter-to-quarter seasonality.

On any given day of the quarter, estimate the remaining C&C amount via pro-rating (more on this below).

If you are thinking, “Wait a second, isn’t it a self-fulfilling prophesy to use the current quarter bookings target?” then you’d be correct. Nonetheless, it works in part of whole because the bookings target was set based on the availability of inputs like marketing spend and sales capacity that in turn tie to C&C.

C&C Method 1B

Should applying the current quarter bookings target not sit well with you, then you can instead scale up your pipeline as follows.

Imagine a company starts the prior quarter with $100 of in-quarter unweighted pipeline. Of this, they convert 30% or $30. In addition, they create & close another $18.4 for a total of $48.4. Since I like to apply the term ‘conversion rate’ to just those opportunities that exist at the beginning of the period, I’m going to call this scaling factor (48.4/100=0.484) the ‘adjusted conversion rate’ since it includes C&C.

Now, if my starting pipeline for the current quarter is $120, then my forecast would be $58.1 = 120 * 0.484. This can be decomposed into $36 (120 * 30/100) of converted starting in-quarter pipeline and $22.1 (120 * 18.4/100) of forecasted C&C bookings.

C&C Method 2

The second method for estimating C&C gross new bookings applies the approach methods 1A and 1B at a more granular level.

Enhancing Method 1A by splitting percentages by opportunity type and (company size) segment as shown in Table 1 below is my recommend approach for estimating C&C. This is the most common split but one need only go down to the level where there are meaningful differences in percentage. Notably for this method, one must have targets for new bookings at this level of granularity; larger companies set such targets as a matter of routine and I advise smaller companies to do the same regardless of how they forecast C&C. Remember, these are percentages of the granular gross new bookings targets; for example, 36% in Table 1 means that C&C is expected to represent 36% of the SMB+New Business target for the quarter.

Opportunity Type	SMB	Enterprise
New Business	36%	21%
Amendment (aka Expansion)	74%	62%

Table 1: C&C as Percent of Granular Gross New Bookings Targets by Opportunity Type and Segment

If extending Method 1B, simply calculate adjusted scaling factors at your chosen level of granularly and apply to your start of quarter pipeline. One challenge with this is that you’ll be using fewer data points to compute each scaling factor which means each factor will be less robust. This may lower accuracy especially for combinations with very short sales cycles; for instance, SMB+Expansion may have an adjusted scaling factor much greater than 1 making it highly sensitive to start-of-quarter pipeline.

C&C Method 3

For Method 3, I’m going to broadly outline a variety of approaches that are best left to data scientists and should only be used if the simpler methods above do not get your Day 15 forecast within +/- 5% of your final number.

Time-Series Models: Estimate C&C by applying time-series forecasting techniques taking into account trend & seasonality. The two most common approaches are auto-regressive integrated moving average (ARIMA) and Holt-Winters Exponential Smoothing. Time series models can rely exclusively on historical C&C amounts or can include other variables.
Input Models: Estimate C&C based on relevant current-state or planned inputs such as ramped sales capacity, demand generation spending, installed base, etc. By way of example, many companies estimate expansion bookings as a simple percentage of Total ARR.

Pro-rating Create & Close

Since you’ll want to regularly update your forecast, you’ll need a mechanism to ‘burn down’ your Day 1 C&C estimate. Go with the easiest approach that still delivers a Day 15 forecast that is within +/-5% of your final gross new bookings.

The easiest way and the one I recommend is to linearly pro-rate by the remaining days in the quarter. For simplicity, I’ve always done this with calendar days in a quarter rather than business days.

Quarterly C&C remaining = (bookings target) * (Day 1 C&C Percent) * (days remaining in quarter) / (total days in quarter)

Though rarely necessary, one could add a bit more complexity by burning down at different rates throughout the quarter. The typical trend is that bookings are distributed 25%-25%-50% across months 1, 2, and 3 of the quarter. The formulas for doing this a simple algebra but are a bit messy:

Quarterly C&C remaining during month 1 = (bookings target) * (Day 1 C&C Percent) * [1 – (month 1 bookings %) * (days in month 1 elapsed) / (days in month 1)]

Quarterly C&C remaining during month 2 = (bookings target) * (Day 1 C&C Percent) * [1 – (month 1 bookings %) – (month 2 bookings %) * (days in month 2 elapsed) / (days in month 2)]

Quarterly C&C remaining during month 3 = (bookings target) * (Day 1 C&C Percent) * [1 – (month 1 bookings %) – (month 2 bookings %) – (month 3 bookings %) * (days in month 3 elapsed) / (days in month 3)]

Additional Triangulation Approaches

I believe the weighed pipeline + C&C approach is so accurate that additional triangulation not only wastes time but also leads to a less accurate forecast, This is particularly true since sales leaders tend to forecast from their ‘gut’ after looking at estimates from one or more approaches. In my experience, their gut is strongly drawn toward the target bookings number even when better analytical approaches suggest otherwise.

Triangulation Method 1: Pipeline Multiplier

Here, look back at each of the last two to four quarters and calculate the ratio of (final gross new bookings) / (unweighted in-quarter pipeline at the start of the quarter). This is virtually same starting point as C&C Method 1B but we will not worry about which deals existed at the start of the quarter and which did not. In our earlier example, the company had $48.4 in final gross new bookings and $100 in unweighted in-quarter pipeline at the start of the quarter. Hence, the pipeline multiplier is 2.07 (= 100 / 48.4).

The formula for computing the forecast is simply:

Forecast for the quarter = (bookings closed won so far) + (pipeline multiplier) * (remaining in-quarter unweighted pipeline)

Just as with the weighted pipeline + C&C approach, one can develop forecasts at more granular levels such as SMB + New Business.

Triangulation Method 2: Commit ‘Divination’

Many sales leaders like to look at several of the following combinations of closed won + unweighted pipeline by forecast category and mystically divine the forecast:

Closed Won + Commit
Closed Won + Commit + Most Likely
Closed Won + Commit + Most Likely + Best Case
Closed Won + Commit + Most Likely + Best Case + Pipeline
(In this context, “Pipeline” is the least likely to close forecast category which is admittedly confusng but a long-standing convention.)

For Enterprise new business with very long sales cycles, this is a decently accurate approach in the last 4 to 6 weeks of the quarter. However, forecast category weighting each opportunities is arguably easier and nearly always more accurate.

Triangulation Method 3: Call Roll-up

In my opinion, the most time consuming and least accurate method of forecasting is asking reps and their managers (and so on up the line) for the number they are ‘committing to’ for the quarter. At each level, leaders use their minds to massage the number. Most often, each of them nudges toward their personal target giving this approach little value. Supporters of this approach argue there is power in getting people to make a public commitment which in turn make the prophesy more likely to be fulfilled.

How to Run a Weekly Sales Forecasting Call

General

Create a culture around this meeting that the only goal is to have an accurate forecast. As such, it is a safe-zone for sharing bad news.
Define the criteria for what constitutes a ‘key deal’ that should be reviewed during the forecast call. This is typically a combination of deal size, close date, and stage/forecast category. (And beware, reps will often ‘hide’ deals but putting the dollar amount just below threshold or the close date just beyond the window)
Leaders at every level must consistently enforce the deal health methodology (ex: MEDDPICC) in every deal review & forecast review
Include representatives, as relevant, from other departments, especially Marketing and Product.
Set an expectation that reps and their first line managers should be able to speak in depth on every key deal

Pre-Meeting

Ensure all deals are up-to-date in CRM before the meeting
Notify attendees of all deals that will be discussed in in what order

During Meeting

Look at the overall forecast versus plan, including the ‘flow’ of changes from the prior week.
Review deals from highest to lowest priority. If needed, impose a time-limit for discussing each deal. You may need a dedicated time keeper for this. The highest priority deals typically warrant longer time slots.
Progression of all key deals discussed last week
- Have there been changes to stage/forecast category, amount, close date
- Discuss blockers and deal gaps relative to your deal health framework (ex: MEDDIPCC); ask reps what help they need to advance each key deal
- Status of / additions to next steps & actions for each key deal.
- Spend a few minutes on closed won and closed lost deals to learn for the future
Top new key opportunities created in the last week
Discuss any changes in market conditions, competition, or internal factors that affect the sales strategy

Post Meeting

Send recap notes – Summarize key points, actions, and deadlines in a follow up email.
Ensure reps get the help they have requested to unblock key deals
Managers should individually follow up during the week to ensure that action items are being carried out.

Forecasting Whale Deals

(best) If whales are a bit more common, then apply a different set of win rates than for ‘regular’ deals. Also, I’ve found it more useful to weight deals using win rates by forecast category than win rates by stage for large deals assuming reps are more on top of the former as is often the case in Enterprise.
(safest) If you have a very small number of whales, esp. whales with uncertain close dates, then simply exclude them from the forecast. When they come in, they are upside surprise. Strive to meet/exceed your company target solely via regular deals.

Renewal Forecasting

Start renewal discussions between 120 and 90 days before contract end
Tune renewal probabilities based on:
- Health scores
- Segment / firmographics
- Contract value
- Cohort tenure (longer-term customers tend to renew at higher rates)
Monitor health scores (requires strong data management and shared systems of record/action)
- Configuration
- Engagement (incl. Monitoring for loss of decision makers, champions, and/or power users)
  - Executive engagement (maintain throughout customer lifecycle; “If anything goes wrong, call me and I’ll make it right.”)
  - User engagement (seat usage and level of utilization)
  - Monitor for M&A activity and have a plan to react
- Value realization / satisfaction
  - Includes CSM sentiment (green/yellow/red) which is actually quite accurate (should be updated at least monthly)
  - NPS (measuring alone is not enough; it is a system of action)
  - Service CSAT (no news is not good news)
Run orchestrated plays to increase health throughout the customer lifecycle
- To QBR or not to QBR? (If your contract was up today, would you renew? What, if anything, might prevent you from renewing?”)
You don’t just want an accurate renewals forecast, you want to proactively drive higher NDR
Automatically generate renewal opportunities in CRM
Triangulate with top-down (NDR applied to renewing CV) and bottom-up (by account)
Role clarity
- Who should own the job to be done of renewing customers
- Who should own the job of expanding customers
Strive for:
- multi-year contracts (often new logo is a single year but renewals should be multi-years)
- Auto-renewals (opt-out rather than opt-in)
- built-in price increases
Consider having an account recovery/escalation team (esp. Prevents CSMs from operating in red alert mode)
- Ensure there is a ‘no shame’ culture in CS for having a ‘red’ account
Customer marketing, product enhancements, and pricing/packaging changes play a key role
If/when renewing is reselling (more often the case than people think), then apply deal inspection rigor as you would for new business (ex; MEDDICC)
- Ties to a culture of celebrating renewal rate rather than framing negatively as churn rate
At what point do you turn off service and mark the renewal as closed lost?
- If client went dark, then turn off at contract end date
- If client engaged, then evaluate every 30 days.

Forecasting Software Vendors

(*) = recommended

Footnotes

¹ Gross new bookings includes new business and expansion but excludes churn and downgrades.

² Disclosure: Jeremey Donovan, the author/editor of this page, is a former Salesloft employee. As of March 2023, he and his employer (Insight Partners) both hold equity in Salesloft.

³ This is an Insight Partners portfolio company. The editor of this page is employed by Insight Partners.

Forecasting