Sales live in their heads: how to get ONE number (without buying a CRM at €400 a user and leaving it empty)

The forecast that is a feeling

You run the Monday sales meeting and you ask: «how are we looking for the quarter?». The sales director gives you a number. But if you ask where that number comes from, the real answer is: from a sensation. A few deals that «are looking good», a few customers who «said they’ll buy», a bit of optimism because the team is pumped. Your company’s forecast isn’t a calculation: it’s a mood. And moods, by definition, get it wrong — usually on the high side, because no salesperson wants to be the one who brings bad news into the meeting.

The reason is simple: the pipeline lives in the salespeople’s heads, not in one place. Each has their Excel, their calendar, their notes, their memory. One knows that deal has been stuck for three weeks, but only they know. Another has a red-hot customer they haven’t told anyone about. The director puts together the pieces that reach them and adds their gut, and out comes a number nobody can verify. When the quarter closes under, everyone is surprised — but the datum that would have prevented the surprise wasn’t there.

This article is for anyone who runs a sales team and wants to stop navigating by feeling. The problem isn’t “buying a CRM” — in fact, you’ve almost always already bought one and it’s sitting there empty. The problem is having a single sales pipeline: one place where the real state of deals lives, actually fed by the salespeople, that produces a number you can trust. Let’s see why CRMs stay empty (that’s the real enemy), what a salesperson has to be able to enter in 40 seconds for them to use it, and how you build the single number — the definition of “won” you don’t have today.

Why enterprise CRMs stay empty

Let’s name names, you already know: you bought (or you evaluated) a famous CRM, one of those at hundreds of euro per user per month. Pretty, powerful, full of features. And after six months it’s empty: salespeople don’t put data in, or they put it in at random every so often, and the forecast is a feeling again — only now you also pay the licences. Why does it happen? It isn’t the “lazy” salespeople’s fault. It’s a misunderstanding about what’s needed.

An enterprise CRM is built to be configured by an operations department, with rigid processes, lots of mandatory fields, reports for management. It’s designed from the point of view of whoever looks at the data, not whoever enters it. For the salesperson, it becomes a second job: after making the call, they have to open a screen with twenty fields, pick the right value, fill in things they don’t need, and all this while they already have the next call in their head. They do the most rational thing in the world: they postpone. And “I’ll do it later” becomes “I don’t do it”, and the CRM stays empty.

There’s a law here that holds for every company software: if the tool is slower than the current method, people go around it. The salesperson already has a method that works for them — the head, Excel, the calendar. If the CRM costs them more time than it saves, they go back to their method, “because it’s faster”. It’s the exact same mechanism by which badly made internal apps get abandoned and people go back to the spreadsheet: it isn’t resistance to change, it’s that the change on offer makes the day worse for whoever is supposed to adopt it. Empty software is the real enemy, and it always comes from the same error: thinking about the management that looks instead of the salesperson who enters.

What a salesperson has to enter in 40 seconds

Let’s flip the problem and start from whoever has to feed the pipeline. A CRM that salespeople actually use starts from a single question: what can a salesperson enter in 40 seconds, right after a contact, maybe from the phone in the car? If the answer to “update a deal” takes more than 40 seconds and more than two or three taps, you’ve already lost.

The 40-second screen contains only the essential, that is the few things that actually tell you where a deal stands:

• Who: the customer/contact (already in the list, not to retype).
• How much: the estimated value of the deal (a number, even approximate).
• Where it stands: the stage (first contact, quote sent, in negotiation, won, lost) — picked from a short list, not a menu of twenty items.
• When: the likely close date (it’s a hypothesis, and that’s fine).
• The next action: what I have to do next time and when (call back on the 15th, send the quote, etc.).

Five things, three taps, forty seconds. Note what is NOT there: the twenty fields management needs but the salesperson doesn’t. Those, if they really are needed, you derive or you ask later, you don’t put them in the path of whoever has to enter fast. The rule is make trivial what you do a hundred times (update the stage, mark the next action) and possible what you do rarely — never the opposite.

And above all, the screen has to work from mobile, because the salesperson lives outside, not in front of a computer. A CRM that updates well only from desktop is a CRM that gets updated in the evening (badly, in a hurry) or never. This is the piece enterprise systems do worst: 1990s screens designed for the big monitor, unusable with a thumb as you leave the customer.

The screen vs the 1990s screens

It’s worth insisting on the difference, because it’s all there. The 1990s screen — the one of so many ERPs and traditional CRMs — is a dense grid of fields, technical labels, endless dropdowns, tiny buttons. It’s designed for a seated clerk who fills in calmly, not for a salesperson on their feet who has thirty seconds. Putting it in a sales team’s hands is like asking a runner to tie their shoes with ski gloves: technically possible, practically no.

The right screen is the opposite: large, simple, few elements per screen, optimised for touch. The stage changes with a swipe or a tap, not by picking it from a menu of fifteen items. The next action is marked in a tap. The value is typed on a large numeric keypad. It’s designed on the real gesture — “I’m leaving the customer, in thirty seconds I update” — not on theoretical completeness. This is the interface design part that decides adoption, and that no off-the-shelf CRM gives you tuned to your salespeople: it gives it to you generic, designed for everyone and therefore for no one.

It’s the same principle of five screens versus fifty that I talk about for internal apps people use only if they remove friction. A CRM is no exception: it’s an internal app used by people who have better things to do than fill in forms, and it lives or dies on the same rule.

The single number: the definition of «won»

Now the piece that turns a fed pipeline into a forecast you trust: the single number, and under it, a shared definition. Because the feeling-forecast problem isn’t only that the data lives in different heads: it’s that everyone has a different definition of what “almost closed” means.

For one salesperson, a deal is “won” when the customer says «we’re there». For another, when the signed order arrives. For the director, when the money comes in. Those are three different moments, weeks apart, and until it’s written which one counts, the forecast adds apples and pears. The condition for the pipeline to produce a real number is write and sign the definitions: what “won” means (signed order? countersigned contract? first payment?), what each stage means, when a deal is considered “lost” instead of “eternal”. From that moment there is one number, and whoever disagrees argues about the definition, not the chart.

This is exactly the same work that sits under a company dashboard the owner trusts: without shared, signed definitions, the prettiest dashboard in the world is only a more expensive way to argue about the numbers. The value of a single pipeline isn’t “having software”: it’s having one truth about the state of deals, that lets you say “we have €400,000 of offers at an advanced stage closing within the quarter” and know what that means, because everyone means the same thing by “advanced stage”.

The bill: the empty CRM costs twice

Let’s put some numbers, because the empty CRM has a double cost that rarely gets looked at. On one side you pay the licences: an enterprise CRM at €300–400 per user per month, for a team of 10 salespeople, is €36,000–48,000 a year — for a system that, if it stays empty, produces nothing. On the other, and this is worse, you keep paying the cost of the wrong forecast: decisions taken on a mood (do I hire? do I invest? do I promise the customer a delivery?), deals that die because nobody called them back in time, leads that get lost in the handoff between marketing and sales.

Item	Estimate
Enterprise CRM licences (10 users × ~€350/month)	~€42,000/year
If the CRM is empty	that cost produces ~zero
Leads lost to missed follow-up (“forgotten” deals)	often the largest item
Bad decisions on a feeling-forecast	hard to see, very real

The biggest hidden item is leads that die. A contact that arrived and wasn’t called back fast, a deal left stuck because nobody remembered it, a hot customer cooled by the wait: that’s revenue you already had in your hand and you lost for lack of a system that said “this one gets called today”. A single pipeline with next actions well kept recovers exactly this: no deal falls into the void, because the system remembers what the head forgets. And it’s almost always there, not in the licences, that the real return sits.

AI: where it cleans notes, not where it invents the close date

The question today always arrives: and artificial intelligence in the CRM? The honest answer distinguishes two worlds. Where AI actually helps: in taking writing work off the salesperson. Transcribe and summarise the notes of a call (the salesperson dictates two sentences, the AI puts them in order), suggest the next action based on history, prepare the draft of the follow-up email, clean and enrich the record. Anything that reduces the 40 seconds to 20 is fine, because it works in favour of adoption.

Where AI must not stick its nose: in inventing the forecast. A model that “estimates the probability of close” or “predicts the close date” produces numbers that look precise and are hot air, because they rest on data that — in a pipeline just born — is scarce and dirty. Worse: they give management the illusion of an objective forecast when it’s an automatic prediction on unreliable input. The forecast is built with real data entered by salespeople and shared definitions, not with an AI that guesses the close date. The rule is the usual one: AI where it takes work away (notes, drafts, suggestions), the person and the rules where you need truth (the real stage, the definition of won). Whoever sells you “the CRM with AI that predicts sales” is selling you a more expensive feeling, not a datum.

Leadership: a dashboard that doesn’t humiliate the team

There’s a political aspect of adoption few consider and that sinks more CRMs than any technical problem: how leadership uses the data. If the salesperson understands that everything they enter becomes a stick to beat them with in the meeting — “I see you only have three open deals, how come?” — they stop entering, or they only enter the things that make them look good. The CRM becomes theatre, and the forecast goes back to lying.

The leadership dashboard has to serve to understand and help, not to humiliate. It serves to see where the pipeline is weak (few deals at the top of the funnel = a lead-generation problem, not laziness), which deals risk dying from missed follow-up (so you intervene, you don’t punish), where the process jams. Used that way, salespeople understand that feeding the CRM is in their interest — because it helps them not lose deals and get help when they need it — and they feed it. Used as a police control tool, they sabotage it. The dashboard for sales leadership is powerful only if the team trusts how it will be used; otherwise it’s a weapon pointed at whoever is supposed to feed it, and nobody loads the weapon that will be fired at them.

It isn’t a tool change, it’s a process change

And here we are at the heart of why so many CRM projects fail even when the tool is good: you think you’re buying a tool, and instead you’re changing a process. Putting in a single pipeline isn’t installing software: it’s deciding how the sales team works — what gets entered, when, with which definitions, how leadership uses the data, what happens to a deal at each stage. If you change the tool but not the process, the tool stays empty.

That means the project has to be done with the salespeople, not dropped from above. You look at how they work today, you understand what they actually need, you design the 40-second screen on their real gesture, you agree the definitions. And you budget a running-in period: the first two weeks the CRM feels like extra weight, then — if it’s done well, if it saves them time and they don’t lose deals — it becomes the tool they won’t drop. As with every app that has to be adopted by people who work, adoption is designed by making the tool an advantage for whoever uses it, not a compliance task. A CRM imposed gets circumvented; one built with the team gets defended.

A typical case: from feeling to a number

A typical profile, architectural, no names. A company with a sales team and a director who every Monday presented a forecast. They had bought, a couple of years earlier, an important CRM: expensive, powerful, and essentially empty. Salespeople kept their deals in personal Excel files and in their heads; the director collected what they told him and added gut. The quarter regularly closed different from the forecast, always short, and nobody really knew why. Every so often it emerged that an important deal had died weeks earlier because nobody had called it back: it was in the head of a salesperson who in the meantime had focused on something else.

What was done. First the uncomfortable work: sit with the salespeople and leadership and agree the definitions. What “won” means (they chose: signed order), what each stage means, when a deal is declared lost instead of staying eternal. Then the 40-second screen was designed with the salespeople, on their real gesture — update from the phone leaving the customer — throwing away the fields that only served management. Five things, three taps. Integration with the ERP, so a won deal became a real order without re-entering anything.

The delicate point was use of the dashboard: it was decided, and communicated to the team, that it would serve to help (see which deals risked dying, where the pipeline was weak) and not to beat people in the meeting. That unlocked adoption: salespeople understood that feeding the system was in their interest, because they no longer lost deals and they got help when they needed it. The first two weeks were grumbling; from the third, the pipeline started filling on its own. At regime, the forecast stopped being a mood and became a number: “this much in offers that can be signed within the quarter”, with everyone meaning the same thing. And deals no longer died in silence, because the system remembered the next actions. The honest note: the salespeople’s talent didn’t change; what they knew no longer lived only in their heads.

Why ready-made stops, and why you need a single pair of hands

Ready-made CRMs — enterprise or cheap — get to 80% and stop exactly on the part that counts: the screen tuned to your salespeople, your stage and “won” definitions, the integration with your ERP (because a won deal has to become a real order, not stay a flag in the CRM), the dashboard that reflects your way of selling. That 20% is exactly what distinguishes a pipeline that gets used from empty software, and no ready-made gives it to you custom.

And here the single-pair-of-hands theme comes back: a pipeline that works is data, interface and integration together. The screen (interface) has to reflect the definitions (data/logic), which have to connect to the ERP (integration). If you split these pieces among whoever “configures the CRM”, whoever “makes the mobile app” and whoever “connects the ERP”, you get three worlds that don’t talk and a system salespeople don’t use. You need someone who holds together the fast screen, the definitions and the hook to real data — not three suppliers who, in front of an empty CRM, blame each other.

Timeline and maintenance

How long does it take? The long part isn’t technical: it’s agreeing the process — the definitions, the stages, what gets entered, how the dashboard is used. Once that’s done, building a focused pipeline (fast screen, mobile, dashboard, ERP integration) is a matter of weeks, not months, precisely because it does few things and does them well. Then a test with a pilot group of salespeople, which brings out the real friction (the missing field, the extra tap) before rolling out to the whole team.

Maintenance has a commercial specificity: the sales process changes (new products, new stages, new commission rules, new markets), and the pipeline has to be updated accordingly. You need to be able to change stages and definitions without rebuilding everything. And you need to watch adoption over time: if after months someone stops updating, it’s a signal that something in the process isn’t working, and it has to be fixed before the CRM goes empty again.

It’s for you if / it isn’t for you if

It’s for you if: you run a sales team and your forecast is in fact a feeling; the pipeline lives in individuals’ heads and Excel files, not in one place; you bought a CRM that stayed empty (or you fear it will); you lose deals to missed follow-up; you want a single number you can trust to decide.

It isn’t for you if: you have very few deals you manage perfectly by voice and you keep them all in your head without losing any (then you don’t need a system); you’re looking for a “complete” CRM with a thousand features more than a tool salespeople actually use (that’s the error that leads to empty software); you aren’t willing to change the process and involve the team — because without that, no CRM gets adopted, however pretty it is.

Frequently asked questions

We already have a CRM but it’s empty: what do we do? It’s the most common situation. The problem isn’t “software is missing”, it’s that what you have isn’t being fed — because it’s slower than the salespeople’s method and/or because leadership uses it to beat people. You start again from the process and the 40-second screen, not from buying another CRM. Sometimes you can save the existing CRM by making it usable; sometimes something simpler and custom is better.

Why isn’t a famous CRM at €400 a user enough? Because it’s designed for whoever looks at the data, not whoever enters it: dense screens, lots of fields, optimised for desktop and management. The salesperson lives it as a second job and goes around it. You pay expensive licences for an empty system. The value isn’t in the CRM’s power, it’s in adoption — and adoption is made by simplicity, not features.

What does a salesperson have to enter, concretely? Five things in 40 seconds, from mobile: who (the customer), how much (estimated value), where it stands (the stage), when (likely close), the next action. No twenty fields. Everything management needs but the salesperson doesn’t is derived or asked later, not put in the path of whoever has to enter fast.

How do I get a reliable forecast? With real data entered by salespeople (adoption) and shared, signed definitions (what “won” means, what each stage means). The single number is born from there, not from an AI that predicts the close date on dirty data. Without common definitions, you add apples and pears and the forecast goes back to lying.

Can AI predict sales? It can help take work away (transcribe notes, suggest actions, write follow-up drafts), and there it’s valuable because it helps adoption. It must not invent the forecast: automatic predictions on young pipelines and dirty data look precise and are hot air, with the aggravating factor of giving an illusion of objectivity. The forecast is made by real data and definitions.

Salespeople won’t use it anyway? They’ll use it if it’s in their interest: if it saves them time, they don’t lose deals, and it doesn’t become a stick in the meeting. A CRM built with the team, with the fast screen and leadership that helps instead of humiliating, gets adopted. One imposed from above and used to control gets sabotaged. Adoption is a design and process choice, not a force choice.

Does it connect to the ERP? It has to. A “won” deal in the CRM has to be able to become a real order in the ERP, without re-entering anything, or the CRM stays an island. Integration is part of the project, not an accessory, and it’s one of the points where ready-made often stops.

How long does it take? The long part is agreeing the process and the definitions; building the actual pipeline (screen, mobile, dashboard, integration) is a matter of weeks because it does few things well. Then a test with pilot salespeople before rolling out. Distrust whoever sells you “the CRM live in a day”: the day after it’s empty again.

In one line

If your forecast is a feeling and the sales pipeline lives in individuals’ heads, the problem isn’t buying a CRM — you probably already have one, empty. The real enemy is empty software, and it comes from an error: thinking about whoever looks at the data instead of the salesperson who enters it. A single pipeline that works has a 40-second screen from mobile, shared signed definitions (what “won” means), leadership that helps instead of humiliating, and integration with the ERP. It’s a process change, not a tool change — and you do it with the salespeople, or it stays yet another licence paid for nothing.

If you want a pipeline your salespeople actually use and a forecast you can trust, look at the projects I’ve built or drop me a line: we start from how your team sells and from what would make them update in 40 seconds, not from a catalogue CRM.