Two fundamentals you must get right before AI can improve your funnel

AI can absolutely create commercial value in B2B, but only if the foundation can support it.

And here’s the part nobody gets excited about:

• Data quality isn’t sexy.
• Signal design isn’t glamorous.
• But they decide 80% of the value AI can produce.

AI today can work with more imperfect data than traditional analytics ever did.

But the quality of the underlying data still determines:

• how early you see churn risk
• how accurately you detect cross-sell potential
• how confident you can be about new business intent
• how well marketing and sales actions can actually be guided

In other words:

• The data defines the quality of your signals.
• The signals define the quality of your decisions.
• And those decisions define your revenue.

Let’s break this down into two fundamentals.

Data quality: the boring reality that decides everything

Nobody celebrates data cleaning.

Nobody high-fives over fixing CRM fields.

But the truth is simple:

AI can’t compensate for chaos; it can only navigate it.

Most B2B organisations share the same pain points:

• inconsistent purchase records
• missing titles, industries or account attributes
• duplicated contacts
• unstructured sales notes
• usage data stored separately from commercial data
• no shared definition of what “a healthy customer” even looks like

Of course, AI can still operate on imperfect data, but the commercial impact scales with the quality of that foundation.

That’s why data work is not just cleaning; it’s a three-step process:

1. Validation: Is the data usable? What’s missing? Are the core signals even detectable?
2. Diagnosis: where are the gaps? Which fields break segmentation? What confuses the model?
3. Clean-up & Enrichment: Fix duplicates, harmonise structures, fill in blanks, enrich with external context.

When this is done well, your organisation unlocks:

• clear customer segments
• predictable buying behaviour models (e.g. RFM)
• visibility into deviations that matter commercially
• a stable base for AI to actually generate reliable insights

For most, it’s not exciting work. But it’s the work that makes everything else possible.

The signal layer: how you actually guide marketing and sales

Once the data foundation is stable, the real value comes from designing the signals that guide decision-making.

Because the purpose of AI in B2B is not to “explain the past”, it’s to direct the next action in the marketing and sales funnel.

And the right signals depend entirely on which commercial goal is most important right now:

Churn signals: detecting behavioural deviation early

As I explained in a previous article on the areas where AI delivers immediate, measurable impact in B2B, churn is simply a customer drifting away from their segment’s expected buying behaviour.

Signal examples:

• extending purchase intervals
• declining product usage
• unusually quiet decision-makers
• behaviour that no longer matches their segment “norms”

If marketing and sales want to act early, they need reliable early warnings.

That is exactly what signals provide.

These signals decide:

• which accounts get proactive outreach
• what message is used (retention, activation, education, expansion)
• whether the issue is usage, value, or changing needs

Cross-sell signals: identifying hidden revenue moments

Cross-sell is B2B’s version of bundling, not as blunt as “add fries for one dollar,”
but structurally the same idea:

Offer the right complementary value at the right moment.

Signals include:

• behaviour similar to customers who historically expanded
• usage patterns that indicate readiness for the next product
• feature adoption curves that align with typical upsell timing
• consumption thresholds that trigger natural expansion conversations

Cross-sell is not a pitch.

It’s recognising when the customer’s behaviour indicates increasing readiness and aligning your commercial motion accordingly.

New business signals: identifying when and where the market is moving

New business is the most expensive growth lever which is why precision matters the most here.

Signals include:

• lookalike accounts that match your best customer segments
• activity spikes from domains with similar behaviour to past wins
• content engagement that reflects early-stage research
• segment-level movements that suggest growing interest

These signals guide:

• which accounts sales should prioritise
• which messages should be used
• where marketing should allocate budget
• when to reach out: timing is often the difference between “too early” and “too late”

Again: Signals are not the insight; they are the action trigger.

Data & signals are the operating system of your go-to-market

You cannot guide a funnel you cannot read.

• Data quality determines what you can observe.
• Signals determine what you should act on.
• Actions determine whether you move revenue forward or not.

To summarise: churn, cross-sell, and new business are all behavioural phenomena, and the above-mentioned signals let you see them early enough to act. And to detect relevant signals, you need qualitative data.

At the end of the day, AI will amplify whatever you feed it.

• Give it noise, and you’ll get noise faster.
• Give it structure, and you’ll get clarity that compounds.