Salesforce World Tour London Takeaways: Fix the Messy Middle in RevOps, Marketing and Field Service

The big theme: disconnected data creates disconnected work

One slide hit hardest because it was so simple:

It also put numbers on it: 1,026 applications in the average enterprise, and 72% of company applications disconnected.

Disconnected data delivers a disconnected experience

You do not need to be technical to feel this. Disconnected work looks like:

• •marketing emails that do not match what sales is doing
• •quoting that depends on someone's spreadsheet
• •service teams arriving without the right info (or the right part)
• •reporting that turns into a monthly panic

Quick diagnostic: are you dealing with disconnected work?

The messy middle: where revenue slows down and trust leaks

The keynote framed revenue in a way that's useful for execs:

Revenue unites every department

That's why it breaks first when handovers are messy.

Another keynote slide showed the core problem visually: CRM on one side, ERP on the other, and then a messy middle of disconnected processes in between.

The messy middle between CRM and ERP

A simple way to find your messy middle in 30 minutes

Pick one revenue stream (renewals is a good one because it is frequent and measurable), then answer:

1. Where does intent start (form, call, partner, inbound email)?
2. Where is the quote created, and who changes it?
3. Where do approvals happen (and how often is it email)?
4. Where does a "deal" turn into an "order"?
5. Where is the invoice created?
6. Where do disputes and credit notes begin?

Now circle where work leaves the system and becomes human glue.

The four questions that cut through confusion

Another slide boiled it down to four simple questions: what you sell, how you sell, how you go to market, and how you market.

Four key revenue questions

Here's the key link to agentic tech: If you want agents to take action in revenue processes, you need to make those actions safe and predictable. That means fixing the messy middle first, even if it is just for one journey.

Field service: where agentic tech has an obvious ROI

Field service is where you can see value quickly, because the work is real-world and time-bound. You either turn up ready, or you do not.

One Field Service slide said teams are stretched, and called out common blockers like poor integration, limited visibility, no asset connectivity, manual reporting, and non-optimised scheduling. It also showed two big numbers:

• •70% facing a resource shortage
• •56% of mobile workers reporting burnout

Teams are stretched

This is exactly where "agentic" can help, as long as you keep it controlled.

What "asset-centric service" means

You are not just supporting a customer. You are supporting the thing they own (asset).

So you need to track:

• •what was installed
• •what changed
• •what parts were used
• •what's due next
• •what might fail soon

What the platform view looked like

There was a clear "stack" view: customer experience, work and asset management, scheduling and dispatch, mobile workforce, and operational excellence, with Agentforce sitting across it.

Field service platform stack view

A real example shown: Agentforce used to cut admin loops

The "Genius Sports: Agentforce in Action" slide was a good example of where value comes from: invoice automation, fewer credit notes, less time wasted on invoice queries, and faster quote creation.

Genius Sports: Agentforce in Action

Marketing Cloud journeys: three patterns you can copy

Marketing Cloud content was practical because it focused on repeatable patterns.

1) New customer journey: get them to value fast

The slide's message was basically: personalise early, use your data (and agents) to make a strong first impression, then guide with a clear next step.

New customer journey

A simple version you can run:

• •Day 0: Welcome plus one next action
• •Day 2: Quick win guide
• •Day 7: Show what good looks like
• •Day 14: Ask one question that improves segmentation

2) Event-triggered journey: act fast and make it relevant

The event-triggered slide had a strong line: show, don't just tell. It also gave an example outcome linked to revenue and recipient volume.

Event-triggered journey

In practice, event triggers should be based on clean signals:

• •quote created but not progressed
• •onboarding step completed (or not)
• •appointment booked, cancelled, or missed
• •renewal window opening

3) Re-engagement journey: be persuasive, then be brave

This slide was honest: you need a reason to return, you should test, and you should not be afraid to suppress disengaged audiences to protect sender reputation.

Re-engagement journey

Quick table: which journey do you build first?

A simple pilot-first delivery plan for Agentforce

This is where most teams go wrong. They try to roll out "AI agents" everywhere, all at once.

A better approach is what your brain already knows is sensible:

This is also exactly how my fixed-price offering is structured:

• •one controlled pilot
• •clear scope
• •guardrails and approval points
• •audit trail
• •measurement pack that tells you whether to scale, adjust, or stop

What a good first Agentforce pilot looks like

From the "best fit" criteria on the offering page, a first pilot works best when the work is:

• •high-volume and repeatable (triage, routing, summaries, booking, FAQs)
• •measurable (you can count time saved, errors reduced, backlog cleared)
• •safe (clear boundaries and human approval where needed)

Pilot deliverables you should demand

If someone proposes an Agentforce pilot without these, be careful.

Scorecard: what to measure so you know it worked

If you want to be clinical about this (and you should), choose a small set of metrics that show value and risk.

Sources

Here are two sources that back the big themes in this article:

• 1.Salesforce Agentforce (what it is, and the "agentic" framing)
• 2.MuleSoft Connectivity Benchmark Report (Salesforce) (the underlying problem of disconnected systems)