The future of car sales: how marketplaces, AI and social media will change

Over the last few months there's one idea I keep turning over in my head: what if the future of the car marketplace isn't really a marketplace at all?

For more than 20 years, the logic has been fairly simple. The more traffic a platform concentrated, the more buyers it attracted; the more buyers it had, the more dealerships wanted to be on it; and the more inventory it accumulated, the more it strengthened its position.

But I think we're entering a different stage. Demand is fragmenting across marketplaces, Google, social media and LLMs, while there's one thing that remains scarce: real, structured, up-to-date inventory that's available to buy.

And that may change where the real moat of the automotive marketplace lies.

From the physical market to the digital marketplace

For decades, buying a car meant physically going to a dealership, a used car dealer, a car lot, a fair or an auction. Then came newspapers and specialist magazines, with publications such as Autopista, Motor16, Auto Trader Magazine or La Centrale des Particuliers.

The information was basic, but a fundamental dynamic emerged: the medium didn't control the cars; it controlled the audience.

The internet multiplied that model. During the 2000s, large general classifieds sites grew, such as eBay, Craigslist, Gumtree, Milanuncios, Leboncoin, Subito, OLX or Marktplaats. In Spain, Milanuncios perfectly represents that stage: a single platform where flats, jobs, furniture, bicycles and thousands of cars sat side by side.

The next leap was verticalisation.

Verticals turn car search into a specialised product

Buying a car isn't the same as buying a sofa. The category needs structure.

This is when we see the rise and consolidation of companies such as Coches.net, AutoScout24, mobile.de, La Centrale, Auto Trader, CarGurus or Cars.com.

The vertical's advantage was clear. It could structure:

• make and model;

• version;

• mileage;

• fuel type;

• gearbox;

• power;

• equipment;

• location;

• price;

• financing;

• market value.

For years, the big win for these companies was becoming the starting point of the purchase.Demand aggregation was the moat.

Then came the next logical step: capturing more parts of the transaction with financing, valuation, trade-in, warranties, reservation, delivery or checkout. The evolution was clear: listing, lead and transaction.

The present is already hybrid

Today a person can start buying a car in many different places.

They can do it on Coches.net, AutoScout24, mobile.de or La Centrale, but also on Milanuncios, Leboncoin or Subito. They can start on Google Search or Bing, discover a car on Instagram, TikTok, Facebook or YouTube, research on Reddit, talk to ChatGPT, Gemini, Microsoft Copilot, Perplexity or Meta AI and end up contacting the dealership via WhatsApp.

The present is no longer “marketplace versus Google”. It's Marketplace + Search + Social + LLMs. And the future will be even more hybrid.

Inventory starts travelling to the buyer

For years we built websites to attract users. Now the opposite is starting to happen too: inventory travels to wherever the user is.

We already have very clear examples:

• Google Vehicle Ads, to distribute specific vehicles from the dealer's feed;

• Meta Automotive Inventory Ads, within the Facebook and Instagram ecosystem;

• TikTok Automotive Ads, including catalogue-based Inventory Ads;

• Microsoft Automotive Ads, within the Microsoft Advertising and Copilot ecosystem;

• ChatGPT Ads Product Feed Campaigns, to use structured inventory within conversational campaigns;

• Perplexity Merchant Program, geared towards conversational discovery and shopping.

The conceptual shift is important: the car stops living solely inside a portal and becomes a distributable unit.

A BMW X1 that comes into stock today can appear simultaneously on the dealer's website, Coches.net, AutoScout24, Google Vehicle Ads, Instagram, TikTok, YouTube, ChatGPT or Copilot. One vehicle, many channels.

Advertising is changing too

We're not simply adding new places to show banners. The very logic of advertising is changing.

Search Ads

The classic model of Google Ads and Microsoft Ads still captures explicit intent: the user searches for “used BMW X1 Madrid” and gets related ads.

Vehicle Inventory Ads

The dealer no longer manually creates one ad per car. It sends the inventory and the algorithm selects which vehicle to show. The clearest examples are Google Vehicle Ads and Microsoft Automotive Ads.

Social Inventory Ads

The same logic is reaching social media. Meta Automotive Inventory Ads and TikTok Automotive Ads use vehicle catalogues, but add behaviour and interest signals. Google may know what you're searching for; Meta or TikTok try to anticipate what might interest you even before you search for it explicitly.

Dynamic Creative Ads

The next step is automatically turning inventory into content. A vehicle comes into stock and the system can generate carousels, Stories, Instagram Reels, TikTok videos, YouTube Shorts, copy and creative variations. It's the natural evolution of social media for dealerships: Inventory → Creative → Distribution.

Conversational Ads and Product Feed Ads

Enter ChatGPT Ads, Google AI Mode, Gemini, Microsoft Copilot and other conversational interfaces.

A keyword like “used SUV Madrid” contains relatively little information. Someone might instead say:

“I have two kids, I live in Madrid, I drive 15,000 kilometres a year and I want to spend 28,000 euros at most. I want something premium but not too big.”

An LLM can understand budget, location, usage, size, preferences and constraints. The logic shifts from keyword targeting to intent matching.

Product feeds take this one step further: structured catalogue, availability and metadata connected directly to a conversation or a campaign. In automotive, the fit is obvious: stock feed + intent + relevant vehicle.

Sponsored Recommendations and AI Explainer Ads

Ads will also start to be integrated into the evaluation phase. Not just “buy this car”, but clearly labelled sponsored recommendations and AI-generated explanations of why a vehicle fits a specific need.

Dynamic Offers

Nor will the ad of the future have to be limited to the car. It will be able to include price, discount, financing, monthly payment, warranty, trade-in, maintenance or delivery. The system will be able to dynamically select the most suitable product and offer for each intent.

Local Inventory Ads

In automotive, location is fundamental. The next step will be to cross-reference intent + inventory + location: not just knowing that a BMW X1 exists, but that there's one available nearby and it's still in stock.

Creator + Inventory Ads

YouTube, TikTok, Instagram Reels and specialist media can increasingly connect editorial content with real inventory. A review of “the best used SUVs for under €25,000” can end up showing units available to buy right then and there.

Messaging Ads

WhatsApp, Instagram DM, Messenger or webchat can become direct entry points to the inventory. The user sees an ad, asks “is it still available?” and an agent with access to the real stock carries on the conversation.

Agentic Ads

And finally, a completely new category appears. The consumer has an agent working for them and tells it: “Find me a car”. The agent checks marketplaces, feeds, dealers, APIs and financing.

In that scenario, the advertiser is no longer trying to convince only the human. It's trying to get its inventory selected by another algorithm.

From Keyword to Agent

The evolution can be summed up like this:

• Keyword: what you've typed.

• Audience: who you are and what interests you.

• Behaviour: what behaviour signals intent.

• Intent: what you really need.

• Agent: which product it should choose for you.

That last step could profoundly change digital marketing.

The future won't be Marketplace or Social or LLM

I think it would be a mistake to try to identify a single winner.

A consumer might discover a Cupra Formentor on TikTok, ask ChatGPT whether it makes sense for a family, look up prices on Google, compare units on Coches.net, check AutoScout24, watch a review on YouTube and end up getting in touch via WhatsApp.

The journey will be omnichannel.

Marketplaces and DMSs need to work much more closely together

For years, the two layers have worked relatively independently. The dealer managed its vehicle in the DMS, exported a feed and the marketplace published the ad. I think we now have the opportunity to go a lot further.

Marketplaces have:

• audience;

• brand;

• trust;

• intent;

• knowledge of the buyer;

• discovery experience.

DMSs and Vertical SaaS have:

• real-time inventory;

• availability;

• price changes;

• CRM;

• operations;

• vehicle information;

• transactional context.

They're complementary assets. I don't think the future is “marketplace versus software”. I think it will be Marketplace + Vertical SaaS.

The marketplace understands and concentrates demand. The software keeps supply alive and structured. The better connected both layers are, the better the experience for the buyer and for the dealer.

If demand fragments, where's the moat?

Today, demand sits with Google, Meta, TikTok, YouTube, OpenAI, Microsoft, marketplaces, classifieds, creators and the dealers themselves. Distribution is multiplying.

But there's something that doesn't multiply.

A VIN only exists once.

A used car has a specific mileage, equipment, condition, price, location, dealer and availability. Its price can change this afternoon, it can be reserved within the hour and sold tomorrow.

That's why I think we're entering a scenario where Supply becomes scarce. Distribution becomes abundant. And that's where the new moat may emerge.

Structured Supply

Having a copy of the ad isn't enough. The important asset will be having structured, enriched supply updated in real time.

Something like a Vehicle Graph, with:

• VIN;

• make, model and version;

• mileage and equipment;

• price and price history;

• photos;

• location;

• dealer;

• availability;

• financing;

• warranty;

• history;

• reservation;

• days in stock;

• demand;

• margin;

• conversion;

• performance by channel.

With this layer, any interface —Google, ChatGPT, Gemini, TikTok, Meta, Copilot, a marketplace or a future agent— can check which vehicles meet certain conditions and get a real answer.

LLMs need supply

An LLM can be extraordinarily good at understanding what car a person needs. But it can't make up which units exist right now, where they are, how much they cost today, whether they're still available, how they can be financed or what warranty they come with.

Models can control the interface, but someone has to control the truth of the inventory.

And that's where Vertical SaaS has a huge opportunity.

Software connected directly to the dealership knows when a car comes in, when its price changes, when it's reserved, when it's sold, how many days it's been in stock, what leads it's getting and how it can be financed.

At Dealcar, for example, the dealer can already manage from a single environment stock, CRM, website and distribution to different channels. At first glance it's operational software. Strategically it can become supply infrastructure.

Multi-posting was just the beginning

For years, the product was easy to explain: enter the car once and publish it on ten portals. That was multi-posting.

The next step could be Intelligent Distribution: automatically deciding which car should appear on which channel, for which user, with which creative, with which offer and with which budget.

And then will come Autonomous Merchandising. If a car has been in stock too long, has an uncompetitive price or gets few leads, a system could recommend a price change, generate content, activate Google Vehicle Ads, launch TikTok Automotive Ads, make it eligible for ChatGPT Ads, reactivate old leads and measure the result.

That's no longer about publishing inventory. It's about orchestrating demand around inventory.

From the omnichannel marketplace to Agentic Commerce

The full evolution could be summed up like this:

• Marketplace 0.0 — Physical: dealerships, car lots, fairs and auctions.

• Marketplace 0.1 — Print: Autopista, Motor16, Auto Trader Magazine, La Centrale des Particuliers.

• Marketplace 1.0 — General Classifieds: eBay, Craigslist, Milanuncios, Gumtree, Leboncoin, Subito, OLX.

• Marketplace 2.0 — Vertical Marketplace: Coches.net, AutoScout24, mobile.de, La Centrale, Auto Trader, CarGurus.

• Marketplace 2.5 — Transaction: financing, trade-in, reservation, warranty and delivery.

• Marketplace 3.0 — Omnichannel: marketplaces + Google Search + Google Vehicle Ads + Meta + TikTok + YouTube + ChatGPT + Gemini + Copilot.

• Marketplace 4.0 — Agentic: the buyer stops searching and simply explains what they need.

The user could say “Find me a car” and an agent would check different sources, analyse thousands of vehicles, compare prices, review history, calculate financing and return a few alternatives.

The interface could belong to OpenAI, Google, Microsoft, Meta, Amazon, a marketplace or a new player. But the supply has to come from somewhere.

Three layers will define the future

I believe the market will end up organising itself around three big layers:

• Demand: Google, Meta, TikTok, OpenAI, Microsoft, Amazon, YouTube, marketplaces and creators.

• Supply: dealers, marketplaces, OEMs, DMS, Vertical SaaS, auctions and inventory platforms.

• Transaction: financing, payments, trade-in, insurance, warranties, documentation, reservation and logistics.

Whoever manages to connect those three layers well will hold an extraordinarily powerful position.

My thesis

I don't think marketplaces are going to disappear. Quite the opposite. Coches.net, AutoScout24, mobile.de, La Centrale or Auto Trader have assets that are extremely hard to replicate: audience, brand, trust, intent and knowledge of the buyer.

But they won't be alone any more. Buyers will be spread across traditional marketplaces + Google + Meta + TikTok + YouTube + ChatGPT + Gemini + Copilot + future agents.

That's why I think marketplaces and software need to get even closer. The marketplace brings demand. The DMS brings real-time supply. Together they can build an experience neither can build alone.

For 20 years the big message was: “I have the buyers.” I think the next big battle may revolve around: “I have access to supply.”

But not to a copy of the inventory or an outdated feed. To the source of truth: the place where the car is born, changes price, is reserved, sold and financed, and from where it can be instantly distributed to Coches.net, AutoScout24, Google Vehicle Ads, Meta Automotive Inventory Ads, TikTok Automotive Ads, YouTube, ChatGPT Ads, Gemini, Microsoft Copilot, WhatsApp and the agents that don't exist yet.

Supply becomes scarce. Distribution becomes abundant.

For decades we built places to bring buyers to. Now we're building infrastructure that can take inventory wherever buyers are.

And perhaps that's the real future of the automotive marketplace.

Not getting everyone to come and look for cars on your platform, but making sure that, wherever they search, the right offer is there.