Building a Sales Pipeline From Scratch
Build a sales pipeline from nothing: define a deal, name stages after what buyers do, set rules, track a few numbers and start in a spreadsheet.

In this article
- Decide what a deal is
- Find the real steps your buyers take
- Name stages after what the buyer has done
- Write an entry and an exit rule for every stage
- The next-step rule: every open deal has an action and a date
- Closing a deal as lost, and why that column matters
- The numbers worth tracking
- The weekly review
- Set it up in a spreadsheet today
- When to move to software
- Where software fits
- Your first pipeline in one week
- Sources
If you have never had a sales pipeline, you have probably still been selling. Customers found you, you quoted, some said yes. What you lack is a record of how that happened, so every month you rebuild the picture from memory: who was waiting on a price, who said "next quarter", who you meant to call back. Memory works with six open conversations. Somewhere past fifteen it starts to fail, quietly. Nobody is rude to you; a few people simply stop hearing from you, and you never learn which ones.
A pipeline fixes that, and it is smaller than the word suggests: a list of the deals you are working, each in one named stage, each with a next step and a date. You can build the first version in an afternoon. This article gives the build order: what to decide first, how to find your stages, what rules to attach, which few numbers are worth computing, and how to review it all in half an hour a week.
It is written for someone starting from zero, in any industry. The examples come from two invented businesses, a small pressure-washing and exterior-cleaning company and a commercial security-systems installer, chosen because their sales look so different. Every company, count and dollar figure in them is hypothetical, and none is a benchmark. Whether you need dedicated software at all is a separate question, covered in When Is It Time for Your Small Business to Get a CRM?. This piece comes first, because you need something to put in the software.
Decide what a deal is
Every later rule depends on this, so do it before drawing any stages. A deal is one thing the buyer could say yes to. Write that sentence for your business and keep it short.
For some businesses it is obvious: one quote for one job. For others it is not. If a property manager has six buildings and you propose a contract covering all six, that is one deal, not six. If you propose each building separately and they could accept any subset, that is six. If a customer buys a one-time job and might later buy a maintenance plan, the second purchase is a new deal with the same company, not a stage of the first.
Two tests catch most mistakes. A deal should be able to end in a clear yes or no. And a deal should have one amount on it, even if the amount is an estimate. If a row has no clear ending or no coherent amount, it is a relationship, and relationships belong on a prospect list or in your contacts, not in the pipeline.
Keep the company and the deal separate in your head. One company can have several deals over the years, and you will want to see the company's history. For the first version, though, a row per deal with the company name on it is enough.
Find the real steps your buyers take
The usual way to build stages is to copy a template: Lead, Qualified, Proposal, Negotiation, Closed. Those words describe a generic seller. Your pipeline should describe your buyers, and you can find out what they do without much theory.
The U.S. Small Business Administration's guidance on marketing and sales plans tells owners to describe how they will sell to customers and to explain each step the customer takes once they decide to buy. Our view is that this is the whole method, provided you use evidence instead of intentions.
Method one: reconstruct your last three to five sales. Take your most recent wins. For each, write the sequence of things that happened between the first contact and the signed job, with dates. Not what you planned, what occurred: an inbound call, a site visit, a quote emailed, a question about insurance, a second visit with the owner, a deposit. Put the sequences side by side. Steps that appear in most of them are candidates for stages. Steps that appear once are the customer's quirks.
Method two: ask. When a customer has just said yes, ask what they did before choosing you. Who else did they talk to, what did they need to see, who had to approve it? An owner who asks every new customer three questions will have a dozen answers within a few months. If you have no recent sales to reconstruct, ask people who recently bought something similar from someone else. Our companion piece on finding your first 10 B2B customers covers how to get those first conversations.
Do not ignore lost deals. Look at the last few quotes that went nowhere and ask where each one stopped. That is where your stages need sharper rules.
Name stages after what the buyer has done
Here is the one idea that separates a working pipeline from a mood board. A stage name should describe something the buyer did or something that now exists, never a feeling you have about the deal.
Compare two lists. "Hot, Warm, Cold" is what the seller feels. "Surveyed, Proposal received, Reviewing with partner" is what happened. The second list can be checked by anyone looking at the record, including you six weeks from now when you have forgotten the details. It also resists the most common failure of small pipelines: deals sitting in "Warm" for months because it is comfortable and costs nothing.
A practical test: for each candidate stage, finish the sentence "The buyer has ..." If you can, it is a stage. "The buyer has seen our price in writing." "The buyer has let us into the building." "The buyer has told us who else must approve." If the sentence has to start with "I think" or "it feels like", rewrite it.
Aim for four to seven open stages. Fewer than four hides information, because you cannot tell a deal that has had one chat from a deal that has a proposal out. More than seven means the stages differ by details nobody can remember, and the pipeline stops being updated. Add two exits that are not stages: won, and closed without a sale. Closed deals leave the working view, and each closed-lost deal gets a reason.
Two examples side by side
The two businesses sell differently, so their stages differ in number, in how long each takes and in who is involved. Treat both tables as illustrations, not templates to copy.
Hypothetical business A: a pressure-washing and exterior-cleaning company selling driveways, siding and small storefront jobs. Sales usually close in days. One person decides.
| Stage | Entry rule: the buyer has... | Exit rule |
|---|---|---|
| Asked for a price | Contacted us or answered us with a specific job in mind (address and surface) | A photo, measurement or visit is arranged |
| Job seen | Shown us the surface in person or by photos we can price from | Written price sent |
| Price sent | Received the price in writing, dated | A yes, a no, or no reply after two follow-ups |
| Booked | Accepted and given a date | Job done and invoice sent |
Four stages, because the sale is short. The follow-up rule is tight: a price that sat for a week has probably been forgotten or lost to a cheaper quote.
Hypothetical business B: a commercial security-systems installer selling access control and camera systems to offices, clinics and small warehouses. Sales take weeks or months, several people weigh in and the contract has a long life.
| Stage | Entry rule: the buyer has... | Exit rule |
|---|---|---|
| First conversation | Talked with us and described a real problem (a break-in, an audit finding, a lease move) | Agreed to a site survey with a date |
| Site surveyed | Let us walk the building; our survey notes are saved | Written proposal sent |
| Proposal received | Got the proposal in writing; we logged who received it | A meeting with the full decision group is held |
| Decision group engaged | Met with everyone who must approve, or told us who that is and why we cannot meet them | Signed agreement, written no, or the stale rule |
| Contract in review | Sent the agreement to their lawyer or finance lead; a decision date exists | Signature or written decline |
Notice the differences. Business A has no "who decides" stage because one person does. Business B has two stages purely about the approval path, because that is where its deals have stalled before. Neither business uses the word "qualified", and neither describes a deal by how likely it feels.
Write an entry and an exit rule for every stage
The tables above show the shape. For each of your stages, write two lines.
- Entry: what must be true, and where you will record the evidence. A date and a few words are enough.
- Exit: the specific event that moves the deal forward, plus what ends it if that event does not come.
Two rules keep these honest. A deal moves forward only when its exit event happens, and you enter the date of the event, not the date you got around to typing it. And a deal can move backward when you learn the earlier stage was recorded wrongly, but that move needs a note.
Writing exits is also how you discover which stages are missing. If a deal sits in "Proposal received" and you cannot write what has to happen next, you probably need a stage between that and "Signed". The installer above added "Decision group engaged" after noticing that proposals died without anyone telling it who the real approvers were.
The next-step rule: every open deal has an action and a date
This is the cheapest rule in the article and the one that does most of the work. Every open deal carries a next step, written as a specific action, and a calendar date on which it will happen. "Follow up" does not count. "Call Dana to ask whether the board has seen the proposal, Thursday December 10" does.
Three reasons it matters. First, it turns the pipeline from a record of the past into a list of what to do this week. Second, a deal with no next step has no owner and no future, and you need to find out which. Third, the date makes silence visible: when a date passes with no action, the deal has been neglected, and the log shows you before the buyer decides you have.
The date should usually be set by the buyer's last statement or by a rule you have chosen. If the buyer said "we decide at the board meeting on the 14th", the next step is a message on the 15th. If they said nothing, your own rule fills in: two business days after sending a price for a short-cycle job, a week after a proposal for a longer one. Pick the interval from how your buyers behave, and adjust it when the evidence says so.
Be careful how you repeat contact. In our view, a few well-spaced follow-ups are a method, while a string of ever-shorter messages is how a prospect ends up asking to be left alone. Email and phone calls are also subject to legal rules, which the FTC publishes guides on; How to Build a Prospect List From Scratch summarizes the relevant points (general information, not legal advice), and Commercial Cleaning Prospecting Templates You Can Actually Use covers follow-up timing and a pre-send checklist. If someone asks not to be contacted, mark the row and keep it visible so nobody contacts them by accident.
Closing a deal as lost, and why that column matters
A pipeline that only moves forward fills with dead deals. Give every deal that ends without a sale a closed-lost status and require one reason from a short fixed list. The list is yours; a starting set could be:
- Chose a competitor (and, if you learned it, the reason)
- Price
- No decision, project cancelled or postponed
- No response after our follow-up limit
- Not a fit, outside our service or area
- Bought something else or did it in-house
- Timing, revisit on a date
The reason has to be a choice from the list, with a free-text note beside it. Free text alone gets filled in with "n/a". A fixed list can be counted, and counting is the point: after twenty or thirty closed deals, the reasons start to say something. A heavy share of "no response" suggests a problem in how you open or follow up. A heavy share of "price" might mean your proposals reach people who cannot approve spend, or that you are quoting before the buyer understands the value. Nobody can tell which from a single deal, and with small counts you should treat the pattern as a prompt to read the individual cases.
Separate "lost" from "not now". If a buyer says the budget resets in July, that is not a loss. It is a parked deal with a review date. The SBA's small-business advice on getting customers suggests reaching out to dormant customers after six months or a year without an interaction or purchase. Your own interval might be shorter or longer, but the principle holds: a parked deal returns on a date you chose, not when you happen to remember.
The numbers worth tracking
You do not need a dashboard. Four numbers are enough at first, and all of them come from the fields you already keep: stage, dates, amounts, outcome. The examples below use hypothetical data, and the arithmetic is shown so you can reproduce it on your own records.
Stage-to-stage conversion
Conversion is the share of deals that reached a stage which then reached the next one. It shows where deals leak.
The weakest step by percentage is the last, but look at the denominator: eight proposals. If a fourth had been signed, the figure would be 4 / 8, or 50%, a jump of 12.5 points from one deal. Small samples make rates unstable. With counts like these, the percentage is a reason to read each lost proposal, not a measurement you can plan around. We did not find a primary source that gives typical conversion rates for businesses like these, and we do not believe any single number would apply across industries, so we give none. Your own history is the benchmark that matters, and it needs time to accumulate.
Average sales cycle
The sales cycle is the number of days from the first real conversation to the signature, measured on won deals. Say the installer's five signed deals over the period took 41, 58, 63, 72 and 96 days.
Sum: 41 + 58 + 63 + 72 + 96 = 330. Average: 330 / 5 = 66 days. The middle value is 63 days. When the two differ a lot, or when one deal is far longer than the rest, report the middle value as well; here they are close, so the picture is stable, but with five deals it still could shift with the next one.
What do you do with it? Use it to set expectations about timing. A proposal sent 20 days before the end of the quarter, to a deal that is already 50 days old, falls inside a plausible window. One sent to a deal that began last week is unlikely to close this quarter, however good the conversation felt. For short-cycle sales, measure in days; the pressure-washing company might find its price-to-booking time is three or four days, and a price that is eight days old with no reply is in a different category from one that is a day old.
Count only won deals for the cycle, and also look at how long lost deals sat before you closed them. A pipeline where lost deals linger for 90 days is carrying dead weight.
Pipeline coverage
Coverage compares what is open with what you need. If you want $120,000 in new annual contract value this quarter and you have signed $40,000 so far, the remaining target is $80,000. Suppose the open deals, using the amounts on their rows, total $240,000. Coverage is $240,000 / $80,000 = 3.0.
That ratio alone is easy to misread, so adjust it in two ways. First, apply your own win rate. The installer's proposal-to-signed rate above was 37.5%, so if every open deal were at the proposal stage, $240,000 x 0.375 = $90,000 might be expected, only just over the target. If many are earlier-stage deals, the expected figure is lower, because each still has steps to survive. Second, remove deals that cannot close in time, using the sales cycle: a $60,000 deal that started eight days ago should come out of this quarter's count. If it leaves, open value drops to $180,000 and coverage to $180,000 / $80,000 = 2.25.
Be plain with yourself about what this is. It is a rough sanity check built on a win rate from eight proposals. It tells you whether you have enough in the pipeline to make the target plausible, or whether you should spend the next two weeks finding more deals. It does not forecast revenue.
What to leave out for now
Skip weighted probabilities per stage ("negotiation is 60%"), because you have no data to set them. Skip lead scores, activity counts and a count of calls made, until you know that calls are what limits you. Add measures when a specific question comes up, not before.
The weekly review
A pipeline that is read weekly stays accurate. One read quarterly describes the past. The review should take thirty minutes and have the same agenda each week, so it becomes routine. Here is one, written as a template you can paste into a calendar invite or a notes file.
Minutes 0-5: Overdue next steps - Sort by next-step date. Anything past its date gets a decision: do it today, change the plan, or close the deal. Do not just push the date. Minutes 5-10: Missing next steps - Find open deals with no next step or no date. Give each one an action and a date, or close it with a reason. Minutes 10-15: Stage truth check - For anything that happened since last week and is not recorded, update the stage and enter the date it happened. - Move back any deal whose stage was recorded early. Minutes 15-20: Slow deals - List deals past your normal time in their stage (use your average cycle as a guide). For each: push, park with a review date, or close. Minutes 20-25: Numbers - Open deals by stage, deals signed and lost this week with reasons, coverage against this quarter's target. - Write one sentence on what changed. Minutes 25-30: This week's plan - Choose the three highest-value next steps and put them on the calendar. - Check how many new deals entered the first stage. If the number is low, schedule time for the top of the funnel.
The last item is a reminder that a pipeline only measures what enters it. If nothing is entering, no stage rule will help, and the fix is upstream. How to Build a Prospect List From Scratch is the method for that.
Set it up in a spreadsheet today
You do not need software for any of this. A spreadsheet with one row per deal and the following columns is enough for the first months.
Deal ID (a number) Company Contact name and role What the deal is (one line: the job, contract or system) Amount (and whether it is one-time or per year) Stage Date entered this stage First conversation date Next step (a specific action) Next step date Who else decides (name and role, or "unknown") Source (referral, inbound, outreach, other) Last real contact (date and what happened) Status (open, won, lost, parked) Closed date Closed-lost reason (pick from your list) Review date (for parked deals) Do not contact (yes or no)
Set up three things so the sheet keeps its shape. Make Stage and Status dropdown lists, so the names never drift ("Proposal", "proposal sent" and "Quote out" all mean different things in a pivot table). Freeze the header row and sort by Next step date, so the top of the sheet is always this week's work. And add one extra tab with the stage definitions (entry and exit rules) so that anyone else who touches the sheet reads the same rules.
From these columns you can compute every number in this article. Days in the current stage is today minus Date entered this stage. Sales cycle is Closed date minus First conversation date for won deals. Conversion needs a count of deals that ever reached each stage, so keep a simple stage-history tab with one line per stage entry (Deal ID, stage, date), or record the date each stage was reached in extra columns. Either way, the records have to be entered at the time, because reconstructed dates make every figure above unreliable.
Keep the sheet to a single place with one person responsible for updating it. Two copies drift apart within a week.
When to move to software
The spreadsheet is a good start and a poor home once the work grows. Signs that you are ready to move:
- More than one person updates it, and you keep finding conflicting edits or stale copies.
- You have more open deals than you can scan in the weekly review.
- Follow-ups slip because the date lives in a column and nothing reminds you of it.
- You need to see a company's whole history, its people, past deals and every message, and the sheet makes you hunt for it.
- You keep adding columns, tabs and formulas to imitate what a tool would do.
Not having reached these signs is a fine reason to wait. A tool imposes its own structure, and if you have not yet decided what a deal is or what your stages mean, software will only record your indecision more neatly. The stage definitions, entry and exit rules and next-step discipline above are what carry over: write them first and they will move to any tool. When Is It Time for Your Small Business to Get a CRM? goes through the decision in more detail.
After a sale closes, keep the row useful. The SBA's guidance on new customers advises building referral-generating activity into the sales process, and the page it appears on gives the example of following up after delivery to ask for a referral. Two post-sale next steps are worth adding to every won deal: a check-in after the first delivery of the work, and a question about who else they know with the same problem. How to Retain Commercial Cleaning Clients is written for one industry, and its follow-up habits carry over.
Where software fits
Good Loux publishes this Journal and sells business development and CRM software, so weigh what follows accordingly. Everything above works in a spreadsheet. Software helps once the rows and follow-ups outgrow your calendar and memory.
Whatever holds it, the discipline is the same: one stage per deal, a move only on an event, and a dated next step on every open row.
Your first pipeline in one week
A schedule that fits around other work:
- Day 1: Write your sentence for what a deal is. List your open conversations, however rough.
- Day 2: Reconstruct your last three to five wins step by step, with dates. Ask your most recent customer how they decided.
- Day 3: Draft four to seven stages named for what the buyer has done. Run the "The buyer has..." test on each.
- Day 4: Write the entry and exit rule for each stage, and the closed-lost reason list.
- Day 5: Build the spreadsheet with the column list above and enter every open deal, each with a next step and a date.
- Day 6: Set your follow-up intervals and how many attempts you make before parking or closing a deal.
- Day 7: Run the first thirty-minute review. Fix whatever the rules did not anticipate, and revise the stages if two deals do not fit anywhere.
Expect to change the stages after a month. That is normal and does not mean the first version was wrong: it was your best reading of your own history, and now you have a pipeline that can show you where it is mistaken.
Sources
We read these directly on November 19, 2026. Pages change, so check the originals.
- U.S. Small Business Administration: Market your business and sales plan for the instruction to describe how you will sell and the steps a customer takes once they decide to buy, and for the suggestion to compare sales and marketing costs with the revenue they generate. The page shows no publication date and gives no step-by-step customer journey of its own.
- U.S. Small Business Administration: 10 Ways to Get New Customers (published April 6, 2017) for the advice to build referral activity into the sales process, to ask for a referral after delivery, and to reach out to dormant customers after six months or a year without interaction.
No figure for typical win rates or sales-cycle lengths appears in this article. The two invented businesses, every count, day and dollar amount, the stage tables and the example thresholds are our own illustrations, not research findings or benchmarks. The build order, the rules and the advice to name stages after buyer actions are our own view, informed by the SBA wording above. The description of Good Loux comes from the product itself. Nothing here is a claim about any real company.


