TL;DR
- Optimise for lasting value: Early billing results and 30-, 60- and 90-day revenue and retention data help merchants buy subscribers who recover their acquisition cost.
- Fix the journey as well as the targeting: Clear pricing, reliable billing and working services reduce avoidable churn; technical failures must not be mistaken for poor customers.
- Use AI with meaningful data: Analytics, audience tools and value-based bidding can improve acquisition, but their usefulness depends on the signals merchants can collect and share.
| View the full session here |
Netflix, Strava and Duolingo are the poster boys for subscriptions, but their customers do not necessarily behave alike. Some use a service every day. Others return occasionally, pause or cancel and come back. Keeping a subscriber is therefore about understanding what makes the service worth paying for, as well as how often it gets used.
For telemedia businesses, there is another question: has that subscriber stayed and paid long enough to cover the cost of acquiring them?
That was the central issue in Telemedia Marbella’s subscription session, where Kilian Espana, CEO and founder of Overdrive Digital, Jennifer Courtney, Senior Google Ads performance specialist and regional team leader at Google, Laurent Baquiast, CEO of Wister, and Smitha Mathew, business development director at Aurora Digital, explored how to turn acquisition into durable revenue.
The message was familiar, but the discussion made it practical. A good conversion rate tells merchants that people are signing up. It says much less about whether they understand the purchase, receive the promised service or successfully pay for subsequent renewals. As Espana put it: “The ad is the onboarding.”
| View the full session here |
The creative promise
The relationship starts with the promise in the creative. If that promise does not match the product, or the price and subscription terms are unclear, the campaign may acquire a customer who leaves almost immediately.
That is why asking an agency for its best markets and cheapest CPAs can miss the point. A competitor’s acquisition cost in the UAE says little about what another merchant can afford. The answer depends on what those customers subsequently contribute.
Espana urged businesses to move beyond acquisition targets and feed back information about subscribers who remain after 30, 60 and 90 days, alongside the revenue they generate. Looking at retention by placement, device and creative can reveal which campaigns acquire lasting customers.
It also changes what the advertising system is being asked to find. Optimising towards a sign-up rewards the delivery of sign-ups. Passing meaningful revenue and retention signals gives the buyer and the platform more evidence about what a valuable customer looks like.
For Wister’s Baquiast, the challenge is making that assessment quickly enough to guide investment. “The key point is to assess the long-term value quick enough according to the campaign,” he said.
Nobody can wait years to discover whether a campaign made money. The first successful billing provides an early signal; later results help establish how well that signal predicted longer-term value. Merchants can then examine performance by campaign, traffic source, placement and device.
But revenue alone is not sufficient. Baquiast also stressed the complaints a subscription generates towards the carrier. Even with good conversion and lifetime value, “if people complain a lot, it’s not a good solution at the end”, he said.
Aurora Digital’s Mathew described a similar approach taken at her company, including looking at performance after 10 days when entering new markets. Her wider point was that the same funnel and price cannot simply be exported everywhere.
She contrasted opportunities to sell higher-priced quality products in European postpaid environments with the importance of affordability in African markets. In more mature Middle Eastern DCB markets, charging capabilities and customer profiles create another set of conditions.
These were practical observations from Aurora’s experience, rather than a universal formula for each region. The implication is that an acceptable acquisition cost must be worked out against local pricing, billing performance and customer behaviour.
“Until the affordability factor and the pricing is taken care, the market is not going to adapt your product,” Mathew said.
How Google can help
The tools to support better decisions already exist. Google’s Courtney’s presentation focused on combining first-party data, website measurement, APIs, customer systems and Google Analytics so campaigns receive a fuller picture of the subscription journey.
Her advice was to begin with a manageable setup: map the journey, choose two or three meaningful points and quantify their value to the business. “It is very important that you sit down, you map out the subscription journey,” she said.
A trial, subscription or important action can provide information, but the business needs to decide what that information means. A click expressing interest and a successful renewal are different outcomes. Treating them as equally valuable would obscure the distinction the merchant is trying to make.
Courtney highlighted Data Manager for connecting data sources, Analytics for events and audiences, and the importance of passing values back into campaigns. Espana said Overdrive was already using value-based bidding, including target return on ad spend, with some clients able to supply the necessary information.
An audience question pushed this further: could merchants feed back the behaviour of customers actually using the content? “Yes, in short, the answer is yes,” Courtney replied.
She suggested specific conversion actions, Analytics events or Customer Match lists as possible routes. The useful step is identifying behaviours associated with valuable subscriptions and making those signals available to acquisition teams.
Feedback loop limitations
But the panel also exposed a danger in that feedback loop. A customer who fails to generate expected revenue may have been let down by the billing system.
Mathew described situations where technical failures were discovered only after somebody noticed that overall billing had dropped. Without adequate monitoring and alarms, revenue could erode for days or weeks. “It is not a profiling issue; it’s a technical issue,” she said.
That distinction matters when performance data goes back into advertising. A good customer who cannot be charged should not automatically teach the algorithm to avoid similar customers. Billing logs, error analysis and clear responsibility between operator and aggregator are part of acquisition optimisation too.
Espana said Overdrive asks clients for error logs to establish which failures can be addressed. Unavoidable losses need to enter the break-even calculation; fixable problems need correcting.
Baquiast also described handset testing to check billing, service operation and the visibility of compliance information. Buying better traffic achieves little if the customer cannot complete the journey or access what they purchased.
Clear consent is another retention tool. Baquiast called for compliance that addresses the “real purchase awareness problem: people need to understand what they are doing”.
Espana argued that displaying prices in advertisements may require a higher acquisition target, but can also bring customers who understand the commitment. Campaigns should adapt to that reality.
Reading the data runes
The difference between a customer leaving immediately in surprise and cancelling after a long, valuable relationship is substantial. As Baquiast said: “A bad churn is an instant churn.”
The conversation about mainstream subscriptions added another qualification. Low usage does not always mean cancellation is imminent. Someone may retain a fitness or learning subscription because they intend to return.
That observation needs care. Continued billing is not, by itself, evidence of satisfaction. The stronger ambition is a customer who understands the recurring charge and continues to regard the service as worthwhile.
Winning former subscribers back also requires more than a generic remarketing list. Mathew distinguished voluntary cancellation, payment-related loss and technical churn. Customers lost through technical problems may be worth approaching again, but the industry does not always receive enough data to identify them.
She also warned that number recycling, particularly in markets with frequent SIM turnover, can mean the same phone number belongs to somebody different later.
Courtney suggested exploring predictive audiences in Analytics alongside remarketing. There are limits: Google’s predictive models require sufficient qualifying data, and its churn probability measures likely inactivity on a site or app, rather than directly predicting cancellation of a paid subscription.
Asked whether there was evidence showing how much better AI was at reducing churn than human analysis, the panel did not provide a quantified comparison.
That leaves merchants with a useful discipline: test what improves their own business. Courtney encouraged gradual experimentation with Demand Gen, stronger creative and accurate data before campaigns launch. The objective is to connect acquisition decisions to commercial results.
Longer-lasting subscriptions emerge from that whole system: an honest advertisement, an affordable proposition, a working service, successful billing and evidence that helps the next campaign find suitable customers. Google’s tools can assist, but the merchant and its partners must supply the understanding.
A subscriber is worth more than a conversion. The business starts earning that value when the promise made before the click continues to hold afterwards.
| View the full session here |
