Every category of marketing software is being stress-tested right now in a way it never has been before. Not by a competitor with a slightly better product. By a foundation model that does the boring 60% of the job for free, run by a buyer who has stopped wanting to pay for anything that looks like a thin wrapper.
I’ve spent the last six months mapping which tools in each category I think actually survive this, meaning that in three years they still have a defensible market position and a customer base that isn’t shrinking. I’ve also been mapping which categories I think get largely consumed, where the standalone tool gets reduced to a feature inside a CRM, a foundation model, or a vertical platform.
This is opinionated. I’ll name specific products and explain what about their positioning makes them durable or fragile. Disagree with the calls and tell me why.
Before going through the categories, the rule that runs through all of this. A marketing SaaS tool survives the AI wave if it owns proprietary data, sits in a workflow with real switching costs, or is a system of record that humans need to coordinate around. It dies if its core value was wrapping a generic capability in a nicer UI. The generic capability is now a $20 a month chat interface, and the UI advantage compounds at zero.
That’s the lens. Now the categories.
Social Media Tools
The category most exposed to commoditization, and the one where I have the strongest opinions.
The job-to-be-done for most social tools is write a caption, schedule the post across networks, and pull a basic engagement report. That entire job is now table-stakes inside ChatGPT, inside Meta and LinkedIn’s native publishing tools, and inside any half-decent agency workflow built on Claude plus a queue.
Hootsuite is the cleanest loss in the category. The original positioning was “manage all your social accounts in one dashboard,” which made sense in 2010 when each network had a primitive web app and no analytics. That positioning never evolved. The product became a kitchen sink priced for enterprise but largely sold to SMBs and agencies who used a fraction of the feature surface. Once the underlying job got cheap, the bloated price-to-value got brutal. There’s no obvious rescue from where they sit.
Buffer is in a different kind of trouble. The product is genuinely good and the brand is loved by SMBs and creators. But the entire positioning is “the simple, beautiful way to schedule social posts,” and simplicity competes directly with free. When the alternative is “ask ChatGPT to write five LinkedIn posts and paste them into LinkedIn’s native scheduler,” paying $15 a month for a prettier scheduler gets harder to justify every quarter. Buffer survives if they pivot harder into creator monetization tooling. They struggle if they keep selling scheduling.
Later sits in the same trap. The link-in-bio asset was a genuine moat for a few years, and link-in-bio is now a feature in every creator tool and platform.
The survivors are the ones that built listening, governance, or vertical depth into the spine of the product.
Sprout Social wins because the positioning evolved past scheduling. The actual sale today is social customer service, social listening, and the cross-functional workflow that connects marketing, support, and brand teams. The Salesforce and Zendesk integrations matter. The Listening data set matters. A buyer can’t replicate any of that with a foundation model and a Zapier connection. The product has a real reason to exist that isn’t “schedule a post.”
Sprinklr at the enterprise end is essentially a workflow and compliance layer for organizations that have hundreds of accounts, regulated industries, and legal review on every public message. Pfizer doesn’t switch off Sprinklr because ChatGPT can write a tweet. The positioning is risk management dressed as marketing software, and risk management is sticky.
Khoros and Brandwatch hold for similar reasons in adjacent enterprise jobs, listening and community respectively, though Khoros has been quieter and the question is whether the parent company invests.
The interesting niche survivors are the ones that own a data set or a vertical workflow. CreatorIQ, Tagger (now Sprout), and GRIN in influencer marketing have a real data moat in creator identification and campaign attribution. Tools like Iconosquare and Dash Hudson hold pockets in beauty and fashion brands because they go deep on the visual analytics those teams need.
If your social tool’s pitch in 2026 is “schedule across platforms and use AI to write captions,” the AI ate you eighteen months ago. You just haven’t had the funeral yet.
Email Marketing Tools
This category splits cleanly. The basic email service providers competing on “send a newsletter, see open rates” are getting pinched from both sides. Substack and Beehiiv took the creator end with a creator-economy business model the old players couldn’t match. The AI-native CRMs are taking the small-business end by bundling email into a broader sales workflow.
Mailchimp is the slow-motion casualty. The brand is iconic, the installed base is enormous, and the product has been lapped on every dimension that matters since the Intuit acquisition. The positioning drifted from “email made approachable for small business” to “all-in-one marketing platform” without the depth to back the new pitch. Klaviyo eats their e-commerce share. HubSpot eats their B2B share. Beehiiv eats their newsletter share. Constant Contact and Mailerlite hold the lower end on price. Mailchimp will exist in 2029, but as a smaller business serving a stickier core, not as a category leader.
Constant Contact and Campaign Monitor are in the same boat with less brand to spend down. Drip had a genuinely interesting product for e-commerce marketers and got squeezed when Klaviyo built scale.
The winners have deep first-party customer data and behavioral triggering that’s expensive to rebuild.
Klaviyo wins in e-commerce on three structural advantages. The Shopify integration is the deepest in the category, the predictive analytics are tied to real purchase data the model can act on, and the SMS and review products extend the customer record across more channels. The positioning is honest. They don’t claim to be everything. They claim to be the customer data platform for consumer brands, and they execute on that.
Customer.io wins in product-led B2B and consumer apps because developers like the data model and the trigger logic is genuinely complex enough that a generalist AI assistant can speed up the work but not replace the operator. The positioning, “messaging built for product teams who care about the data layer,” scares off the wrong buyers and attracts the right ones. That’s a healthy positioning.
Braze and Iterable hold the enterprise B2C tier on a similar argument. Real-time data activation across mobile, web, push, and email is harder to stand up than it looks, and a Fortune 500 lifecycle marketing team is buying confidence as much as software.
Kit, formerly ConvertKit, survives in the creator economy by shaping the product around how creators actually work, with subscriber tagging by source, automations built on what subscribers do, and commerce features that turn email into a revenue tool. The positioning is creator-first in a way the bigger players can’t credibly fake.
The new entrants worth watching are the developer-infrastructure email companies. Resend, Loops, and the next generation of API-first sending platforms are growing because the buyer shifted. Marketing engineers and PLG ops teams want infrastructure they can drive with code and AI agents, not a drag-and-drop builder. Whoever wins this layer will be a meaningful business by 2028.
The piece of the email category that AI swallows is the writing layer and the basic flow building. Subject-line generators are toast. Standalone deliverability copywriting services are toast. The infrastructure layer, including sending, deliverability, segmentation, and identity resolution, gets stronger because it’s exactly what AI agents need to call into when they take over campaign execution.
Marketing Automation
The classic mid-to-enterprise marketing automation stack includes HubSpot, Marketo Engage, Pardot (now Marketing Cloud Account Engagement), ActiveCampaign, and Eloqua.
This category survives almost intact, but for a counterintuitive reason. The actual workflow building work, including designing nurture programs, building landing pages, and scoring leads, is increasingly something an AI agent can do faster than a marketing ops person. So why don’t the tools die? Because they’re systems of record. They’re where the lead lives, where the campaign attribution lives, and where the integrations to Salesforce, Snowflake, and the data warehouse live. Switching off Marketo isn’t a copywriting decision. It’s a six-figure project with risk attached.
HubSpot is the strongest survivor in this group, and the reason is positioning more than product. They committed to being the operating system for sub-$50M-revenue companies’ go-to-market, and then built every adjacent product (CRM, service, content, payments, commerce) into the same data model. By the time the AI wave arrived, they had a full workflow that the AI couldn’t easily disintermediate because the AI would have to replace four products at once. That’s a structural moat built on positioning discipline.
Marketo Engage holds the enterprise because Adobe is selling the bundle and large buyers want one throat to choke for the marketing cloud, the analytics, and the experience platform. The product has stagnated relative to where it could be, but the switching cost is real and Adobe’s enterprise sales motion keeps the renewals coming.
Pardot is the weakest brand in the bunch. Salesforce keeps half-burying it under Marketing Cloud rebrands, the product hasn’t kept pace, and the customers who’ve stayed have done so because ripping it out of Salesforce is a project nobody wants to staff. The positioning is muddled. Buyers don’t know what to call it, what it does, or who it’s for. Salesforce will keep collecting the revenue, but it isn’t winning new logos against HubSpot or Marketo.
Eloqua is in a similar quiet-decline mode inside Oracle. The installed base pays the bills.
ActiveCampaign occupies a tricky middle ground. The product is good. The positioning is fuzzy. They survive if they get very specific about who they serve, which is small businesses that need real automation but can’t afford or operate HubSpot. If they keep trying to compete with HubSpot on breadth, they get crushed because HubSpot will out-resource them on every front.
Drift and Qualified, the conversational marketing tools that adjacent to this category, are the most interesting case study in positioning collapse. They built a product around the idea that website chatbots and human SDRs handing off conversations would replace forms. The AI wave broke that thesis in the worst possible way. Now the buyer can run a far better website agent on a foundation model, and the legacy SDR-handoff workflow looks dated. Drift got acquired by Salesloft. Qualified is repositioning hard around AI agents for inbound. Whether the brand and the buyer mindshare survive the pivot is genuinely unclear.
The new entrants worth watching are the AI-native CRMs, including Attio, Day.ai, Folk, and the latest Salesforce Agentforce variant. The thesis is that an AI-first system of record has a structural advantage over a system of record that bolted AI on. I think that’s directionally right. The switching costs on installed-base CRMs are so high that the incumbents have a lot of time to respond, but the wedge is real and Attio in particular is winning specific buyer segments (technical founders, GTM engineers) that historically went to HubSpot by default.
Demand Generation
Demand gen is where the most interesting fight is happening, because the category was already mid-disruption before AI arrived.
The category includes intent platforms (6sense, Demandbase, Bombora at the data layer, RollWorks), data providers (ZoomInfo, Apollo, Clay, Cognism, Lusha, RB2B, Warmly), and the prospecting and outbound layer (Outreach, Salesloft, plus dozens of newer tools).
Intent data wins. 6sense and Demandbase have proprietary signals (bidstream data, third-party publisher activity, co-op data, account-level intent) that AI cannot generate by being clever. They’re effectively data brokers with a workflow on top, and the data is the moat. Both survive. Bombora sits underneath much of this as the actual data source for many platforms, which is a quiet but durable position.
Pure prospecting tools without proprietary data are in trouble. Apollo has a real data set and a strong product, but it’s being squeezed by buyers who can use ChatGPT plus a $30 LinkedIn scraper to get most of the same coverage for SMB prospecting. Apollo’s response, building out the workflow and the engagement layer, is the right move. The question is whether they can re-position from “cheaper ZoomInfo” to “the GTM platform for the bottom and middle of the market” before the price-driven buyers churn.
Cognism and Lusha are both fighting on the same battlefield with the same problem. Cognism’s GDPR-compliance positioning gives it a real edge in Europe that pure American players can’t easily match, which is enough to keep growing. Lusha is the most exposed of the bunch.
Clay is the most interesting story in the category. The positioning is exactly aligned with where the market is going, which is a workbench for GTM engineers who want to compose data sources, AI enrichment, and outreach into custom workflows the legacy tools can’t express. Clay grows. They’ve built the right product for the right buyer at the right moment, and the moat is the workflow library and the integrations more than the data.
ZoomInfo is the most interesting bear case in marketing software. The data is real, the moat exists, the financials still work, but the buyer narrative has turned actively hostile. Renewal conversations are brutal. Pricing power is eroding. The CEO has been honest about the AI threat, which is more than most public companies in this position do. ZoomInfo survives, but it gets meaningfully smaller before it stabilizes, and the eventual shape of the company is closer to a data layer feeding AI agents than a tool that humans log into.
The newer warm-prospect tools, including RB2B, Warmly, Common Room, are interesting because they’re building on a different premise. They identify visitors and intent signals at the individual level and route them into the existing CRM and sales workflow. They survive as features that get acquired into the bigger platforms, and a couple of them probably go public on their own.
The outbound execution layer (Outreach, Salesloft, Apollo’s sequencer) survives because sales teams need a system of record for sequences, calls, and pipeline activity. The integration into the CRM is sticky, the manager reporting is the actual sale, and the AI piece writes the email but the platform routes, tracks, and reports.
Content Marketing
This is the category where the biggest carnage happens.
Most AI writing assistants, including Jasper, Copy.ai, Writesonic, Rytr, and the dozens of “AI-powered” content tools that raised in 2023, were already commoditized by ChatGPT and Claude. Anthropic and OpenAI’s pricing alone makes most of them economically irrational. Jasper had a real shot because they were early and built brand templates and team workflows that the foundation models didn’t have natively. Then the foundation models got chat memory, projects, custom GPTs, and team plans. The category that calls itself “AI content” has, ironically, the worst defensibility against AI.
The survivors fall into three buckets.
First, the SEO data platforms (Ahrefs, Semrush, and to a lesser degree Moz) survive because they own crawled data sets that AI doesn’t have. You can ask Claude what a keyword’s search volume is, and it’ll guess. Ahrefs knows. The tools that own a data set the AI needs to call into get stronger, not weaker. Ahrefs is the cleanest example of patient compounding in marketing software. They built a data moat over a decade, never raised outside money, and they’re now perfectly positioned because the AI assistants need someone to source the underlying SERP data from. Semrush is in the same position with a bigger sales motion and more product surface.
Second, the enterprise governance and brand-voice layer survives. Writer.com is the cleanest example. Their positioning is enterprise AI for content teams with brand, legal, and compliance guardrails, which is exactly the job that gets bigger as more content is AI-generated, not smaller. Grammarly, after the Coda acquisition and the shift to enterprise productivity, sits in an adjacent zone where the embedded surface across every employee’s writing is the moat.
Third, the workflow and content operations players. Conductor and BrightEdge, the legacy enterprise SEO platforms, survive in their installed bases on switching costs but won’t win much new logo growth. Contentful and Sanity at the headless CMS layer survive because the underlying content infrastructure has to exist regardless of who or what is generating the content.
The middle gets squeezed hard. Surfer, Clearscope, MarketMuse, Frase, and the rest of the SEO content optimization tools are in a hard spot. The job of “tell me what to add to this article to rank for a keyword” is increasingly something the foundation model does natively when prompted with the SERP. The SEO data platforms above have started doing the optimization layer themselves. One or two of these tools survive by going deeper into a specific use case, like programmatic SEO at scale or enterprise content ops with workflow approvals. The rest get acquired or quietly shut down.
Optimizely’s content marketing module and similar all-in-one suites are in a different version of the same trap. The pitch was “manage your whole content operation in one platform,” and AI is collapsing the operation into fewer steps that don’t need a heavy platform.
If you’re a content marketer reading this, the implication is uncomfortable. Most of the tools you’re paying for to make content faster don’t have a future as standalone businesses. The work survives. The tools mostly don’t.
Local Marketing Tools
The local marketing software category, including Birdeye, Podium, NiceJob, Yext, Reputation, Vendasta, Thryv, Housecall Pro, and Jobber, looks completely different in three years, and most operators haven’t internalized it yet.
The piece of this category that survives strongest is the players bundling messaging, payments, reviews, and booking into one operational tool for a local service business. Podium is the cleanest example. The positioning isn’t “we manage your reviews,” it’s “we are the system you use to talk to your customers and get paid.” That’s a system of record for a plumber, a med spa, or a roofer. AI threatens the review-response feature inside the product. AI doesn’t threaten the system, because the system holds the customer relationship and the payment workflow.
Birdeye is in the same neighborhood with a slightly different positioning. They’ve leaned harder into the multi-location enterprise SMB segment (auto dealer groups, healthcare networks) where the buying decision is centralized and the analytics across locations matter. That’s a defensible position because the AI assistant can’t replace the cross-location reporting layer.
NiceJob and the cleaner SMB reputation tools probably hold their ground because they kept the product focused, kept the price reasonable, and serve a buyer who genuinely just wants more reviews and a simple way to ask for them. That’s a real job that doesn’t go away.
The bigger reputation suites, including Reputation and the upper end of Birdeye’s product, are under more pressure because they sold a complex platform at enterprise prices for a job that’s getting cheaper. They survive in the multi-location enterprise segment, but the SMB-and-mid-market push gets harder every quarter.
Pure listings management is in a hard place, and Yext is the case study. The original pitch, “syndicate your business listings to fifty directory sites,” made sense in 2014 when those directories actually drove traffic. Google increasingly trusts its own data, AI overviews bypass directories entirely, and the value of a Foursquare listing is approximately zero. Yext has been repositioning hard around enterprise brand answers and AI search visibility, which is the right strategic direction. The question is whether they can grow into the new positioning faster than the old positioning shrinks. The leading indicators are mixed.
Vendasta is interesting because the positioning is white-label software for agencies that sell to local businesses, which is a workflow position that survives the AI wave on the agency side even if individual products in the suite get pressured. The agencies aren’t going away, and they need a platform.
Thryv is positioned at the smallest end of the local SMB market, and the positioning has always been “everything you need to run your service business in one place.” That’s similar to Podium’s pitch with a different vertical bias. They survive in their core, but compete with Housecall Pro and Jobber, which are vertical-specific operating platforms for trades that are pushing back into the marketing layer from the operations side. ServiceTitan extends downward toward smaller HVAC and plumbing operators with the same compression effect. Vertical operating systems eating horizontal SMB marketing tools is the multi-year story to watch in this category.
The wild card is what local discovery and booking look like when AI agents do the job. If a homeowner says to ChatGPT “find me a good plumber near me with availability for Tuesday morning and book the appointment,” whoever owns that interface and that fulfillment layer wins. None of today’s local marketing tools are obviously positioned for that future. Whoever figures it out, whether that’s Google, a vertical operating platform like ServiceTitan extending into discovery, or a startup that doesn’t exist yet, eats the high-value layer of the category.
Pull the six categories together and the pattern is consistent. Tools survive if they own proprietary data the AI cannot generate, if they are a system of record that a team or a customer relationship coordinates around, or if they sit deep enough in a workflow that ripping them out is a project rather than a decision. Tools die, or at best get reduced to features, if their value was a UI on top of a generic capability that the foundation models now do for free or close to it. Positioning is what determines which side of that line a product lands on, because positioning shapes which buyers stay loyal when their costs go up and which ones bolt the moment a free alternative is good enough. If you’re a marketing leader making a 2026 software budget, that is the question to ask about every line item. If you’re a founder building in this category, that is the question that determines whether you have a business in three years or a feature someone else will eventually charge for.
Which of these calls do you think I have wrong?