Jonathan Becker
Founder and president, Thrive Digital
Where to hear and read about Jonathan Becker1 episode or test
Jonathan Becker is founder and president of Thrive Digital, a Vancouver agency that has deployed over $3.5 billion in paid budgets for clients like Uber, Asana, Square and MasterClass. The most useful lesson from him is to treat paid growth as one channel in a diversified mix, run on disciplined testing, known unit economics and attribution that never trusts a single source of truth.
I don't recommend putting all of your money into a single, you know, performance marketing channel
Mastering paid growth | Jonathan Becker (Thrive Digital)
Jonathan Becker: the story and the lessons
1How Jonathan Becker built Thrive and won Uber
Becker moved from SEO into paid search because clients wanted provable ROI. He started as a web developer who structured pages to rank in organic results. Clients kept asking about the ROI of SEO and how to scale it, and the answers pointed to paid search.
Google was removing the data SEOs used to guess and test, so organic felt intangible. Paid search gave him a tangible lever.
A freelance side business that began in a walk-in closet became a 130-person agency. Thrive now manages about $500 million a year in ad spend for clients including Uber, Asana, Tempur-Pedic and Lululemon. It is based in Vancouver, though most clients are in the United States.
Becker won Uber by confessing the loophole he was exploiting. In 2013 he went to the TED conference in Long Beach while running a 10-person agency. Garrett Camp asked to share his cab, and Becker told him he had bought Uber's branded keywords with a snippet that looked like Uber's organic listing, earning $20 per referral signup and tens of thousands of dollars.
He told Camp it added zero value and someone should close it. Uber's team first told him to stop, then hired him to fix it. Uber stayed a client for 10 years and moved Thrive from local bars and clothing stores to sector leader.
Becker says you make your own luck by risking embarrassment. He knew no one in the TED community and had little to lose, so he spoke up.
2Becker's case for a diversified paid growth mix
Becker runs client budgets like a fund manager, spreading across channels. He compares one-channel spending to putting life savings in one stock, exposed to CPC swings and market conditions nobody controls. His mix includes email, direct mail, SEO, linear TV and performance marketing.
He calls the goal "responsible channel mix diversity". Loopholes in one platform come and go, and he has seen brands scale on one and crash when it closed.
Becker says dependence, not paid itself, is what makes paid dangerous. Paid becomes a drug only when it is the sole source of revenue, which turns performance into a roller coaster.
Becker checks product-market fit and other working channels before betting on paid. Early on, he asks whether product-market fit exists and whether the company risks over-relying on paid. If email, SEO, influencer reach or direct mail already work, paid is highly likely to work.
Later-stage companies need creative resources, in-house marketers and engineers who can fix broken tracking.
Becker says know your maximum acquisition cost before scaling spend. Winners model how much they can afford to pay for a customer on any channel before it turns unprofitable. Grammarly and Athletic Greens know their impressions, their average quality and the spend level where ROI hits a cliff.
The focus is bottom-line profitability, not breakneck growth. Slow-payback businesses model LTV to CAC.
Becker recommends SEO and paid together, and likes paid for traceability. Finance and C-suite teams prefer paid because spend maps to revenue, so they can budget and project. With SEO you can only correlate actions with traffic gains.
Google runs content through about 200 signals and only 20 to 30 have been disclosed. Becker started in SEO and still rates it highly.
Becker rejects the idea that paid was once easy. Every 2010s unicorn used performance marketing, but not everyone who used it scaled. Thrive had projects that flatlined in that period.
3How Becker tests ad creative
Becker tests creative by changing one variable between two near-identical ads. In Meta he puts one audience in one ad set and runs two nearly identical creatives that differ in a single variable such as copy or an image. Because ad servers serve uneven impressions, he compares on a leveling metric like click-through rate or impressions until conversion.
One copy test found a version converting 50% more often at a given funnel stage, which became the base to beat. No exotic tool is needed.
Becker builds a testing process instead of waiting for one person's brilliance. He holds targeting constant, isolates variables and finds which style works at which funnel stage. Results differ dramatically brand to brand.
Airbnb found listing photos without people converted much better, likely because guests don't want strangers in the place they will sleep.
Becker matches different creative to each funnel stage. Brand teams often hand paid a file of random assets, so everyone sees the same ad. He plans a conversation that opens at the top of the funnel, shifts as people move through site behavior, and ends in an action.
One message to all targeting causes banner blindness and wastes spend.
Unpolished phone-shot creator ads beat polished brand ads in Thrive's tests. Brand-guideline assets underperformed on Instagram against iPhone videos of creators saying they tried and love a product.
Becker says UGC works because a third party validates the product instead of the brand saying "trust us".
A dog on the couch doubled or tripled ROAS for a furniture client. The client struggled on paid social with highly styled, model-filled room shots. An art director suggested putting the owner's dog on the couch to make ads more playful and approachable.
Becker points founder-led DTC brands to TikTok but warns about creative cost. Founder-led brands filming handheld iPhone videos, and brands that partner well with influencers, crush it there. TikTok's ad platform resembles Facebook ads 6 to 7 years ago, with cheap CPCs and weak attribution.
Brands launch net new assets several times a week, and reaching about $1 million a month takes a huge volume of creative. Few Thrive clients spend $750,000 a month on it. B2B SaaS has not cracked it yet.
4How Becker measures what paid really earns
Becker scores leads by downstream revenue instead of chasing the cheapest lead. He pipes CRM revenue into a database with a connector like Supermetrics, joins it with channel data and builds a lead scoring model for bidding. A higher cost per lead can bring a higher ROI.
This solves slow payback when leads take 2, 6 or 12 months to become revenue. Thrive also built an in-house tool, ThriveStack, that feeds a database and visualizes in Looker Studio. It is not commercially available. "Not all opportunities and sales are equal."
Becker calls attribution a subjective art with some science in it. Last click, first touch and multi-touch each give credit to a different click, and the weights come from the business model and goals. The first decision is whether the company is chasing profitability or growth.
He cites John Wanamaker in 1919 saying half his marketing was wasted and he didn't know which half.
Apple's IDFA change forced Becker's team to validate from several angles. In mid-2021 Apple let users decline to share the advertiser ID, which broke the validation Facebook relied on. Cookie-based attribution is now one input beside media mix modeling, statistical models and customer and population surveys.
Recast is an off-the-shelf MMM tool, and many firms build bespoke models.
Media mix modeling has shown Becker's own attribution models were measuring campaigns that don't work. Thrive built sophisticated attribution models that MMM then showed were crediting failing campaigns.
He treats that as a useful finding because it lets a client move budget elsewhere. "There's no single source of truth."
5How Becker hires, pitches and uses AI
Becker hires technical problem-solvers over years of experience. A first performance hire must solve tracking, attribution, data visualization and campaign management problems. Thrive staff came from nuclear physics, mathematics, finance and full-stack engineering.
He took apart computers as a child. He has seen 10-year veterans be mediocre and one-year people be phenomenal.
Becker asks how many windows are in New York City to test composure. He borrowed the question from Jeff Bezos. The answer matters less than whether candidates notice cars and trucks have windows and can do the arithmetic.
He watches whether they stay composed or get frustrated, as a sign of how they'd act with a client.
Becker advises hiring an in-house marketer first, then adding an agency. Without a marketing point of contact, agencies end up reporting to a busy CEO who can't give approvals fast enough. Thrive has no project without one.
Early on, an agency runs new experiments on audience sequence, creative and channel mix. Later it maintains a machine and adds testing and attribution sophistication.
Becker won Snap by refusing to answer its RFP questions. The RFP had 20 to 30 zigzagging questions. At 2:00 a.m. the night before, he and partner Brent McArthur decided to build a bespoke deck, worked until about 6:00 a.m. and pitched at 10:00 a.m.
Facing about 20 executives, he said if they agreed with his logic they should hire Thrive, and if not they shouldn't. The room went silent, and about two hours later Snap called to say they were hired. "I am not going to answer any of your questions today."
Becker uses generative AI for copy variants, RFP drafts and mock-ups. ChatGPT drafts RFP responses that are 80% good, and about 10 hours of editing replaces a week of work by five or six people. DALL-E and Midjourney produce mock-ups in about 1% of the time.
Thrive has laid off no one and has more people than ever, spending time on modeling, validation and creative levers rather than keyword-level bid modifiers. The user still has to know which questions to ask.
Mastering paid growth | Jonathan Becker (Thrive Digital)
Paid growth works best as one part of a diversified channel mix, run with rigorous creative testing, lead scoring tied to revenue, and attribution that never relies on a single source of truth. Thrive has deployed over $3.5 billion in paid budgets for Uber, Asana, Square and MasterClass, so the playbook comes from real scale.
Becker moved from SEO into paid search because clients wanted provable ROI. He started as a web developer who structured pages to rank in organic results. Clients kept asking about the ROI of SEO and how to scale it, and the answers pointed to paid search.
Google was removing the data SEOs used to guess and test, so organic felt intangible. Paid search gave him a tangible lever.
A freelance side business that began in a walk-in closet became a 130-person agency. Thrive now manages about $500 million a year in ad spend for clients including Uber, Asana, Tempur-Pedic and Lululemon. It is based in Vancouver, though most clients are in the United States.
Becker won Uber by confessing the loophole he was exploiting. In 2013 he went to the TED conference in Long Beach while running a 10-person agency. Garrett Camp asked to share his cab, and Becker told him he had bought Uber's branded keywords with a snippet that looked like Uber's organic listing, earning $20 per referral signup and tens of thousands of dollars.
He told Camp it added zero value and someone should close it. Uber's team first told him to stop, then hired him to fix it. Uber stayed a client for 10 years and moved Thrive from local bars and clothing stores to sector leader.
Becker says you make your own luck by risking embarrassment. He knew no one in the TED community and had little to lose, so he spoke up.
Becker runs client budgets like a fund manager, spreading across channels. He compares one-channel spending to putting life savings in one stock, exposed to CPC swings and market conditions nobody controls. His mix includes email, direct mail, SEO, linear TV and performance marketing.
He calls the goal "responsible channel mix diversity". Loopholes in one platform come and go, and he has seen brands scale on one and crash when it closed.
Becker says dependence, not paid itself, is what makes paid dangerous. Paid becomes a drug only when it is the sole source of revenue, which turns performance into a roller coaster.
Becker checks product-market fit and other working channels before betting on paid. Early on, he asks whether product-market fit exists and whether the company risks over-relying on paid. If email, SEO, influencer reach or direct mail already work, paid is highly likely to work.
Later-stage companies need creative resources, in-house marketers and engineers who can fix broken tracking.
Becker says know your maximum acquisition cost before scaling spend. Winners model how much they can afford to pay for a customer on any channel before it turns unprofitable. Grammarly and Athletic Greens know their impressions, their average quality and the spend level where ROI hits a cliff.
The focus is bottom-line profitability, not breakneck growth. Slow-payback businesses model LTV to CAC.
Becker recommends SEO and paid together, and likes paid for traceability. Finance and C-suite teams prefer paid because spend maps to revenue, so they can budget and project. With SEO you can only correlate actions with traffic gains.
Google runs content through about 200 signals and only 20 to 30 have been disclosed. Becker started in SEO and still rates it highly.
Becker rejects the idea that paid was once easy. Every 2010s unicorn used performance marketing, but not everyone who used it scaled. Thrive had projects that flatlined in that period.
Becker tests creative by changing one variable between two near-identical ads. In Meta he puts one audience in one ad set and runs two nearly identical creatives that differ in a single variable such as copy or an image. Because ad servers serve uneven impressions, he compares on a leveling metric like click-through rate or impressions until conversion.
One copy test found a version converting 50% more often at a given funnel stage, which became the base to beat. No exotic tool is needed.
Becker builds a testing process instead of waiting for one person's brilliance. He holds targeting constant, isolates variables and finds which style works at which funnel stage. Results differ dramatically brand to brand.
Airbnb found listing photos without people converted much better, likely because guests don't want strangers in the place they will sleep.
Becker matches different creative to each funnel stage. Brand teams often hand paid a file of random assets, so everyone sees the same ad. He plans a conversation that opens at the top of the funnel, shifts as people move through site behavior, and ends in an action.
One message to all targeting causes banner blindness and wastes spend.
Unpolished phone-shot creator ads beat polished brand ads in Thrive's tests. Brand-guideline assets underperformed on Instagram against iPhone videos of creators saying they tried and love a product.
Becker says UGC works because a third party validates the product instead of the brand saying "trust us".
A dog on the couch doubled or tripled ROAS for a furniture client. The client struggled on paid social with highly styled, model-filled room shots. An art director suggested putting the owner's dog on the couch to make ads more playful and approachable.
Becker points founder-led DTC brands to TikTok but warns about creative cost. Founder-led brands filming handheld iPhone videos, and brands that partner well with influencers, crush it there. TikTok's ad platform resembles Facebook ads 6 to 7 years ago, with cheap CPCs and weak attribution.
Brands launch net new assets several times a week, and reaching about $1 million a month takes a huge volume of creative. Few Thrive clients spend $750,000 a month on it. B2B SaaS has not cracked it yet.
Becker scores leads by downstream revenue instead of chasing the cheapest lead. He pipes CRM revenue into a database with a connector like Supermetrics, joins it with channel data and builds a lead scoring model for bidding. A higher cost per lead can bring a higher ROI.
This solves slow payback when leads take 2, 6 or 12 months to become revenue. Thrive also built an in-house tool, ThriveStack, that feeds a database and visualizes in Looker Studio. It is not commercially available. "Not all opportunities and sales are equal."
Becker calls attribution a subjective art with some science in it. Last click, first touch and multi-touch each give credit to a different click, and the weights come from the business model and goals. The first decision is whether the company is chasing profitability or growth.
He cites John Wanamaker in 1919 saying half his marketing was wasted and he didn't know which half.
Apple's IDFA change forced Becker's team to validate from several angles. In mid-2021 Apple let users decline to share the advertiser ID, which broke the validation Facebook relied on. Cookie-based attribution is now one input beside media mix modeling, statistical models and customer and population surveys.
Recast is an off-the-shelf MMM tool, and many firms build bespoke models.
Media mix modeling has shown Becker's own attribution models were measuring campaigns that don't work. Thrive built sophisticated attribution models that MMM then showed were crediting failing campaigns.
He treats that as a useful finding because it lets a client move budget elsewhere. "There's no single source of truth."
Becker hires technical problem-solvers over years of experience. A first performance hire must solve tracking, attribution, data visualization and campaign management problems. Thrive staff came from nuclear physics, mathematics, finance and full-stack engineering.
He took apart computers as a child. He has seen 10-year veterans be mediocre and one-year people be phenomenal.
Becker asks how many windows are in New York City to test composure. He borrowed the question from Jeff Bezos. The answer matters less than whether candidates notice cars and trucks have windows and can do the arithmetic.
He watches whether they stay composed or get frustrated, as a sign of how they'd act with a client.
Becker advises hiring an in-house marketer first, then adding an agency. Without a marketing point of contact, agencies end up reporting to a busy CEO who can't give approvals fast enough. Thrive has no project without one.
Early on, an agency runs new experiments on audience sequence, creative and channel mix. Later it maintains a machine and adds testing and attribution sophistication.
Becker won Snap by refusing to answer its RFP questions. The RFP had 20 to 30 zigzagging questions. At 2:00 a.m. the night before, he and partner Brent McArthur decided to build a bespoke deck, worked until about 6:00 a.m. and pitched at 10:00 a.m.
Facing about 20 executives, he said if they agreed with his logic they should hire Thrive, and if not they shouldn't. The room went silent, and about two hours later Snap called to say they were hired. "I am not going to answer any of your questions today."
Becker uses generative AI for copy variants, RFP drafts and mock-ups. ChatGPT drafts RFP responses that are 80% good, and about 10 hours of editing replaces a week of work by five or six people. DALL-E and Midjourney produce mock-ups in about 1% of the time.
Thrive has laid off no one and has more people than ever, spending time on modeling, validation and creative levers rather than keyword-level bid modifiers. The user still has to know which questions to ask.