TL;DR
- Cheap traffic can prove costly:Low acquisition costs can hide failed payments, rapid churn and complaints that lead to refunds, fines or service suspensions.
- Measure what happens after conversion:Feeding billing, renewal and complaint data back into campaigns helps attract profitable customers, but technical and product problems need separate investigation.
- Test the next wave of advertising:Value-based bidding and Google Demand Gen offer opportunities to improve acquisition, with success depending on reliable data and careful experimentation.
A cheap customer can turn out to be an expensive mistake. For mobile content and subscription businesses, the acquisition dashboard may show a campaign beating its cost-per-acquisition target while, further down the line, those same customers fail to bill, cancel or complain.
That disconnect was at the heart of “Cheap Traffic, expensive lessons – reading the data behind mobile growth” at Telemedia Marbella. Mesh’s Christopher Henseler and Alessio Matteucci, MCP Insight’s Thomas Tinker and Google’s Francesco Beninato explored why the industry needs to look beyond the conversion and start feeding the commercial consequences of acquisition back into its marketing.
“Just looking at CPA and cheap acquisition is not enough anymore,” Henseler told delegates. Even tracking the first successful billing, something Mesh had been encouraging partners to do, no longer tells the whole story. Renewal payments, churn, complaints, refunds and potentially content usage all help establish whether a campaign is producing a viable business.
The challenge is turning those signals into better buying decisions without teaching the advertising algorithm the wrong lessons.
Matteucci offered an example from the UAE, where broad audience targeting delivered acquisition costs of around €2–€2.50. From the media buyer’s perspective, that looked good. Conversations with clients revealed a different picture.
Feeding billing information back into Google exposed placements, devices and sections of traffic that were generating conversions without generating profitable customers. Tightening the campaign raised acquisition costs, but improved what clients actually received.
“The traffic that we were delivering was not as cheap as before, but it was for sure much more effective,” he said.
For an industry accustomed to negotiating around a headline CPA, that is a significant change. Paying more for a subscriber who pays and stays can make considerably more sense than buying a succession of cheap sign-ups that never recover their acquisition cost.
Are complaints useful?
The consequences also extend beyond an individual campaign’s profitability. Tinker warned that complaints attract the attention of operators and regulators, potentially bringing suspensions and carrier fines.
“Complaints are the only things that the MNOs don’t want to see growing,” Matteucci said.
Tinker argued that operators do not always assess complaints against the wider picture, including transaction volumes. MCP Insight is working with some carriers on that approach, but merchants still need to prevent problematic traffic from becoming an operator problem.
Matteucci described campaigns in Norway and Switzerland that performed well against CPA targets but prompted clients to report high complaint rates. Investigating the placements allowed Mesh to exclude problematic sources. Acquisition costs were affected, but the reduction in complaints left clients happier.
The missing ingredient is often speed. Henseler pointed out that agencies can learn about refunds weeks or even months after the acquisition, by which time the opportunity to intervene has largely passed.
Tinker said closer integration between compliance providers and agencies is improving that situation. MCP Insight can feed information back in near real time through its analytics and anti-fraud tools. It is also using AI to assess banners and landing pages for compliance issues and issue alerts, extending work that previously required human inspection.
AI sniffs out anomalies
Mesh, meanwhile, is using its Mesh One AI agent to identify anomalies across the acquisition funnel and direct account managers and media buyers towards emerging problems. Matteucci described daily notifications highlighting where attention is needed.
The wider analytical and campaign decisions still require human judgement. Finding an unusual pattern is useful; understanding its cause determines whether the response improves the business.
That distinction becomes especially important as advertisers explore value-based bidding. Beninato explained that target CPA effectively treats customers as having the same value, whereas a value-led approach can help distinguish between someone who quickly churns and someone who continues paying. A fixed acquisition target can therefore leave a business overpaying for poor customers while missing better customers worth spending more to acquire.
Moving to value requires experimentation, Beninato cautioned. Campaigns may need more learning time than advertisers expect when working towards a familiar CPA target.
Henseler suggested that making the first successful billing the primary optimisation event could offer a more meaningful starting point than assigning an identical value to every sign-up. Establishing subscription lifetime value takes time, he said, while the first payment provides an earlier indication of commercial potential.
But even that signal needs careful handling. “The credit in the SIM card of a user is unfortunately not a marketing KPI,” Matteucci warned.
A failed billing attempt can reflect an insufficient prepaid balance rather than a lack of interest. Similarly, churn and complaints may originate in poor content, confusing journeys or technical failures. Feeding those outcomes back indiscriminately risks blaming acquisition for something it did not cause.
“You can also confuse the data if you now think Google is responsible for all your high churn,” Henseler said. Tinker illustrated the point with MCP Insight’s optimisation testing, which examines user journeys across different SIM and phone scenarios. He reported an 8% revenue improvement in a UK example from analysing those scenarios and identifying technical obstacles.
A customer who wants to subscribe but cannot complete the journey is lost revenue, however good the campaign that brought them there.
Detailing Demand Gen
The panel also discussed the move towards Google’s Demand Gen campaigns and what it could mean for advertisers accustomed to Display. Beninato encouraged businesses to test early and build experience before being forced into a change, while making clear that Google’s plans remain dynamic.
Matteucci said Mesh had already begun testing Demand Gen and campaign transformations, with no major performance impact in its experience so far. His advice was to experiment with campaigns at an early stage or those needing more traffic.
Across both bidding and campaign formats, the message was to build evidence. There is no single valuation model that every subscription service can simply adopt.
Mobile growth now depends on knowing what happens after the click, after the conversion and after the first bill. The cheapest traffic may still have a place. But until the business can see what it earns, what it costs and what problems it creates, “cheap” is little more than a number on a dashboard.
