Understanding AI deal scoring: the four factors behind the number

The AI Deal Score in Advanzo is a number from 0 to 100 that estimates an open deal's chance of closing – calculated from four equally weighted factors: engagement, momentum, timing and size. The number does not replace judgement, but it makes visible which deals deserve attention and which are quietly at risk.

This article explains how the number comes about, what the colour bands mean, where the honest limits of scoring lie – and how to use the score day to day without trusting it blindly. It is the look behind the concrete number that general deal-scoring guides usually skip.

What is the AI Deal Score in Advanzo?

The AI Deal Score is an AI assessment of an open deal's closing chance, shown as a number from 0 to 100. You see it directly in the deal header and as a badge on the Kanban card – so the pipeline shows at a glance where things are going well and where they are not.

Important for context: the score rates deals, not contacts. The question is not «does this lead fit us?» – that is lead scoring – but «how likely is this specific deal to close?». It is answered from what is visible in the CRM: activities, email history, meetings, deal data.

For SME teams without a sales operations role this is especially valuable: nobody has to build reports or maintain spreadsheets to see the health of the pipeline. The assessment sits where the work happens – on the deal itself and on the board.

Which four factors sit behind the number?

The score is composed of four equally weighted factors – none dominates, and each tells a different part of the story:

• Engagement: how involved is the customer? Are they replying to emails, asking questions, showing up to meetings? A deal where only you are active is weaker than it looks.
• Momentum: is the deal moving? Stage changes, fresh activities and shortening response times speak for momentum – standstill, by contrast, is one of the most reliable warning signs there is.
• Timing: how does the timeline look? A realistic, near-term close date supports the score; a missed or endlessly postponed date drags it down.
• Size: how does the deal value compare with your typical business? Unusually large deals tend to close less often and more slowly than deals in your usual range.

The value of the breakdown is diagnostic: two deals scoring 55 can have completely different problems – one lacks engagement, the other's timing is slipping. If you know the factors, you know not just that a deal is weakening but where to intervene.

What do the green, yellow and red bands mean?

Advanzo translates the number into three bands: green covers 70 to 100 – the deal is on track. Yellow covers 40 to 69 – there are weaknesses that deserve attention. Red covers 0 to 39 – the deal is seriously at risk or has effectively gone to sleep.

The bands are deliberately coarse. Their purpose is not precision but at-a-glance prioritisation: on the Kanban board, the red badges catch the eye immediately, and a weekly review sensibly starts exactly there. The difference between 62 and 68, on the other hand, should not be over-interpreted – an AI estimate is not built for that.

How current is the score?

The score is a snapshot on demand: it is calculated when you request it and does not continuously update itself. If the situation changes – a new reply arrives, a meeting falls through – it is worth running a fresh calculation before basing decisions on it.

Honesty about precision belongs here too: between two runs the value can vary by a few points, even without major changes to the data. That is the nature of AI estimates. So treat the score as an order of magnitude with a direction – «solidly green», «deep yellow», «freshly tipped to red» – and not as an exact measurement to the point.

What can the score not do?

Three limits are worth knowing before you base decisions on the number. First: the score only rates open deals – won and lost ones are out of scope. Second: it does not train on your historical win/loss record; it judges the visible signals of the individual deal, not the patterns of your past. Third: it only knows the CRM. The good conversation at the trade fair, the customer's internal budget approval – whatever is not recorded does not flow in.

From this follows the most important rule of use: the score is a starting point for your judgement, not a substitute for it. A low score on a deal you rate highly for good reasons is a prompt to record the missing information – not to write the deal off.

How do you use the score in daily sales work?

The score works best as a prioritisation and diagnosis tool inside fixed routines. Day to day: before allocating your time, glance at the badges on the board – red and freshly tipped deals first. In the weekly review: walk through the yellow band and ask, per deal, which factor is weakening – missing engagement calls for a different action than slipping timing.

The interplay with Autopilot is also strong: while the score measures the state on demand, Autopilot speaks up by itself with at-risk flags and concrete proposals – say, a follow-up task for the deal that has been silent for two weeks. Score and copilot complement each other: one shows where it is burning, the other proposes what to do. Both are part of Advanzo's AI features, available with the AI add-on at CHF 9 per user/month – details on the pricing page.

How do you improve a weak score deliberately?

By working on the weakest factor rather than the symptom. If engagement is missing, another standard follow-up will not help – a new occasion will: a concrete question, a genuinely relevant piece of content, a proposal with two meeting slots. If momentum is missing, the deal needs a defined next step with a date – vague «we will stay in touch» deals are exactly the ones that die quietly.

If timing is slipping, the most honest measure is often an open conversation about the real timeline – a postponed but realistic close date is worth more than an optimistic one the score rightly punishes. And on the size factor, structure sometimes helps: an unusually large deal can be split into stages that are individually faster to decide.

A second lever is mundane but effective: data upkeep. Log the phone call, link the emails, update the close date – often the score rises simply because the deal's reality is now visible. A score based on complete data is the precondition for work on the real weaknesses to pay off.

Frequently asked questions (FAQ)

How is the AI Deal Score calculated?

From four equally weighted factors: engagement (customer involvement), momentum (movement in the deal), timing (realism of the timeline) and size (value relative to your typical business). The result is a number from 0 to 100.

What does a red deal score mean?

Red covers values from 0 to 39: the deal is seriously at risk – usually due to missing customer response, long standstill or a missed close date. Red does not mean lost; it means act now or let go consciously.

Does the score update automatically?

No, the score is a snapshot on demand. It is recalculated when requested and can vary by a few points between runs. Before important decisions, a fresh calculation is worth it.

Does the score learn from my won and lost deals?

No. The score does not train on your historical win/loss record; it assesses the visible signals of the individual open deal. Your historical data stays untouched – and your judgement stays essential.

What is the difference between deal scoring and lead scoring?

Lead scoring rates how well a contact fits your ideal customer profile – before the deal. Deal scoring rates the closing chance of a specific open deal – mid-process. Advanzo has both: lead assessment via Autopilot, and the deal score on the deal itself.

Can I trust the score if my data is patchy?

Only partly. The score knows nothing beyond the CRM data – if activities and emails are missing, a healthy deal quickly looks endangered. The cleaner the upkeep, the more meaningful the number. Patchy data is the most common cause of misleading scores.

Start Advanzo for free and see your pipeline with deal scores: four factors, clear colour bands, honest limits – with Swiss hosting and no credit card.