Pipeline Management for SMEs: Stages, Probabilities, Forecast

Why is the revenue forecast almost never right?

The question “What will come in next quarter?” is among the most important in a business. Hiring, purchasing, and liquidity depend on the answer. In many small and mid-sized companies, the honest answer is: we will know when we get there.

The reason rarely lies in the arithmetic. It lies in the list the arithmetic is based on. It contains opportunities nobody has touched for months. It contains “looks good” next to an amount, and nobody can say how you would tell. It lacks cases that exist only in a colleague’s head. Add up this list and you get a number, but not a forecast.

The question this article sets out to answer: how does the list of open opportunities become a pipeline that a revenue forecast can rely on? To answer it, we first clarify what belongs in a pipeline, then define stages by evidence and maintain closing probabilities, and finally turn to the forecast, keeping the list clean, and whether a spreadsheet is enough.

A pipeline from which nothing is ever deleted is an archive of hopes.

What is a sales pipeline and what belongs in it?

The pipeline is the list of all open sales opportunities. A sales opportunity is a specific possible order from a particular customer. Five details belong to each opportunity: customer, estimated value, stage, next step with a date, and the person responsible.

The pipeline is not an address book and not a list of all inquiries. It begins where an inquiry has become a verifiable opportunity. How to separate good inquiries from bad ones before that point is described in the article on lead qualification. The workflow the stages represent is the subject of the article on how to optimize your sales process. The pipeline makes that workflow visible for every single case.

How do I define pipeline stages? The Evidence Stages model

The most common mistake with stages is defining them by the salesperson’s feeling: “interested,” “hot,” “about to close.” Two people mean different things by these, and the same person means something different on Friday than on Monday.

The Evidence Stages model turns this around. It is the mental model of this article. Every stage is defined by something the customer has verifiably done. An opportunity only advances once the evidence exists. What you did yourself, such as sending a quote, counts only together with a reaction from the customer.

The starting values are freely chosen assumptions to begin with. They are not industry figures. You may adapt the stage names. What matters is that for every stage there is a sentence beginning with “The customer has …”.

How do I maintain win probabilities?

The win probability of a stage is the share of opportunities that are eventually won from that stage onward. It belongs to the stage, not to the individual case. As soon as every salesperson assigns a personal percentage to every opportunity, the estimate is back to gut feeling.

Research shows how strongly the salesperson’s judgment shapes the outcome. Xu, van der Borgh, Nijssen, and Lam (2022) examined in three studies how salespeople decide when prospecting. The relationship between the size of an opportunity and conversion followed an inverted U-shape: salespeople focus on medium-sized opportunities and tend to leave very small and very large ones unattended, because they weigh effort against benefit and conserve their resources. In an experiment with 216 experienced salespeople, willingness to pursue a prospect was highest at an order value of 50,000 dollars, which was typical for the territory. At 1,000 dollars and at one million dollars it was lower. A simulation with the data from the first study showed: if a new opportunity is one standard deviation above what the salesperson is used to, the odds of conversion drop by more than 15 percent. In solution selling, salespeople also account for uncertainty across their whole portfolio of opportunities. According to the authors, models that ignore this overestimate or underestimate conversion rates by up to 100 percent. Past success and experience shift the judgment further. The authors acknowledge that size and uncertainty are not the only influences. For your pipeline this means: an unusually large opportunity is won less often than the percentage of its stage suggests.

In practice this means: rely on counted rates per stage. This is how you get from starting values to your own values:

1. Keep records for one quarter

   For each opportunity, note which stages it reached and whether it was won or lost. Lost and dormant opportunities count too.

2. Count per stage

   How many opportunities reached the stage, and how many of them were won in the end? The quotient is your win probability for that stage.

3. Replace the starting values

   Enter the counted values into the table. Round to five percent. A few dozen cases do not support more precision.

4. Measure again annually

   Repeat the count once a year or after major changes to your offering and target group.

How do I build a revenue forecast from the pipeline?

The simplest robust forecast is the weighted pipeline. You multiply the value of each opportunity by the probability of its stage and add up the results. The following worked example uses freely chosen numbers and the starting values from the table above.

In the example, 96,000 euros are on the list, while roughly 33,000 euros are robust. Anyone planning with 96,000 euros is planning with almost three times what is to be expected based on their own rates. The chart shows a second anomaly: more than a third of the list value hangs on a single early opportunity.

For quarterly planning, the expected closing date comes into play. Count only opportunities whose decision falls within the period. From the forecast you can also calculate backwards how much pipeline a revenue target requires. That calculation is described in the article on how to plan sales targets.

How do I keep the pipeline clean?

A forecast is as good as the list behind it. The list only stays good if it is reviewed regularly. Twenty minutes a week are enough for that in most small teams. You can attach the review to your weekly meeting, as described in the article on weekly team planning.

• Every opportunity has a next step with a date in the future.
• Every stage is backed by evidence recorded in the note.
• Opportunities without contact for more than 30 days get a date or are closed.
• Lost opportunities receive a reason from a short fixed list.
• New opportunities from the week are entered, including those from phone calls.
• The weighted total is noted and compared with the previous week.

Closing is hard, but it is the most important part. Claro, Plouffe, and Vieira (2023) found in salesperson data that too many opportunities pursued at once overload salespeople and reduce goal commitment and performance. Too few opportunities are harmful as well. A tidy pipeline therefore protects attention for the cases that are genuinely open.

Loss reasons are your cheapest learning material. In the case study by Surdziel and Gdowska (2026) on a Polish IT consulting firm, a form for analyzing won and lost bids was among the simple aids that supported coordination, learning, and monitoring. Four to six reasons are enough, for example price, timing, competitor, no need, no reply.

Do I need a CRM for the pipeline, or is a spreadsheet enough?

To begin with, a spreadsheet with one row per opportunity and the five details from above is enough. It has one advantage: you learn which stages and fields you really need before you spend money. The principle behind this is set out in the article on why the process comes before the tool.

A CRM, meaning software for managing contacts and sales opportunities, pays off when several people work on the list, when reminders for dates get lost, or when evaluating the rates by hand takes too long. What small teams should look for when choosing is compared in the article on CRM for small businesses.

As the volume of data grows, automated methods come into play. Wu, Andreev, and Benyoucef (2026) present a two-stage model that first groups leads by observable characteristics such as company size, title, and management level and then ranks them within each group by likelihood of conversion. It is based on 784,493 leads from 40 inside sales companies in three service industries in North America, recorded from 2011 to 2024. In the held-out test data, the average conversion rate of the leads given priority rose from 6.2 to 8.9 percent, through the grouping alone and without behavioral data. This is an analysis of existing data, not a field experiment. As limitations the authors name the small number of characteristics and the restriction to service providers. Habel, Hartmann, Wiseman, Ahearne, and Vaid (2025) found that automated lead nurturing raised the probability of conversion by 23 percentage points on average in one study with 13,525 leads, and had no effect on average in two further studies. Both papers presuppose data that only exists once a pipeline has been kept cleanly over a longer period.

How to get started

1. Create the list: Enter all open opportunities with customer, value, next step, date, and person responsible into a spreadsheet.
2. Define stages: For four to five stages, write one sentence each that begins with “The customer has …”.
3. Assign stages: Assign each opportunity to the stage whose evidence exists, not the one that feels right.
4. Clean up: Close everything that has been without contact for 30 days and gets no new date.
5. Weight: Use the starting values to calculate the weighted total and note it with the date.
6. Set the appointment: Fix 20 minutes per week for the review.

After one quarter you have your own rates per stage and a series of weekly values. Then you see for the first time how close your forecast was to actual revenue, and you can adjust the probabilities. Which further numbers can be read from the pipeline is shown in the article on sales metrics for small and mid-sized businesses.

In short: a revenue forecast becomes more reliable when each opportunity is assigned to the stage whose evidence exists, when opportunities without contact are closed, and when probabilities are adjusted to your own rates. The arithmetic itself is simple. What carries the forecast is regular maintenance of the list.