Pipeline forecasting for manufacturers: next quarter’s orders and plant load from open quotes
Pipeline forecasting predicts orders and plant load from the quotes that are still open.
What pipeline forecasting means for a manufacturer
Pipeline forecasting comes from sales teams that track deals in a CRM. Each deal sits at a stage, and each stage carries a probability of closing. HubSpot’s list of deal properties shows the idea: its weighted amount is the deal amount multiplied by the deal probability. The probability updates when someone moves the deal to a new stage.
In a plant that quotes custom or engineered products, the pipeline is the list of open quotes. Each quote has a value, a stage, a customer, and a date, whether it sits in an ERP, a CRM, or a spreadsheet. A pipeline forecast turns that list into two answers: the orders likely in the next few weeks, and the shop hours they will need.
The weighted pipeline formula
The formula has three parts:
- Weighted value = quote value × win probability.
- Expected order date = quote date + the usual days from quote to order at that stage.
- Weighted pipeline for a month = the sum of the weighted values expected in that month.
A $100,000 quote with a 30% win probability adds $30,000 to its month. Nobody expects a $30,000 order. Across many quotes, the weighted values add up to an expected total, and the wins and losses partly cancel out.
The fewer and larger the quotes, the less that averaging works. One large quote can swing a month, as the worked example shows.
Win probabilities from your own quote log
CRM defaults are generic. HubSpot’s default deal pipeline starts at 20% for Appointment scheduled and rises to 90% at Contract sent. They are the software’s starting values, not anyone’s history.
Your quote log holds the real rates. A win rate is quotes won divided by quotes won and lost, and the guide to win-loss analysis sets out the counting rules. Compute it for each stage, then split it further wherever your log shows a difference:
- by customer, such as repeat against new
- by quote size or product family
- by lead time, if long lead times lose orders
Close quotes at a fixed age, such as 90 days, as the win-loss guide suggests. A budgetary quote can take longer than that, so keep it open until its expected order date has passed. A stale quote left open keeps its probability and inflates every month it sits in.
Forecast next quarter’s orders from your open quotes
Tell Derik where your quotes are logged and how long they usually take to become orders. He will tell you whether that history is enough to weight your open quotes.
Start a conversationWhen the order lands: the expected order date
The probability says whether a quote turns into an order. The forecast also needs when. Take the median days from quote to order for each stage from the same log, and add it to each quote’s date.
When a quote’s expected date has already passed, count it in the current month and flag it for a follow-up call. An overdue quote is either about to close or already lost, and only the customer knows which.
A pipeline only sees as far ahead as the quote cycle. If quotes take six weeks to become orders, the pipeline covers about the next six weeks. Orders after that will come from quotes nobody has written yet.
Worked example: 14 open quotes on October 1
A plant builds custom electrical enclosures and control panels. On October 1, 2026, its quote log holds 14 open quotes worth $844,200. The numbers below are made up to show the method, and a script computes every result.
The plant’s own history gives these rates:
| Stage | Win rate, new customer | Win rate, repeat customer | Median days from quote to order |
|---|---|---|---|
| Budgetary | 5% | 12% | 110 |
| Firm quote | 18% | 32% | 45 |
| Negotiation | 40% | 58% | 28 |
| Verbal yes | 75% | 90% | 12 |
Two quotes are over 90 days old and past their expected order date. The plant closes them as stale, and $95,000 leaves the pipeline. The other 12 are weighted and dated. An overdue quote passed its expected date before October 1, so it counts in October:
| Quote | Stage, customer | Value | Win probability | Expected order | Weighted value |
|---|---|---|---|---|---|
| Q‑3088 | Verbal yes, repeat | $82,000 | 90% | Oct 7 | $73,800 |
| Q‑3094 | Negotiation, new | $126,000 | 40% | Oct 6 | $50,400 |
| Q‑3097 | Verbal yes, repeat | $26,300 | 90% | Overdue | $23,670 |
| Q‑3051 | Negotiation, repeat | $37,200 | 58% | Overdue | $21,576 |
| Q‑3080 | Firm quote, new | $94,000 | 18% | Oct 17 | $16,920 |
| Q‑3069 | Firm quote, repeat | $15,900 | 32% | Oct 4 | $5,088 |
| Q‑3072 | Firm quote, new | $23,400 | 18% | Oct 12 | $4,212 |
| Q‑3040 | Firm quote, new | $9,800 | 18% | Overdue | $1,764 |
| Q‑3108 | Firm quote, repeat | $57,500 | 32% | Nov 13 | $18,400 |
| Q‑3101 | Firm quote, repeat | $48,500 | 32% | Nov 6 | $15,520 |
| Q‑3055 | Budgetary, new | $210,000 | 5% | Nov 22 | $10,500 |
| Q‑3112 | Budgetary, new | $18,600 | 5% | Jan 18 | $930 |
October carries $197,430 of weighted orders, November $44,420, and December nothing yet. For the quarter, 11 quotes with a face value of $730,600 weigh in at $241,850, or 33.1%. Three quotes are overdue, Q‑3097, Q‑3051, and Q‑3040, so each gets a call this week.
From dollars to shop hours: the plant load
The shop plans in hours, not dollars. Convert each weighted value with the plant’s own ratio of shop hours per $1,000 of sales, from routings or past jobs. In this plant, enclosures take 7.2 hours per $1,000 and control panels 9.6. Each order is built in the month after it arrives.
| Build month | Booked backlog | Expected new orders | Total | Share of 3,300 h capacity |
|---|---|---|---|---|
| November 2026 | 2,240 h | 1,760 h | 4,000 h | 121% |
| December 2026 | 1,180 h | 389 h | 1,569 h | 48% |
| January 2027 | 420 h | 0 h | 420 h | 13% |
Capacity is 22 people at 7.5 hours a day for 20 working days. If the expected October orders arrive, November runs 700 hours over. The plant can plan overtime, move work forward into October, or quote later dates on new orders. Capable to promise checks a date against capacity before it is promised, as the guide to available to promise explains.
December and January look light because their quotes are not written yet. That is the edge of the pipeline’s horizon, not a forecast of an empty winter.
Weighting also hides lumps. Q‑3055, a $210,000 budgetary quote at 5%, adds 101 hours to the forecast. If it is won, it brings 2,016 hours, or 61% of a month’s capacity. Plan the weighted case, and keep a list of the large quotes that would change it.
Sales forecasting methods compared
The weighted pipeline is one of several sales forecasting methods. Here are the plant’s fourth-quarter orders under three of them, plus the pipeline counted in full:
| Method | How it works | Q4 2026 orders | Fits when | Misses |
|---|---|---|---|---|
| Weighted pipeline | Open quotes × win probability, by expected order month | $241,850 | Orders are large and arrive through quotes | Orders from quotes not written yet |
| Run rate | The average of the last three months, carried forward | $621,000 | Many small orders arrive at a steady pace | Seasons and turning points |
| Same quarter last year | Last year’s orders for the same months | $495,000 | Demand follows the calendar and customers stay the same | Growth and this year’s quotes |
| Every quote in full | The face value of all open quotes | $730,600 | Never, as a forecast | Every quote that will be lost |
The run rate averages July to September, which took $212,000, $168,000, and $241,000 of orders. Same quarter last year is what the textbook Forecasting: Principles and Practice calls the seasonal naïve method.
The numbers disagree because each method reads different data. A practical forecast uses the weighted pipeline for the weeks the quote cycle covers, and a time series after that. The guide to forecasting in Excel covers the time series side for one item, and AI demand forecasting extends it across every item.
Many sales teams add judgment on top. Salesforce’s pipeline forecasting guide lists the standard forecast categories as Pipeline, Best Case, Commit, Omitted, and Closed, assigned from each opportunity’s stage.
Where AI does the work
The formula is simple. Keeping its inputs current is the work, and that is where AI helps:
- Reading the email trail. A language model reads the messages on each quote and suggests a new stage or date. A buyer writing that the purchase order goes out next week is one example.
- Flagging quotes that need a call. It lists overdue quotes, quotes past the 90-day limit, and quotes with no contact in weeks.
- Filling gaps in the quote log. It reads the quote document for missing fields, such as the product family or whether the customer is new.
- Estimating each probability from more than stage. A model trained on the quote log can weigh customer, size, lead time, and product together.
- Drafting the weekly forecast. Orders and shop hours by month, with the quotes behind each number.
A sales manager or planner approves each change before it reaches the forecast, and the log keeps the reason. Human in the loop covers how to design that approval step.
How ThriveAI builds your pipeline forecast
ThriveAI builds the forecast from the systems you already use. It reads the quote log, the CRM, and the ERP’s orders and routings, with connections that only read data. It works out win rates and quote-to-order times from your own history, then drafts each week’s forecast of orders and shop hours by month. A named person approves the forecast and every change to a quote’s stage or date. The system is designed to keep your data on its own server in Canada. The same history feeds a win rate report, as win-loss analysis describes, and the dates you promise on new orders.