Marketer Stories50 interviews, mapped

Each person is linked to the two to four concepts their interview belongs to and to their company, so the map pulls people towards the themes they actually talked about and shows who shares an employer: DeepAgent appears twice, five people are independents. Switch on channels and tools to see who runs Meta, who swears by HubSpot. The filters in the strip above apply here and to every table; clicking a concept, company, channel or tool, here or in a table row, lists the people behind it; hovering a row lights its node up here.

Ask marketers about channels and they answer about data. The most crowded cluster is not paid media, it is data and attribution: 38 of 50 interviews land there, almost always with the same complaint. Platforms over-attribute to themselves, dashboards get built and never read, and the number that matters, margin, LTV, the conversion down the funnel, is rarely the one on the screen. Maria Chiara Barbaro said it bluntly: attribution is a problem everyone wants to solve and nobody does.

The second surprise is how often the conversation drifts to craft: curiosity as a habit (Andres Cuadros), struggle tattooed on the knees (Claudia Capovilla), unlearning as a senior skill (Valeria Cossalter), precision as the thing to hire for (Giulia Mastrantonio), psychology instead of a marketing course (Barbaro), mastery of your own product in deeply technical B2B (Edoardo Giaquinto). The people who run the most technical stacks are the ones most insistent that the edge is human.

Meta is the most named channel and the most conditional. It wins for discovery and e-commerce starts (Antonio Martino, Marco Salvaggio), it is the main engine for a €500-a-year legaltech because everyone from CEO to intern is on Instagram (Stefano Quinzii), it took Psicologoinchat from zero to €10K MRR alone (Barbaro), and it fails for leads in automotive while working for awareness (Michele Poiega). Google is the channel nobody would cut: Francesca Pietrobon keeps it above everything, Ilenia Giovannini pays €20 a click when the click sells an iPhone, Mastrantonio calls it the best targeting and the most exhausting automation. LinkedIn is the only channel Giaquinto would keep for Formula 1 engineers, all organic, and the one Quinzii dropped for its cost per lead.

Two channels are described as undervalued by people who measured them. WhatsApp: an open rate above 85% and double the conversion of newsletters in e-commerce (Pietrobon), a human on WhatsApp beating an AI chatbot in a dealership (Poiega), and the whole pivot that got iSendU acquired (Mario Miletta). Events and trade shows: a lab on the customer for a hardware founder (Claudio Carnevali), a qualified channel for HR consulting where 40 decision makers beat 100 attendees (Cossalter), essential at the right stage for AI B2B (Mario Nicotra). And two channels that cut spend instead of adding it: six German marketplaces at once halved Baldinini's performance budget (Giuliano Torelli), and Meta's own ad library, read as an intent signal, gave Adriano Matarrese 66 appointments in three days.

ROAS is the metric everyone mentions and the one most distrusted. Jessica Gotti replaces the average with marginal ROAS: is the next dollar efficient? Pietrobon replaces it with MER, because the platform grading its own homework will always pass. Giovanni Tresca shows why: a €3M e-commerce with 5× on Meta and 8× on Google was sending 40% of its Meta budget to a low-margin line, and its real margin was 28%, not the 45% on paper.

The pattern across both editions is a move down the funnel. Conversion rate, not CPC or CTR, decides what Lorenzo De Stefano scales. Sales-qualified leads by BANT, not MQLs, are Matarrese's north star. Enrico Celotto measures from anonymous traffic to closed opportunity, so that a 5% sales conversion instead of 20% points at sales, not at leads. Davide Bessi's cascade runs contracts, onboarding, activation, expansion, MRR. Two people defend the opposite move: Mattia Di Angelo keeps CTR as the earliest signal of message-market fit, and Mastrantonio watches the no-show rate on booked calls because it is the one number she can act on the same day.

Every one of the 50 interviews touched AI without being asked twice. Reading them together, three stances emerge, and they are not about age or seniority.

What unites all three: nobody lets the model decide. Igor Papo caught himself unable to structure a speech without an LLM and went back to the Business Model Canvas. Antonio Martino's home-made AI reporting tool told him to pause a working campaign. Francesco Urso screenshots the dashboard before asking the model to read it. Robb Clarke, whose AI handles 99.3% of support, stopped a flood of tickets by changing one word by hand. The most AI-forward interviewees, De Stefano, Lombardi, Tresca, Matarrese, Cossalter with agents on HubSpot, are also the most explicit that judgement stays human. Paolo Franzese put it bluntly: AI increases the need to understand tech, it does not diminish it.

There are about 15,000 martech tools, Luca Barboni reminds us, and no all-in-one. The stack these marketers describe is smaller and humbler than the market would suggest: spreadsheets appear as often as BI platforms, Igor Papo's count is brutal, one client in twenty has a BI tool at all, and Carmelo Abate, who used to teach a whole lecture on tools, now calls them almost irrelevant.

Two recurring moves: build your own, dashboards in Lovable, reporting in n8n and Notion, data pushed to Supabase and read back through a hidden subdomain (Matarrese), a weekly Excel of predictions versus actuals that takes 45 minutes (Barboni); and centralise before you scale, a single source of truth before the next campaign (Tresca, Celotto, Cossalter). Nobody argued for more tools.