Personalised Landing Pages at Scale — Just Eat for Business

Discovery / Design / Development / Data Import

In 2019, Just Eat for Business experienced a phase of rapid growth. Having previously operated solely in London - Just Eat for Business were now offering corporate catering to a total of 8 cities across the UK with various other cities forecasted to be added to the offering in the months.

Due to the expansion, we rapidly expanded the range of PPC keywords that we were bidding on. We were now targetting a total of 760 keywords and phrases across 8 different cities.

With 6000+ different combinations of PPC adverts across various platforms but only 20 variations of landing pages to link the adverts to, the Marketing team were forced to send customers to generic landing pages that weren't personalised to the advert that the customer had engaged with.

Our solution was to ensure that customers clicking any one of our 6000+ PPC adverts were served a landing page that mirrored the information they had seen on the ad.

4.74%

Increase in form submissions

85%

Increased average time on page

Ideation / Design

Having discovered that three colleagues in the Marketing team had been working with Just Eat for Business throughout some of its key growth years; I arranged a whiteboarding session seeking to understand the existing pain points.

It was clear from this meeting that the team were aware of the value of a personalised proposition but didn’t have access to a solution.

"As a consumer whom has clicked a PPC advert and landed on a Just Eat for Business landing page, I want to be shown recommendations that are local to me so that I know I can offer my teams the food being advertised"

Technical Discovery

I came across the following article and became particularly interested in HubDB’s dynamic pages and ‘child table’ feature.

Not only could you create landing pages that were generated by a single row in a database table, you could also link each row to a ‘child table’.

This meant that the data in the child table could be combined with its parent table and in theory generate dynamic pages whereby any single row from each of the two tables could be combined.

I created a parent table for our ‘types’ (e.g. ‘pizza’) and linked each row to a ‘locations’ child table.

I then set up a landing page to be powered by the parent table meaning that any ‘type’ could be combined with any ‘location’ - /pizza/london. I then created a third table for ‘suggestions’, added restaurants as rows and added a column for ‘tags’.

With a Hubl function that checked the users url and compared it to the tag's column we were able to display restaurants that were specifically relevant to that particular ‘type’ and ‘location’.

HubDB table with child tables defined

Output of combined HubDB tables

Design

The sole purpose of these particular landing pages was to generate leads arriving from PPC adverts.



With this in mind, we set out to create a simple layout that formed a seamless journey when arriving from an advert but fundamentally, also encouraged action.



Suggested vendors were linked directly to contact forms, content flowed naturally into CTA’s and the page was kept as short as possible whilst communicating Just Eat's B2B offering.

Data Import

Over the course of 2 weeks, the Marketing team compiled and uploaded a spreadsheet containing all 760 ‘types’ as rows (including hero imagery for each) and separately a spreadsheet that contained the 8 locations.

This resulted in 6,080 variations of the landing page each displaying 1-6 vendor suggestions relevant to that specific page.

Example of a ‘type’ being readily available on any row in the ‘locations’ table -



/bagels-at-work/london
 /bagels-at-work/manchester /bagels-at-work/birmingham /bagels-at-work/leeds /bagels-at-work/reading /bagels-at-work/liverpool /bagels-at-work/edinburgh /bagels-at-work/glasgow