Why This Matters to Cross-Border Sellers
You probably haven’t thought about link previews since the last time a product listing URL you pasted into Slack showed a blank gray card instead of a crisp image and price. But that broken preview is more than a cosmetic annoyance — it’s a failure in the data pipeline that powers your affiliate links, your product research scraping, your automated social posts, and your competitive analysis. Every time your team manually checks an Amazon product page to grab the title and image, you’re burning minutes that add up to hours. The new Link Preview API from the team behind Exabase solves exactly this pain point, and it does it with a free tier that’s generous enough to let you experiment without a budget request. For anyone running a multi-marketplace operation — Amazon, Shopify DTC, TikTok Shop, even Temu — the ability to reliably pull structured metadata from any URL is a quiet force multiplier. Let me tell you what this thing actually does, where it beats the old tools, and where I think it’s not ready for prime time yet.
What Link Preview API Actually Does (and Why It’s Not Just Another Scraper)
At its core, this is a hosted API that takes a GET request with any URL and returns structured JSON: title, description, Open Graph image, favicon, image dimensions, and site name. That sounds boring until you realize how many edge cases break the naive approach. Founder Johnny lays it out clearly in his launch post: “It’s extremely tedious and painful to cover all of the weird edge cases on the internet, and it shouldn’t be something you need to pay for when starting out.”
The API handles JavaScript rendering — critical when a site loads its metadata via client-side frameworks — and proxy rotation to avoid IP blocks. It also includes purpose-built integrations for sites that are notoriously unreliable: YouTube, Amazon, Twitter/X, and others. According to the team, they maintain a 95%+ success rate across the web. That number sounds plausible if they’re already running this at scale for their other app, Fabric, which is a bookmarking/reading tool. The free tier is capped at 20,000 requests per month, after which you presumably need to move to a paid plan (pricing beyond free is not disclosed on the launch page, but the business model is explained as a loss leader to draw users into Exabase’s broader extraction platform).
For a cross-border seller, think of this as a universal metadata extractor. You can feed it a Shopify product page, an Amazon listing, a TikTok Shop link, or an Etsy item — and get back a clean JSON object. No regex, no headless browser maintenance, no worrying about captchas. The API normalizes the output so you’re not writing fallback logic for every site that serves its OG tags differently.
How It Stacks Up Against Incumbents
The link preview space isn’t empty. There’s Embedly (acquired by Medium, often rate-limited), Microlink (strong but with a different pricing model), and the old standby of running your own headless Chrome with something like Puppeteer. Here’s where Link Preview API differentiates itself:
Free tier depth. Most services give you a few hundred free requests. 20,000 monthly requests for zero cost is unusual — and it’s enough to power a small-scale product research pipeline or an affiliate link enrichment system without a credit card. The founder explicitly said this is an entry point to the Exabase platform, which includes deeper document parsing and search. That’s a smart acquisition play, but for the user it means you get real value before you’re upsold.
Site-specific handlers. Andras Czeizel, a commenter on the launch, nailed it: “The dedicated handling for sites like YouTube, Amazon, Twitter/X, Airbnb, etc. is probably the real value here.” Generic scrapers often fail on Amazon because the page structure changes frequently, and the OG tags can be inconsistent. Having a handler that already accounts for Amazon’s quirks saves you weeks of trial and error. This is especially relevant for cross-border sellers who pull data from regional Amazon marketplaces (DE, JP, etc.) — though I’d want to test if the API handles localized URLs correctly.
TypeScript SDK. A commenter asked for a Node.js or Python SDK, and Johnny pointed to the existing TypeScript SDK. That makes it easy to integrate into a server-side Node app for batch processing. If you’re building a tool to prefetch product previews for your Shopify store’s blog posts or to generate social cards for your Amazon product links, you can drop in the SDK and start cronning immediately.
Where it falls short compared to incumbents: no obvious cache-control API. A commenter asked about forcing a cache refresh when a site updates its OG image. Johnny acknowledged the feedback and added it to the roadmap, but right now you’re at the mercy of their periodic recrawl. For use cases where freshness matters — like monitoring price changes on Amazon, where the OG image might update with a sale badge — that’s a real gap. Embedly and Microlink both offer force-refresh parameters.
What Cross-Border Sellers Can Borrow: Use Cases
This API isn’t going to replace your full-stack product research tool like Helium 10 or Jungle Scout. But it can fill specific cracks in your workflow:
Affiliate link enrichment. If you’re running a DTC brand with a blog that posts roundups of products you recommend (e.g., “Best travel backpacks for 2025”), you often embed Amazon affiliate links. Those links need rich previews — title, image, price. Instead of manually curating each, you can hit the Link Preview API to auto-generate the preview data. It’s a one-liner in your CMS.
Competitive product research. Pulling metadata from dozens of competitor product pages on SHEIN, Temu, or eBay to track descriptions and images over time. A cron job can snapshot the OG data daily. This is cheaper and simpler than running a full scraper with rotating proxies.
Automated social sharing cards. When your Amazon product goes viral on TikTok, you want the share link to display a beautiful card. Passing the product URL through the API ensures your social media manager doesn’t have to manually upload images for every post. This works for TikTok Shop links too.
Why Amazon Sellers Should Care More Than Shopify Ones
It’s counterintuitive, but Shopify merchants actually have less need for this. Shopify product pages are relatively predictable — you can scrape OG tags yourself in a few lines of JavaScript. Amazon, on the other hand, is a nightmare. The page is heavily dynamic, the metadata is often missing or incorrect, and Amazon actively blocks scrapers. The built-in Amazon handler in the Link Preview API is the killer feature for Amazon sellers. If you’re a brand owner selling on multiple Amazon marketplaces, you can use this API to normalize product data from .com, .de, .co.uk, and .jp without maintaining separate scrapers for each locale. The 95% success rate claim is more credible for Amazon than for the general web because the team has already battled that specific beast.
Where the Math Breaks: Limitations and Risks
I’m not sold on everything. Here are three points where the rosy picture gets murky for serious operators:
Scale costs are opaque. The free tier of 20,000 requests is generous for tinkering, but if you’re pulling 500 product links per day across 10 categories (that’s 15,000/month), you’re already bumping against the ceiling. What happens after? The founder says the cost is subsidized by Exabase’s paid platform, but there’s no published pricing for the link preview service alone. If you build a dependency on this API and it becomes expensive or gets shut down as a loss leader, you’re stuck rewriting your integration. I’d want a public pricing page before I commit to a production pipeline.
Cache freshness delays. For cross-border sellers, timing can be critical — especially around seasonal sales where Amazon slaps “Lightning Deal” badges on images. If the API’s periodic recrawl takes hours to pick up a change, your automated social cards will show outdated pricing. The lack of a force-refresh endpoint means you’re relying on their cache cadence. Until that’s added, this tool is best for static product data, not real-time monitoring.
Regional and language handling. The API may handle YouTube and Amazon well, but what about Temu (which is notoriously aggressive with bot detection) or Etsy (which often serves localized OG tags based on user IP)? The launch comments mention “purpose-built integrations for YouTube, Amazon, Twitter, and a fair few other sites.” That list may not include the Chinese or European platforms you rely on. Before scaling, you need to test URLs from your specific marketplaces. The 95% success rate across the web is impressive, but it’s an average — the long tail of problematic sites could include your most important ones.
The Cache Freshness Problem (A Deeper Concern)
This deserves a sidebar because it’s the most common failure in link preview tools. Gal Dayan’s comment on the business model highlights the tension: “proxy rotation and JS rendering at scale isn’t cheap, so is this free tier subsidized by Exabase’s paid platform?” The answer is yes — it’s a loss leader. That means the cache strategy is optimized for cost, not freshness. The API probably uses a shared cache across all users to reduce origin fetches. That’s fine for a blog post, but if you’re using it to populate a product feed that updates hourly, you’ll see stale data. My advice: combine the API with a fallback headless browser call for high-importance URLs, at least until the force-refresh feature is released.
What I’d Watch / Test Next
Here’s my concrete action plan for any cross-border seller reading this:
Sign up and test your marketplace URLs this week. Hit the Link Preview API with product pages from Amazon, eBay, Temu, and Etsy. Compare the output to what you’d get by manually inspecting the page’s OG tags. Pay special attention to images — Amazon often returns low-resolution or placeholder thumbnails. If the API consistently returns high-res images, it’s a win.
Cron a batch of 100 URLs and measure success rate. Use the TypeScript SDK to automate a quick test. Log the number of 200 responses vs. failures. If the failure rate exceeds 5% for your specific sites, that’s a red flag for production.
Review the Exabase platform for deeper extraction. The API is part of a larger suite. If you’re already doing product research and need full-page text extraction or document parsing, Exabase might replace multiple tools. But don’t upgrade until you’ve validated the free tier with your actual use case.
Set a reminder to check for cache-control endpoints in 30 days. The founder added it to the roadmap. If it ships, this tool becomes a serious player for real-time workflows. If it doesn’t, plan to layer a small caching proxy on your side to force fresh fetches.
Watch the pricing. If the free tier remains unlimited in spirit but starts throttling, move on to Microlink or Embedly. But for now, 20,000 free requests gives you a low-risk sandbox to see if link preview automation can save your team even one hour per week. That hour is worth more than the API call cost — and this week, it’s free.
