of sales leaders and sellers have high confidence in their forecast accuracy
Gartner press release, 2020for Michigan’s commission to issue a final order on a rate case, from the filing date
Michigan Public Service Commissionaverage buying cycle in 6sense’s 2025 buyer survey (median purchase $200–300k)
6sense, 2025Stage by buyer evidence and give the approval its own date
Record an outside approval in its own fields, with its own date, and keep the stages for evidence the buyer has given. A deal that waits on a commission order runs on 2 clocks. The commercial clock belongs to you and the buyer: the evaluation, the terms and the buyer’s internal sign-offs. The approval clock belongs to someone else, such as a regulator, a council or a grant body. Neither the rep nor the champion can speed it up.
Most CRMs give a deal a single close date, so that date ends up carrying both clocks. A common patch adds the approval as a stage, such as “awaiting commission order”. Then a deal can sit at a late stage for months while every report calls it nearly won. Write each exit criterion as something the buyer did and the CRM can show: a written selection, a redline closed by procurement, the budget owner’s sign-off. Then give the approval the fields below.
| Field | What it holds | Where it comes from |
|---|---|---|
| Approval type | Commission order, rate case, council or board vote, grant award | The buyer’s approval path, recorded at qualification |
| Reference | Docket or case number, agenda item or grant ID | The public record |
| Decision date | When the approval is due, and the basis for that date | A statutory deadline, the procedural schedule or the meeting calendar |
| Status | Filed, scheduled, heard, decided, modified or denied | The public record, checked weekly by the deal owner |
| Approval owner | The person at the buyer who tracks the approval, such as regulatory affairs | Your seat map for the account |
| Lag to signature | The median and the longest time from approval to signature | Your closed-won deals with the same approval type |
| Close date | The later of the readiness date and the decision date plus the median lag | Calculated in the CRM, never typed |
The close date follows from those fields. It is the later of 2 dates: the readiness date, when the commercial clock should finish, and the decision date plus your median lag. Measure the lag on your own closed-won deals with the same approval type, and keep the median and the longest. Reps update the inputs, with evidence, and the CRM recalculates the date. Every move carries a reason: the approval moved, the buyer’s step moved or yours did.↗
The forecast category then follows a rule:
- Commit: the evidence for the current stage is complete, and the close date falls inside the quarter even at your longest lag. The seats that sign, and the person at the buyer who tracks the approval, have had a two-way exchange with you in the last 90 days.↗
- Best case: the same, except that the date falls inside the quarter only at your median lag.
- Pipeline: everything else.
Commit that slips without a competitor is a data problem first
When deals slip with no competitor involved, we think the problem is usually the data before it is the reps. Start with what commit means and what the CRM records. Only 45% of sales leaders and sellers have high confidence in their forecast accuracy.↗ We read that as a system problem: when most teams doubt their number, a few weak reps can’t be the whole story.
In our experience, commit on approval-gated deals usually means the champion is keen. The champion can’t speed up a docket, and coaching the rep won’t either. Coaching still matters on the commercial clock, but it does nothing for the approval clock. Clean definitions won’t make a regulator punctual. They make the forecast say what is known.
Baseline before you change a definition. Take the last 4–6 quarters of forecast snapshots and measure the error of the forecast you held in week 4 against what closed.↗ Then count the share of open deals whose close date moved twice or more.↗ Our audit scores a week-4 error under 5% as 5 and an error over 30% as 1.↗ It scores a share of deals moved twice above 40% as 1, and under 10%, with a reason required for every move, as 5.↗
Report every quarter after the change against both numbers, with the reason for each slip. Without the baseline, almost any new number can be framed as progress.
With 10 deals a year, forecast each deal by name
With about 10 deals a year, list each deal by name with its derived date. Report a floor and a ceiling rather than a weighted total. A weighted total assumes enough deals for the odds to average out. A quarter of 3 deals never gets there. Take 3 deals of €400k, each 60% likely to close this quarter: the weighted forecast is €720k.↗ The quarter can’t end there: it ends at €0, €400k, €800k or €1.2M.↗
A quarter of 3 deals of €400k at 60% each ends at €0, €400k, €800k or €1.2M, never at the weighted €720k
Data behind this chart
| Item | Value |
|---|---|
| €0: no deal closes | 6.4% |
| €400k: 1 deal closes | 28.8% |
| €800k: 2 deals close | 43.2% |
| €1.2M: all 3 close | 21.6% |
The most likely result, €800k, comes up 43% of the time, and €400k or less 35% of the time.↗ More deals narrow the swing without removing it. In our earlier calculation, 20 deals of €75k, each honestly called at 30%, still left a typical swing of about a third either way.↗
So the board pack shows 3 things: the floor (commit), the ceiling (commit plus best case) and the deals in between, by name. A miss then comes with a deal name and a reason.
The weekly forecast call changes with it. Reps stop moving close dates by hand and report what changed in the inputs: new evidence, a moved decision date, a seat that went quiet. The CRM recalculates the dates, and the call spends its time on the deals whose inputs moved.
Finance and the board read the CRM’s weekly snapshot
The CRM, the finance model and the board pack agree when finance and the board read a dated snapshot of the CRM’s deal fields. Neither keeps deal facts of its own. Take the snapshot at the same time every week. It holds each open deal’s stage, category, amount, approval fields and derived close date, and every move with its reason. Finance names the snapshot date in its model and adds its own rules on top. Those decide when a signature becomes a booking and when revenue starts. The board pack uses the snapshot the finance model used, so both show the same deals on the same date.
| Report | Reads from the snapshot | Adds | Never holds |
|---|---|---|---|
| CRM forecast | Stage, category, amount, approval fields and the derived close date | A reason for every close-date move | A close date typed by hand |
| Finance model | The same fields, with the snapshot date named | Booking, revenue and cash timing rules | Deal dates or amounts of its own |
| Board pack | The snapshot the finance model used | The floor, the ceiling and the named deals that moved, with reasons | A number that isn’t in the snapshot |
When the reports still disagree, someone has retyped a deal fact. Trace the difference to the snapshot and fix the field in the CRM, where the next snapshot picks it up. Finance can still show a cautious case, built from the same snapshot at your longest lags. It shouldn’t re-date deals in a spreadsheet, because that is how 3 versions of the same quarter begin.
What happens to a commit list waiting on a rate case?
About half of it leaves commit once its close dates are derived. The example below is illustrative: the vendor, its utilities and its lags are fictional, and only the Michigan deadline is real. A meter data analytics vendor goes into Q4 2026 with 4 deals in commit worth €1.4M, half of them waiting on a state commission.↗ Its own closed deals set its lags. From a commission order to signature, the median is 6 weeks and the longest 11; after a council vote, 2 and 4 weeks.↗
The market adds its own steps and seats. At a utility buying software worth €100k or more, 6–12 people sit on the panel, by our estimate.↗ The seats that can stop the deal include finance and regulatory finance, such as a director of rate case.↗ That seat knows the docket’s dates, so we’d make it a required seat before commit. A rate case takes about a year by our estimate, and approved spend usually follows the order within 12 months.↗ In Michigan, an administrative law judge sets the schedule and writes a proposal for decision.↗ The commission has 10 months from the filing to issue its final order.↗ If it misses that deadline, the rate increase is approved automatically.↗
| Deal (fictional, Illustrative) | Waits on | Close date at median lag; at longest | Q4 2026 category |
|---|---|---|---|
| City utility, €300k | A council vote on 17 November 2026 | 1 December 2026; 15 December 2026 | Commit |
| Co-op, €400k | Nothing outside: its board approved in September | 20 November 2026, from the mutual plan | Commit |
| Investor-owned utility in Michigan, €450k | A rate case order, due by 15 December 2026, 10 months from the filing | 26 January 2027; 2 March 2027 | Moves to Q1 2027 |
| Investor-owned utility, €250k | Commission approval of its metering programme, expected on 19 November 2026 in the docket’s schedule | 31 December 2026; 4 February 2027 | Best case |
The Michigan deal shows the rule at work. Its utility filed on 15 February 2026, so the order is due by 15 December.↗ The median lag puts the signature on 26 January 2027, and an order 2 weeks early still lands it in January.↗ Commit falls from €1.4M to €700k, and best case adds €250k.↗ The €450k deal becomes a Q1 2027 deal, with a date the board can check.↗
How much coverage does a 9–15 month cycle need?
Set coverage for each close quarter separately, count only deals that can close in it and take the multiple from your own in-quarter close rate. Cycles in energy and utilities are long. By our estimate, software deals with utilities typically take 9–15 months, and up to 36 at a US investor-owned utility.↗ Deals with energy companies take 6–15 months.↗ Buyers in 6sense’s 2025 survey, with a median purchase of $200–300k, reported an average buying cycle of 10.1 months.↗ We have no separate figure for AI software; plan it on the same cycle as other software until your own closed deals say otherwise.
Generic multiples are only a start. Salesloft puts enterprise teams with 15–25% win rates at 4–7×.↗ It tells teams with cycles of 120 days or more to measure across a rolling 2-quarter window.↗ On a 9–15 month cycle we’d count a deal towards a quarter only when it passes the commit tests apart from the date. It needs complete evidence for its stage, the required seats engaged and a derived close date inside that quarter.↗
The multiple is 1 ÷ the share of such deals that closed in their quarter over the last 4–6 quarters.↗ If 1 in 3 did, a quarter needs 3× its target in qualifying deals.↗ Count deals as well as euros: at €400k a deal, a €1.2M quarter is 3 deals, and 3× is 9.↗
The calendar matters more than the multiple. With cycles of 9–15 months, deals that close between July and September 2027 are created between April and December 2026.↗ Most of that quarter’s pipeline is already in your CRM or isn’t coming. Set a coverage target for each of the next 4 quarters and check it every month.↗
In practice
How we do it at Panelhop
A Panel Check (GTM audit · 2–3 weeks) baselines your forecast from your own CRM before anything changes: the error of the forecast held in week 4 against what closed, close-date moves and stage discipline, from the last 24 months of deals and the forecast snapshots you have.↗
In a Leak Fix (we build the fixes), the deal-process module writes the stage exit criteria and the approval fields into your HubSpot or Salesforce, calculates the close date, requires a reason for every move and stores the weekly snapshot that finance and the board read. Panel Ops (we run it monthly) then reports slips and forecast error against the baseline. Everything we build is documented and stays in your accounts.
Questions buyers ask about this
How long is the sales cycle for selling AI software to energy companies?
By Panelhop’s estimate, most software deals with energy companies take 6–15 months, up to 36 months for a German municipal billing tender, and deals with utilities typically take 9–15 months. We have no separate figure for AI software, so plan it on the same cycle until your own closed deals say otherwise. Buyers in 6sense’s 2025 survey reported an average buying cycle of 10.1 months.
Can a deal be commit while the commission order is still pending?
Yes, if the stage evidence is complete, the seats that sign are engaged and the order is due early enough: your longest approval-to-signature lag must still put the signature inside the quarter. If only your median lag gets it in, it’s best case. An order due in the last weeks of the quarter usually makes the deal next quarter’s commit.
How do I fix a forecast call spent moving close dates, with about 10 ETRM and billing deals a year?
Stop typing close dates: derive them from the readiness date and the decision date plus your measured lag, and require a reason for every move. Use the call to review what changed in the evidence, the approval dates and the seats. With about 10 deals a year, review each deal by name and report a floor and a ceiling.
What if the commission modifies or denies the request?
Record the outcome in the approval status and move the deal back to the stage the order reopens: a cut in approved spend usually reopens scope and price. Give these outcomes their own reason code, apart from timing slips, because they point to qualification while a timing slip points to the calendar.
Should finance keep its own haircut on approval-gated deals?
We’d rather it didn’t, because a separate haircut is how the CRM, the finance model and the board pack drift apart. Finance can show a cautious case from the same snapshot, using your longest lags, and apply its booking and revenue rules on top.
