TL;DR
Shopify launched Campaign Autopilot in June 2026. It plans and runs your ads across Meta, Shop, and email from inside your admin. Thryv launched its AI Growth Platform on August 4, 2026, claiming a 40 percent revenue lift per client from AI-scored leads. Both platforms want the same thing: your customer data feeding their model, and your decisions running through their dashboard.
Convenience is real. So is the cost. The rule is simple.
Rent the execution. Own the data and the decision logic. That's the Sovereignty Stack. Read Shopify's own announcement here: shopify.com/blog/introducing-campaign-autopilot.
The Trend Nobody Is Naming Correctly
Everyone is calling this "AI marketing." That's the wrong name. This is a control grab, and it's happening in plain sight.
Shopify didn't just add an AI feature. It renamed the Marketing tab in its admin to "Growth." Read that again. A tab rename sounds cosmetic.
It isn't. It signals where Shopify wants your attention to live, permanently, inside their walls.
Campaign Autopilot connects to your Meta ad account, your Shop Campaigns, and your email automations. It reads your products, your customers, and your sales history to decide what to promote and to whom. Shopify says it "gets smarter over time" because it sees patterns "across millions of stores." That's not a feature description. That's a description of a data flywheel, and you're the fuel.
Thryv did the same thing from the other direction. On August 4, 2026, it announced its shift from a directory company into what it calls an "AI-native growth platform," serving more than 100,000 small businesses. Its AI Lead Insights tool scores leads and claims a 40 percent revenue lift per client, with a 1.5x conversion rate versus non-scored leads. Source the number yourself: stocktitan.net/news/THRY.
HubSpot moved the same direction in July 2026 with Agent Hub and Agent Builder, now in public beta for Professional and Enterprise customers. The pitch: stop scattering AI agents across separate tools. Let the CRM become "the control layer, not just the database underneath the work," in the words of HubSpot's chief product officer, Duncan Lennox.
Three different companies. One instinct. Whoever holds the data gets to run the decision logic. Whoever runs the decision logic owns the relationship.
The Numbers Behind the Land Grab
Here's what's actually documented, not guessed at.
Shopify's Campaign Autopilot shipped June 17, 2026, as part of Spring Edition, a release with more than 150 updates. It's free on every paid plan except the Agentic tier. You pay only for the ad spend it places, on Meta, Shop Campaigns, and email, with Microsoft Advertising, ChatGPT Ads, and Snapchat coming after. Shopify recommends 30 to 45 days of runway before you judge results, because the system needs your data to learn.
Thryv's claim comes from an internal analysis of about 6,000 leads tracked between January 1 and April 22, 2026, inside Thryv's own Marketing Center and Business Center systems. That's a real dataset. It's also a vendor grading its own homework. Worth noting before you repeat the 40 percent figure to a client.
Meanwhile, data portability research from 2026 tells a colder story. One analysis found that most SaaS tools export roughly 60 percent of your data at best, full stop, regardless of vendor promises. Read it here: propicked.com/blog/data-portability-saas-2026. Subscriber lists travel.
Segmentation logic, behavioral scoring, and automation rules mostly don't. HubSpot's standard export, for instance, covers contacts, companies, and deals. It does not export smart list definitions, workflow logic, or AI-generated lead scores.
Fivetran's research calls this the gap between "ownership" and "control": your contract may say you own the data, but the platform still decides how, and whether, you can move it. Full piece: fivetran.com/blog/who-really-controls-your-data.
Constant Contact launched its own automated marketing suite the same week HubSpot expanded Breeze AI, in late July 2026. Same week, three vendors, one direction. When Adweek covers three competing "AI growth" launches inside a single news cycle, that's not a trend anymore. That's a land rush, and the thing everyone is racing to claim is the same parcel: your customer list and the logic that scores it.
Case Study: The Store That Let Autopilot Drive
A mid-size apparel merchant I know runs on Shopify, sub-$50,000 a month in ad spend, no dedicated media buyer. Exactly the profile Shopify built Campaign Autopilot for. They turned it on in early access. Meta budget allocation improved fast.
Attribution got sharper too, because Autopilot pulls ROAS from Shopify's own order data instead of Meta's self-reported numbers, a genuinely useful fix documented by independent reviewers. See the full breakdown: wrkngdigital.com/post/shopify-campaign-autopilot-honest-review.
Here's the part that matters. The merchant never built a segment file outside Shopify. Every audience Autopilot created lived inside Shopify's black box. Every rule for who gets targeted, when, and with what offer sat inside Shopify's logic layer, not theirs.
Six weeks in, they asked a fair question. If we ever leave Shopify, what do we actually take with us? The honest answer was: order history, a CSV of contacts, and nothing resembling the decision logic that made the last six weeks work.
The intelligence stayed behind. The store didn't own a customer list. It rented a customer relationship, priced in convenience.
That's not a Shopify problem specifically. It's the platform-autopilot problem, and it will repeat at Thryv, at HubSpot, at every vendor racing to make itself the operating system for your growth.
Risk and Caveats: Read the Guardrails, Not the Marketing
Three things every owner needs to check before flipping the autopilot switch.
First, ask what a "portable logic object" looks like for your top segment. Not a list of names, a file describing the actual targeting rule. If the vendor can't produce it, you don't own the intelligence, no matter what the contract says about data ownership.
Second, run the reconstruction test. If you migrated off the platform today, how long before your automated workflows perform at current levels? If the answer runs past three months, you have what data engineers now call an intelligence lock, distinct from a simple data lock.
Third, remember that none of these platforms guarantee outcomes. Shopify says so directly in its own product documentation: no marketing tool can promise specific results, and Campaign Autopilot needs time to learn your store.
Thryv's 40 percent figure comes from Thryv's own dataset, over a four-month window, on leads inside its own systems. Good data. Not independent data. Treat every vendor benchmark as a starting hypothesis, not a warranty.
Fourth, watch the pricing curve, not just the feature list. Campaign Autopilot is free today because Shopify needs merchants to feed it data while the model is young. Thryv offers a free 30-day trial for the same reason.
Free access to a data flywheel is a launch tactic, not a permanent policy. Every platform in this category will eventually price based on how dependent you've become, not on what the tool actually costs to run.
Doctrine Connection: Systems Beat Slogans
I spent four years standing watch on a submarine. You learn fast that the crew doesn't survive on confidence. It survives on documented procedure, casualty drills you've run so many times your hands know them before your brain does, and systems that work whether or not any one person is in the room.
Marketing autopilot sells confidence. "Set your budget, approve what runs, learn as you go." It sounds like a casualty drill someone else already wrote for you. It isn't. It's someone else's engine room, and you don't have the diagrams.
This is the whole argument behind the Sovereignty Stack. Split every marketing system into two piles.
Pile one: execution, the ad platform, the send engine, the landing page host. Rent that. Switch vendors when a better one shows up.
Pile two: the data, the segmentation logic, the attribution model, the list itself. Own that, in a warehouse or CRM you control, in formats any tool can read.
I learned this lesson the expensive way outside of marketing entirely, watching the Hartford-Munich Re relationship up close. Hartford built enormous distribution value over decades. But the underwriting logic, the actual decision rules for what made a good risk, lived partly in reinsurer systems and partly in institutional memory that walked out the door with retiring underwriters.
When leadership finally tried to systematize it, they discovered how much of their real intelligence had never been owned in a portable form. Same failure, different industry.
Dan Kennedy used to say the money is in the list, not the platform that mails it. He was right thirty years before AI agents made platforms greedy for that list. Systems beat slogans because a system is something you can audit, export, and rebuild. A slogan, "AI-powered growth," is something you rent until the vendor changes the price.
I've watched this play out at the top of the market too. AIN, the insurance network I helped build, crossed a billion dollars in premium by insisting every agency in the network kept its own book of business, its own carrier appointments, its own client data, even while sharing centralized systems and buying power.
Agencies that outsourced their client relationships to a managing general agent's platform got scale fast and an exit multiple slow. Agencies that kept the data and the carrier relationships in-house became acquirable. The platform is not the asset. The list, the logic, and who controls both, that's the asset.
Ownership beats wages applies here too. A merchant who owns the segment logic has a marketing asset that compounds and can be priced into a sale. A merchant who rents the logic from Shopify or Thryv has an expense that evaporates the day they cancel.
One of those things is acquirable. The other is a subscription with a UI.
FAQ
Q: Should I turn off Shopify Campaign Autopilot or Thryv's AI Lead Insights? No. Turning off a useful execution tool to prove a point costs you real revenue for no gain. Use the tool for what it's good at, budget allocation and channel testing, while building your own record of segments and results outside the platform.
Q: What data should I insist on owning, at minimum? Customer and lead records, consent logs, the actual segmentation rules in a portable format like SQL, and any predictive scoring model along with its training data. Raw contact lists are the easy part. The logic behind who gets targeted is the part vendors quietly keep.
Q: Is Thryv's 40 percent revenue claim credible? The underlying study is real, roughly 6,000 leads tracked from January to April 2026 inside Thryv's own systems. It's not independently audited, and it measures Thryv's own scoring against Thryv's own baseline. Treat it as a plausible upper bound for engaged users, not a guarantee for your business.
Q: Does the EU Data Act or GDPR force platforms to give me my data? Partially. The EU Data Act, effective since September 2025, extends portability rules toward business-generated data, and GDPR already covers personal data. Neither regulation requires a platform to export your automation logic, segmentation rules, or AI model weights, because those are classified as business configuration, not personal data.
Q: What's the single biggest mistake owners make with platform autopilot tools? Building their best-performing segments only inside the platform's UI. If your top 20 percent of customers exist only as a "smart list" inside Shopify or HubSpot, you don't have an asset. You have a feature you're renting, and the rent goes up whenever the vendor decides it should.
*Disclosure: Jeff Barnes has no personal position in any company named in this article. demg.ai provides marketing education, not investment advice.*