Making Tech Transform Hospitality Marketing

Digital Director Mat Bowden was interviewed in August 2024 about some of the trials and tribulations he has faced when developing clients digital marketing infrastructure and some of the creative ways he has had to use to MacGyuver solutions.

In this instance, he recalls being the retained digital marketing agency for a high-profile restaurant group:

THE PROBLEM

What was the original client brief?
The client engaged us to manage and optimise their digital marketing. Our job was to drive bookings and brand growth across multiple channels. On the surface, that’s straightforward, but the data setup didn’t support meaningful optimisation.

And what was the real problem once you got stuck in?
Each restaurant had its own unique website on separate domains, all funnelling into a single, centralised third-party booking system. That booking engine was closed off, no custom tracking, no data visibility, no integration options. So we were spending across Meta, Google, TikTok etc, but had no way to see which activity actually led to a confirmed booking.

If we couldn’t close the loop, we couldn’t optimise and that’s what forced us to rethink the infrastructure.

That then led to a further problem of development, without any budget for development.

IDEATION

What were your first steps?
We began with a deep technical audit of the digital estate. Mapping every user journey, noting where tracking dropped, and identifying what could be standardised and optimised. We also logged every paid media touchpoint to understand what was being sent vs. what could be tracked.

How did the shape of the solution come about?
We knew we couldn’t touch the booking system often, so we quickly dropped in a GTM tag to allow us to iterate without need for continual access to the third party site, we then focused on reducing steps for users and stitching together everything before that final handoff. That meant tightening up UTM hygiene, using shared GTM containers across the restaurant sites, and creating a rules-based attribution model that could infer conversion intent through pre-booking behaviour.

Were there any major hurdles in ideation?
Yes, CMS inconsistencies were a problem. Some platforms rewrote query strings or blocked certain scripts outright, causing inconsistencies in the data and identifying the same user across multiple domains is a bit trickier than most would realise.. 

This led to some hefty re-coding, and so we had to resist the instinct to track everything and instead focus on what mattered most to media optimisation: source clarity, behaviour patterns, and booking indicators. It wasn’t about collecting every signal, it was about identifying the ones that drove action.

Where did you look for inspiration?
We didn’t want to reinvent tracking. So we borrowed tried and tested principles from affiliate tracking, broadcast attribution, and old-school direct response: measure what you can, infer what you must, and make sure it holds under scrutiny.

We weren’t building a product, we were building signal intelligence for better media performance

PROTOTYPE & DESIGN

What stood out in the design/prototyping process?
We treated the solution like a performance layer. It had to be invisible, fast, GDPR-compliant, and deployable across multiple platforms with minimal fuss. We prototyped the unique identifier’s first and cross domain tracking, followed by more detailed attribution logic, treating it  like code: versioned, testable, and debuggable. Every data point had to earn its place. Then validated it using live traffic simulations.

What was the technical grit?
Safari’s privacy settings. Non-compliant UTMs. Inconsistent cookie behaviour. We solved it by building in fallbacks, layered timestamp tracking, referrer chaining, even using session duration as a proxy in some cases, just to ensure that we were measuring correctly.

Who was involved?
We had our media and strategy teams aligned with UX, analytics and dev support. No silos, the value came from keeping planning, execution, and tracking fully integrated.

Any creative or technical risk?
Absolutely. We were building this to make marketing decisions. The process only works if it’s well built and brutally tested. This wasn’t a guess, it was logic, stress-tested against real journeys and validated repeatedly. But in a volatile market like hospitality and with a celebrity chefs name to consider, there was little room for error at any stage. 

What were the most interesting or frustrating challenges?
One major issue was CMS variation, some sites stripped or altered UTM parameters, which broke attribution. Other CMS wouldn’t push triggers into the DOM, so there were many work arounds required. We built fallback logic and layered in referral and timestamp stitching to keep our model intact.

Who did you collaborate with?
The client had no dedicated web team, so we picked up the mantle,  we worked with the third party vendors web teams to implement and brought in analytics specialists to help model the logic. Our media planners also fed into the process, because the entire point was to use this data to make smarter campaign decisions.

Any creative or technical risks involved?
Plenty. We effectively reverse-engineered attribution using external signals,  which meant that in early iterations, a browser update, cookie policy shift, or CMS tweak could potentially break our model. But we designed for flexibility and documented everything for futureproofing; which eventually led to actionable insights that increased bookings.

LIVE

How did testing and iteration fit in?
It was baked in. We ran parallel tests, tracking alongside live campaigns and monitored discrepancies in real time. We also used heatmaps, user journeys, and manual spot-checks to test the attribution logic under pressure.

Any spicy back-and-forths?
Ha! Yes, A campaign launched with inconsistent naming conventions, which temporarily broke our attribution chain. We patched it fast and used it as the case study to enforce strict UTM protocols across all platforms.

What did you learn from the data once it went live?
A few surprises. For instance, we found that in a couple of locations Friday night bookings were more influenced by in-house entertainment than digital promotions, a valuable insight that helped shift the comms and budget strategy. We also managed to link the bookings to receipts and so identified certain dishes and offers that were underperforming.

What was the most personally interesting part of the project?
Honestly, the clarity it brought. Solving a real problem with no fluff. This wasn’t about a flashy tool, it was invisible, functional infrastructure that let us track performance. 

There’s something deeply satisfying about solving an invisible problem, building an elegant solution that lets a marketing team do what it’s best at: optimising in real time, not guessing post-campaign.

What impact did it have on the client’s business?
We delivered clear, actionable insight at speed. The client quickly dropped a long-running promo that wasn’t performing and redirected spend into an offer that was. Returns started landing immediately, and by year-end, they’d hit six figures.

Faster optimisation cycles kept performance climbing. But more importantly, the client had full confidence in the approach, because they could see what was working, why it worked, and how it was driving real results.

All backed by data, not gut feel.