Mastering paid growth with Jonathan Becker of Thrive Digital

Executive overview

Performance marketing works best as one channel in a diversified mix — not a singular growth engine. Over-reliance on a single channel or loophole creates fragility when conditions change.

Creative testing is now the primary lever for paid growth efficiency, as bid management and targeting have been largely automated by platforms.

Attribution has always been subjective and remains so after iOS privacy changes — the right approach is treating it as an ongoing investigation using multiple signals, not a single source of truth.

When paid growth is the right channel

• Establish product market fit before scaling paid; performance marketing amplifies what already works, it doesn't validate an idea
• If other channels (organic, email, direct mail) convert, paid acquisition almost certainly will too
• For D2C and e-commerce, fast payback cycles feed the flywheel naturally
• For B2B and lead gen, model LTV-to-CAC and build lead scoring to bid on high-value prospects rather than optimising for lowest cost per lead
• Early-stage focus: does it work at all? Later-stage focus: at what scale can it work?
• Key inputs to assess: creative resources, technical staff, professional marketers in-house, attribution capability

Channel mix and diversification

• Treat the marketing budget like a portfolio — diversify across channels to reduce volatility
• Companies that crash are typically those who bet everything on a single channel or short-lived platform loophole
• Grammarly and Athletic Greens show that paid-primary scaling still works when unit economics are rigorously modelled
• Google and Meta remain the most advanced and ubiquitous platforms; nearly every performance marketing programme runs on at least one
• TikTok is promising — cheap CPCs, high engagement — but the ads platform is early-stage and creative-asset demands are very high (new assets several times per week)
• TikTok works best for D2C, founder-led brands with authentic video content, and influencer-driven campaigns; B2B SaaS applications are still emerging
• Amazon ads are powerful but narrow — primarily relevant for D2C and retail
• SEO and paid search are not mutually exclusive; a diversified media mix should include both

Creative as the primary performance lever

• Platform automation has removed most manual bid and targeting controls — creative is now one of the last levers performance marketers can meaningfully influence
• A common failure: brand design teams hand off polished assets without understanding funnel stages or channel mechanics
• Structure creative around the funnel: different messages at top (generating intent), middle, and bottom (capturing intent) — one homogenous ad causes banner blindness
• Run rigorous A/B tests by isolating a single variable across two otherwise identical ads within the same audience and ad set
• Use ratio metrics (click-through rate, impressions-to-conversion) to compare ads that received unequal impression volumes
• Send learnings back to the creative team and iterate: identify the winning variant, then challenge it with a new variant
• User-generated content (authentic, iPhone-filmed) consistently outperforms highly produced brand assets on social channels
• Seemingly small creative changes — like adding a dog to furniture photos — can double or triple ROAS
• Most testing can be done inside Meta Ads and Google Ads natively; exotic third-party tools are not required

Attribution: managing a permanently unsolved problem

• Attribution is how you determine the relationship between what you did and what revenue resulted — and it has always been imperfect
• John Wanamaker's 1919 observation still holds: half the marketing spend is wasted, but which half remains unclear
• Classic models: last-click (most common historically), first-touch, and multi-touch weighted attribution — all are subjective
• Apple's iOS 14.5 privacy changes removed IDFA access, fragmenting Facebook's ability to validate campaign attribution and audience targeting overnight
• Third-party cookie deprecation compounds this, removing another previously reliable attribution signal
• The modern approach combines multiple signals: platform-reported data, media mix modelling (MMM), customer surveys, and incrementality testing
• Media mix modelling (regression analysis, popularised in the 1950s) is now widely used to find causal relationships between spend and revenue; tools like Recast offer off-the-shelf MMM
• There is no single source of truth — anyone claiming otherwise is wrong
• Thrive has used sophisticated attribution models only to find via MMM that specific campaigns were not working; invalidating spend is a valid and important outcome
• Attribution is an ongoing investigation, not a solved problem

Lead scoring for B2B and long-cycle businesses

• Optimising for lowest cost per lead is a common mistake — cheaper leads often mean lower-quality revenue
• Build an ETL pipeline (tools like Supermetrics) to join CRM revenue data with channel data, then model which lead cohorts generate the most revenue
• A lead scoring model enables real-time bidding on audiences statistically likely to become high-value customers
• This addresses the slow-payback problem: predictive scoring lets you optimise today's bids against latent future revenue

AI's impact on performance marketing teams

• Platforms have been automating bid management and targeting with AI for over a decade — this is not new
• The result: teams do less manual campaign implementation and more modelling, validation, and creative strategy
• Generative AI (ChatGPT) now accelerates RFP responses dramatically — what took a week for five or six people now takes a day
• Image generation tools (DALL-E, Midjourney) reduce creative mockup time to roughly 1% of what it previously required
• AI democratises creative ideation — anyone who can articulate a thought can now generate visual concepts
• The analogy: like the printing press or industrial loom, AI displaces specific tasks but does not reduce the total number of creative roles; the nature of work shifts

Hiring for paid growth

• Technical aptitude is essential: look for backgrounds in mathematics, finance, engineering, or other quantitative disciplines
• Hands-on platform experience matters: can they run Meta Ads and Google Ads and explain what they are doing?
• Understanding of creative's role, attribution mechanics, and business unit economics separates strong candidates from weak ones
• Years of experience is a weak signal — one-year performers can outperform ten-year veterans
• For agencies and client-facing roles, composure under pressure matters: the "windows in New York City" logic question tests structured thinking and client-service temperament
• Mid-level practitioners often want to move into management after two to three years; factor this into retention planning
• Job title conventions: "Growth Marketing Manager", "Paid Acquisition Manager", "Performance Marketing Manager", or "Media Planner and Buyer" — benchmark against LinkedIn and Glassdoor

Agency vs. in-house

• Agencies and in-house teams are not mutually exclusive — Thrive requires an in-house point of contact on every engagement
• A professional marketer as a liaison (not necessarily a paid specialist) is needed for approvals and information flow
• Early-stage value of an agency: rapid experimentation, discovery of what works, access to cross-client pattern recognition and built-out processes that small teams cannot replicate
• Later-stage value: managing complexity at scale, sophisticated attribution, testing infrastructure, and capabilities difficult to hire for internally
• Uber retained Thrive for 10 years because deep institutional knowledge and proprietary tooling were hard to replace