Customer experience is everybody's job, but if everybody owns it, does anybody?

That was the question we put to the room when we hosted the CX Collective at Beauty Pie HQ in London. We wanted to open the conversation beyond CX and customer care, inside the office of a brand living these questions every day. Expectations are shifting, technology is shifting, and the role every function plays in the experience is shifting with them.

The panel spanned commercial, marketing, engineering and CX: four very different vantage points on the same customer. Here is what stood out to us as hosts.

Ownership works when it's embedded, not orbiting

The strongest models we heard weren't the ones with the grandest CX title; they were the ones where customer sat inside a profit and loss (P&L). At Bulk, CX reports into the commercial leadership team, which means margin, product and brand conversations happen with a customer lens in the room before decisions are locked, not after.

The failure mode is the opposite: a customer leader with no budget line, circling the business asking "but what about the customer?" Without data and modelling that shows a P&L owner why their pound is better spent here, that role burns out, and the mandate goes with it.

Ownership also has to be enforced and enabled. Customer targets carried the same weight as revenue targets in the businesses where "everyone owns the customer" genuinely worked.

"Unless a customer leader can show you clearly, with data and modelling, how your pound is better spent on customer than on something else, you're never going to do it. That's a recipe for frustration on both sides."

- Rhea Fox, Marketing Director, Moonpig

Buy-in is a coalition, built before you need it

Matt Williams's approach at Bulk: set the strategy publicly at the start of the year: mission, principles, outcome metrics, goals, so every leader knows what they're being held to. Then work the relationships one by one, building a coalition of stakeholders before the business case ever reaches the CFO.

Translate for your audience. The same problem statement needs a tech version, a brand version and a commercial version. Keep it short, keep it costed, and land the urgency of doing nothing.

Rhea Fox offered the top-down counterpart: the value-at-risk case. If churn moves from 6% to 8.5%, here's how many members we lose next year, here's the revenue gone, and here's what replacing them costs in media. Get the CEO to agree they don't want the status quo, then arrive with the fix.

"You don't need 100% of the data. If you've got 60% and sensible assumptions, you can get to 100% of a decision."

- Rhea Fox, on building the case

Proximity changes priorities

Every function makes decisions that land on the customer, but most just never see the landing. At Beauty Pie we've opened up shadowing across the whole company: be an agent for a day, sit with Member Happiness, watch your decision arrive in the queue.

Vaishali Patel has run the same play with engineering teams: a full day in customer service every quarter, consistently, not as a one-off. The result was pressure from the bottom up: engineers who had seen the problem became vocal about fixing it. Paired with member happiness data in the prioritisation deck, that's how CX work makes the roadmap.

The most quoted example of the morning: a former CTO who, after a shadowing session, built the frontline team a custom app for stock checks, order updates and member status in 48 hours. Work everyone had assumed would take months.

"That's the point they realise: oh, that's the decision I made, and that's the impact it had."

- Vaishali Patel, Head of Engineering, Beauty Pie

Voice of customer: stitch it, shorten it, and question NPS

The feedback is everywhere: tickets, chatbot, Trustpilot, product reviews, social, and almost always siloed. Bulk has started pulling it together with an MCP connection into their chatbot, plus BigQuery feeds for Trustpilot and automated Zendesk reports, working toward a single sentiment dashboard. The practical win: flagging an anomaly the same day instead of dissecting last week on a Monday morning.

On metrics, the panel was blunt. Nobody could point to a convincing correlation between NPS and revenue, and touchpoint surveys rarely reach the root cause. What's rising in value instead: online reputation: Trustpilot, and increasingly how your brand shows up in Reddit, forums and LLM answers, because that's where decisions get made now.

Two metrics people would actually defend: Trustpilot score by market (4.3+ as the bar) and contact-to-order ratio, which forecasts resourcing and exposes friction in one number.

And a note for anyone building the reporting: nobody wants the ninety-minute insight deck. Short, sharp, commercial. That's what gets acted on.

"Beyond whether it's accurate, online reputation is how other people make decisions now."

- Rhea Fox, on why she'd track reputation over NPS

AI: the boring plumbing decides the experience

The AI conversation was refreshingly unglamorous. Authentication, privacy and system integration came first: if your platforms can't identify a member and pull their data cleanly, AI won't work in your business, however good the model is.

Where it does work, the bar has moved. The example everyone recognised was a returns flow: reply in three minutes, courier booked in ten, refund processed in days, all of it reading like a human wrote it. That's the standard customers now quietly expect.

The provocation of the session: AI may make customer service, as a discipline, smaller, and that's not a threat if you point the capability elsewhere. Fewer contacts because needs were anticipated. Fewer returns because sizing and fit were solved upstream. IKEA was cited as the model: automate the inbound, then reskill the team into design advice, and report a billion in incremental revenue for it.

The organisational pattern to steal: an AI super-user group. Cross-functional, sharing prompts, scripts and integrations. Bulk and Moonpig are both running one, and both found teams were solving the same problem in parallel without knowing it.

"Immerse yourself in AI, not to replace roles, but to see what it creates. That's the part we don't know yet."

- Matt Williams, Head of Customer Satisfaction, Bulk

What great looks like next

Exceptional experiences are rare, and rare is what makes them talkable. The examples that stuck were small and frictionless: a debit card reprinted on the spot in Edinburgh; a return accepted from someone who wasn't even the buyer. Both were moments where resistance was expected and didn't come.

The forward-looking ask from the panel was proactivity. Brands hold enough data to anticipate a member's next need: replenishment, sizing, a problem forming, and mostly wait to be contacted instead. Use the data for the customer, not only the bottom line, and the bottom line follows.

Closing thought from the room: it costs roughly five times less to keep and re-engage an existing customer than to acquire a new one. Worth repeating in your next acquisition-heavy planning meeting.