How do we (try to) measure and track user engagement ?

I’d already heard about RFV (Recency, Frequency, Volume) as an indicator from the Financial Times, and our version synthesizes subscriber and member activity, considering regularity and consumption.

Given L’Équipe’s years-long focus on a membership-driven strategy, we now have over 2 million active monthly members and subscribers, along with a significant amount of behavioral data for analysis.

The most complex phase of implementing this project wasn’t developing the formula but rather the iterative testing phase. This allowed us to refine the metric, replacing “Recency” with “Regularity.”

One of the initial conclusions (unsurprisingly!) was a clear correlation between customer lifetime value and engagement level.

The more we increase customer engagement levels, the more subscriber lifetime extends and the more members convert to subscribers

Why is this metric more accurate than page views or time spent?

It takes into account the regularity, frequency, and volume of content consumed, while adding a weighting based on the “importance” of the content: the greater the “value” of a piece of content in terms of editorial significance, revenue generation, and the time spent producing it, the higher its weight will be.

• Regularity, frequency, and content volume: factors in the weight of consumed content—giving more weight to editorially and financially valuable content.
• Frequency of visits: How often a user returns to the platform (both web and apps)
• Content diversity: The range of distinct content consumed
• Tenure: How long someone has been a member or subscriber
• Interaction with content: With higher weighting for Flagship content

For instance, a user who primarily reads “live match” content with low consumption of premium or flagship content will get a lower RFV than someone consuming more higher-value articles or documentaries.

Triple objectives to boost engagement (and of course retention!):

1. Increase recurrence
2. Improve frequency
3. Encourage consumption of high-value editorial content

Key insights from RFV analysis and actions taken:

#1 Correlation between app downloads and high RFV

We developed, in partnership with POOOL, various formats to promote app downloads.

#2 Email openers have higher RFV

We launched a newsletter project targeting subscribers with a low RFV (less than 25) —particularly those with weaker weekend engagement— to provide a weekend news summary and insights from our editorial team.

#3 Consuming Flagship content drives high RFV

Although we defined the content weighting in our algorithm, our initial analysis emphasized the importance of ‘original’ content in customer lifetime value.

By ‘flagship content,’ I mean the very best of L’Équipe — the finest work from our journalists, whether in the form of videos, podcasts, written pieces, or documentaries.

One of the challenges for a media organization is ensuring its content thrives on the homepage of the site. News moves quickly, one story replaces another, and an incredible piece can fail to reach its audience because it was showcased for too short a time, paired with an unclear image, or given a misunderstood headline.

To address this, we adopted an idea suggested by our subscribers: releasing a monthly preview of upcoming content, similar to how platforms present their programming schedules.

Both on site:

And in the newsletter:

These campaigns resulted in notable RFV increases:

• +9% for platform clickers
• +12% for email clickers

> Subscriber (re)engagement strategies from L’Équipe and Les Echos-Le Parisien

#4 Interactive features do have an impact on RFV (Especially for members)

We aimed to foster interactivity among readers. With over 500K monthly comments, the “comments” section on the platform is essential. So to maximize its value, the first step was to update the space with a redesign and new features, such as emoji reactions, AI-generated comment summaries, etc.

Our second objective was to foster greater proximity between audiences and the brand: bringing journalists and readers closer together to develop a sense of belonging. As part of the Olympic Games, we introduced several types of content, including Q&A sessions, which we plan to gradually roll out across our platforms.

> Going further with interactive engagement: DER SPIEGEL’s debating feature

Priority #2: Targeted retention and re-subscription offers in an marketing automated cycle

Prioritizing former subscribers in your strategy, especially reaching out to them within the first few days after they cancel, is crucial:

• 70% of re-subscribers re-engage within seven days of cancellation.Former subscribers acquired during promotions are particularly responsive to targeted relaunches with attractive pricing.
• On average, former subscribers are 18 times more likely to re-subscribe than members, while members are 18 times more likely to subscribe than anonymous users.

Retention and re-subscription scenarios:

We identified two main phases, each divided into distinct analysis periods:

Retention Period: The client has made his cancellation request but still has access and is informed of its impending end.

Re-Subscription Period: The client loses access, prompting relaunches during key periods:

• Day 0–7
• Day 8–30
• Day 30+

We deployed cycles of retention and re-subscription across all customer touchpoints:

• Cancellation form (called in France “loi MUPA”)
• Service-client cancellation confirmation
• Promotional messages on digital platforms
• Apps, push notifications, in-app messages
• Paid media

To achieve the highest possible results, we first implemented offers (initially the same for all users) across all customer touchpoints.

Spoiler alert: doing this across all channels and levers has a MASSIVE impact!

All this communication is coordinated using audiences created within our CDP, which are then shared with all our third-party tools.

A culture of testing as a growth driver

Over the past 18 months, we tried a growth marketing approach, which, in our terms, means ultra-personalization and precise targeting of our subscription messages, starting with our retention and re-subscription offers.

A multidisciplinary team (data, acquisition, and CRM) was formed to define targeting strategies and, more importantly, activate them across our platforms. Each step is tested, analyzed, and then optimized.

Among the initiatives:

• Over 300 campaigns tested annually, featuring variations on:
  • Marketing angles (what are the benefits? Content? Which words should we use to define the benefits?)
  • Pricing approach
  • Trial period and its price

In 2024, enhanced automation of campaigns enabled us to manage around 15 specific segments with tailored offers and messages. The primary advantage of this automation is its ability to significantly improve our conversion rates.

Key learnings from these tests:

• We discovered that a significant portion of former subscribers are quite receptive to retention offers at higher price with commitments that do not negatively impact our ARPU but instead come with a one-year commitment.
• This approach meaningfully increases the LTV of our subscribers and, by extension, boosts our margins.

Thanks to these strategies, 75% of customers who reached the -100 days mark of their promotional offer period are still subscribers as of November 2024.

Romain L'hote is Marketing Director of L'Équipe, the emblematic French sports newspaper with a daily digital readership of 2.5 million, and over 190,000 subscribers to its paid digital offer. Subscriber acquisition is, as it is for many, at the heart of its monetization strategy. Today, Romain shares how, by trial and error, his team managed to halve acquisition costs on advertising platforms, while increasing both attribution and last-click subscriptions.

Over the past 3 years, our subscriber portfolio has grown significantly, thanks to a number of strategic choices, notably the shift from mass marketing to more precise reader targeting.

But, despite the power of our own channels (site, app and CRM, which account for the lion’s share of our subscriber recruitment), we continue to invest in paid to support our strategy. In 2023, 90% of our media investments were made on Google and Meta, with an almost even split. We are also combining this with other, less well developed, types of spending (X, TikTok, Twitch, influence).

Changing the way we do paid

Halving our subscriber acquisition costs didn’t happen overnight, of course. It took us two years to make significant progress on our two main KPIs:

• Conversion & cost of acquisition platform attribution (at 7 and 30 days)
• Conversion & last-click acquisition cost

There are several methodological biases in the measurements we use, not least because we don’t have an attribution tool:

• We don’t capture information on non-consenting audiences
• We have to rely on platform figures and their notion of post-click and post-view consent (7 days post-click, 1 day post-view for Meta, up to 30 day post-click for Google)
• With possible data duplication between different platforms…

However, our work on paid acquisition has been structured around three axes: 

• Firstly, segmenting audiences as finely as possible
• Secondly, to restructure our accounts on advertising platforms
• Finally, testing and personalizing our creative messages

1- Segment audiences as finely as possible

Generally speaking, the trend in paid marketing is to work broadly, i.e. with large audiences regardless of specific criteria. 

But we’re a special case. Thanks to our strong audience, and therefore large data flow, segmenting our audiences seemed a crucial first step. 

Historically, targeting was based on a readers’ favorite sports, but the results were rather uneven. We therefore chose to segment users in two ways: 

• According to a reader’s status: anonymous, member, former subscriber, etc; 
• According to their content consumption habits: volatile direct users, readers who remain on free articles, paid articles, videos…

Going from a segmentation with a myriad of divisions, often with very narrow targeting, we instead opted for 3 campaigns based on level of interest, each broken down into only 4 or 5 segments, in addition to a few one-off campaigns.

Given the human resources allocated to the subject (1.5 FTEs), our primary goal was to have something controllable and measurable; without having to micro-pilot segments of 500 people, which may perhaps have a competitive CPA but will ultimately generate few conversions.

Of course, not all groups have the same volume, profile or propensity to subscribe… 

You’ll also enjoy: The Risk of Personalization: do people want and trust it?

2. Restructuring our accounts on advertising platforms

In order to implement our targeting strategy, a lot of behind-the-scenes work went into structuring accounts around our 3 main target audience families, 3 “campaigns” in the platform sense.

We structured accounts using the TOF-MOF-BOF method, a funnel-based targeting system based on third-party and proprietary data: 

TOF (Top of Funnel) audience: 

The TOF audience refers to people whose understanding of our paid product, and the value it provides, is not determined. These readers are generally presented with subscription awareness ads. 

Trigger: third-party data, mainly based on centers of interest, possibly non-qualified visits (visits to our home page and direct mail).

MOF (Middle of Funnel) audience: 

The MOF audience has already shown some interest in the L’Équipe brand and is aware of our paid subscription. 

Trigger: these are logged-in members, and/or browsers on specific pages such as our live match commentary.

BOF (Bottom of Funnel) audience: 

The BOF audience is made up of people who are already interested in our premium offer, are highly engaged and are close to converting. 

Trigger: visit to our subscription landing page, tunnel entry, ex-subscribers, etc.

Our Google Ads campaign page

The data is captured from our platforms and from our CDP. We’ve also created a bridge between these and subscription management tools to be able to send ‘ex-subscriber’, ‘subscriber to X offer’ information.

Example of a campaign for a BOF (Bottom of funnel) user

Trigger: a user has seen our subscription landing page:

This reader is then targeted by a “BOF” advertisement:

Example of a campaign for a MOF (Middle of funnel) logged member: 

Example of a campaign for a BOF (Bottom of funnel) ex-subscriber: 

3. Personalizing messages

After structuring our audiences, it was time to think about how to personalize our messages. Visuals, pricing, offers, testing every word… it’s been a long, whole year of testing to find the right hook for the right customer! And of course… we don’t have the perfect combination.

Here’s an example of message personalization: contextualized offers for the Tour de France and the Rugby World Cup. 

We tested very regularly… Here are three things we’ve learned, some of which are counter-intuitive:

#1 Stop pushing sponsored articles 

Betting money on a specific “paywall” article, in the hope that it will convert readers into subscribers, is no longer a strategy we follow: the acquisition costs are too high in attribution, and in last-click. What’s more, their lifespan is too short to promote them long enough, and native posts already bring enough audience.

#2 Set up awareness campaigns to generate subscriptions

On very small targets, we sometimes found it difficult to distribute our campaigns. So, in order to improve our conversion volume, we tested converting on awareness campaigns… and it works!

The notoriety campaigns enabled us to distribute our campaign more widely, but also to maintain an acquisition cost similar to that of a classic conversion campaign.

#3 Don’t mess with the algorithm

This has become our biggest fear. 

In the case of any big promotional campaigns, we could always increase our investment very sharply for the first few days. But now, for each campaign, we try to scale gradually to avoid upsetting the algorithms and maintain stability on our account and our performance.

Paid strategy 2.0: the results

Thanks to our new strategy, we have substantially increased both attribution and last-click subscriptions, with a 2-fold reduction in acquisition costs in both cases.

Cost per acquisition (attribution) 

Of course, not everything has been a success, and there have been a few flops, particularly with our paid influencer campaigns, where the ROI has never been proven. 

Example of a failed influencer campaign to promote the L’EQUIPE gift card for Christmas 2023, with… 1 conversion 

Now what?

Our projects for 2024: 

• Continue testing, in a rapidly changing ecosystem with the end of third-party cookies, and with deadlines this summer that are likely to shuffle the deck a little on targeting, performance and metrics

• Develop video formats where we’re not particularly ahead of the game, and in certain verticals with an offbeat creative focus for TikTok to support one of the pillars of our overall brand strategy: reaching young people. At the same time, we want to promote subscriptions in a different way, and understand how algorithms work. 

• Try to recruit more qualified leads. Recruit less, but better, with higher ARPU. We want to target our offers even more, giving priority to our former subscribers, whose conversion rate is 13 times higher than that of our members.