Why Now<\/h2>\n\n\n\n
Two things pushed us to build this.<\/p>\n\n\n\n
First, our MCP server<\/a> (the natural-language interface to SparkToro that plugs into Claude, ChatGPT, Cursor, and any other Model Context Protocol client) has been on fire since launch. Customers are using it to build audience reports, explore section data, and reason over their audiences in plain English. That validated something we already suspected: people don’t just want to see<\/em> SparkToro data, they want to use<\/em> it \u2014 in the tools they already work in.<\/p>\n\n\n\n Second, the space around us has changed. LLMs, workflow tools like Zapier and n8n, AI agent builders, and modern data warehouses have all made the “pipe data between things” problem an order of magnitude easier, as long as the source has an API. Every workflow, every AI agent, every automation now assumes it can call any tool it needs. Making SparkToro one of those tools was overdue.<\/p>\n\n\n\n We’re not the first product to add an API. But we’re the first to make behavioral audience data a building block any workflow can call on demand<\/strong>: where a defined group of people spend time online, what they read, watch, listen to, and engage with.<\/p>\n\n\n\n One thing to be clear about up front: the SparkToro API is a separate product from the SparkToro web app<\/strong>. If you’re a Personal, Business, or Agency subscriber today, you’re getting a hands-on research tool tuned for the individual and team analyst use case. The API is priced and delivered independently.<\/p>\n\n\n\n We chose this split deliberately. App subscribers and API customers are two different kinds of people, using the data in two different ways:<\/p>\n\n\n\n Trying to serve both with a single subscription would compromise both. So the API is standalone: your own credits, your own billing, your own dashboard.<\/p>\n\n\n\n Here are the three starting points we think are most interesting. These all can be what happens when this data becomes something you can compose.<\/p>\n\n\n\n Let\u2019s say you run a marketing agency and a prospect just filled out your contact form. You’ve got 48 hours to deliver a standout, homerun pitch. Every other agency will show up with “here’s who we are and what we do.” You want to show up with “we already know your customers and how to reach them.”<\/p>\n\n\n\n Feed the prospect’s audience description into the API, pulled from their intake form or reverse-engineered from their website:<\/p>\n\n\n\n “Women 25-45 in the US who buy skincare from Sephora, Ulta, and independent beauty retailers. Follow beauty creators on TikTok and Instagram, care about clean ingredients and sustainability.”<\/p>\n\n\n\n Twenty minutes of scripted work later, you’ve got a branded one-pager for the pitch call: the top podcasts this prospect’s customers listen to, the YouTube channels they watch, the publications they read, the creators they follow. Not “here’s what we could do.” Instead: “we spent time on your buyers before this call, and here’s where their attention actually is.”<\/p>\n\n\n\n It turns a generic capabilities pitch into “we already researched your customers.”<\/strong> That’s a real close-rate lift, and it costs about 23 credits instead of an analyst’s afternoon.<\/p>\n\n\n\n You’re a marketing lead making budget calls for next quarter. A creator’s manager just pitched you a $30K sponsorship because their audience “aligns with your brand.” How do you actually know? What if the overlap with your ICP is 15%, not 60%?<\/p>\n\n\n\n Build your ICP audience once and cache the result:<\/p>\n\n\n\n “US-based indoor plant hobbyists aged 28-45 who visit thesill.com and pistilnursery.com, follow plant creators on TikTok and Instagram, listen to Bloom & Grow Radio, and shop Home Depot and Etsy for plant supplies.”<\/p>\n\n\n\n Now you’ve got a ranked list of newsletters, podcasts, and creators sorted by measured audience overlap with your ICP. Hand it to your media buyer as the sponsorship shortlist. Every new pitch that lands in your inbox gets sanity-checked against it: does this influencer’s audience actually match ours, or do they just show off vanity metrics?<\/p>\n\n\n\n Rerun the audience quarterly and diff against last quarter to catch shifts before your competitor does: a podcast climbing the top 10, a publication losing ground to a Substack, a creator whose audience just changed. It turns “who should we sponsor next quarter” from a gut-feel debate into a data question.<\/strong><\/p>\n\n\n\n You’re on the growth team at a B2B SaaS company. A few hundred companies sign up for your free tier every week, and every one of them gets the same onboarding sequence, because nobody has time to research a free signup.<\/p>\n\n\n\n An agent does. Wire one into your signup flow: when a new account comes in, the agent takes the signup’s company domain, asks the API who that company’s audience is, and writes the enrichment straight into your CRM. Where that audience spends time, which podcasts and publications reach them, which of your templates and case studies actually fit. Your onboarding email now references the customer’s actual market instead of “your industry.” Your sales-assist queue sorts itself by audience fit.<\/p>\n\n\n\n This is the use case where the difference between the MCP server and the API gets obvious. The MCP server is for a person asking questions in a chat window. This agent isn’t asking questions; it’s running the same research playbook hundreds of times a week, unprompted, while your team does something else.<\/strong> No one would (or should) sit in a chat tool doing that by hand.<\/p>\n\n\n\n Notice what these three have in common: none of them end with “and then export the data to a spreadsheet.” SparkToro data is worth more in motion (attached to a lead, informing an agent, driving a workflow) than sitting in a CSV.<\/p>\n\n\n\n An API doesn’t make audience data infallible. This is the same modeled, affinity-based data that powers the web app: directional by design, and best used to decide where to look and what to test. The pipeline can run unattended; the decisions it feeds shouldn’t. If a single number is about to move a five-figure budget, treat it like any research source and do a sanity-check.<\/p>\n\n\n\n The API is credit-metered and priced independently from any SparkToro subscription. The principle: you pay for exactly the data you use, and nothing else.<\/strong> If all you want is the AI usage data for an audience, that’s what you pay for. <\/p>\n\n\n\n Credits are consumed per call, and different endpoints have different weights. Creating an audience report costs 10 credits. From there, every section is a la carte: quick reads like demographics, networks, and podcasts cost 1 credit each; standard sections like websites, social, and subreddits cost 2; and premium endpoints like TAM and keyword extraction cost 3 to 5. A typical full pull (report creation plus eight sections) runs about 23 credits. There’s a credit calculator in the docs<\/a>: check the sections you need, and it shows your cost per report before you write a line of code.<\/p>\n\n\n\n Every new API account starts with 200 free credits, enough to prototype a real integration end-to-end, no card required. When you’re ready to scale, credit packs start at $500 for 5,000 credits, and volume bundles bring the per-credit price down from there (the Agency Bundle is 50,000 credits for $3,000). Credits stay good for 12 months, and there’s no monthly bill and no auto-renewal.<\/p>\n\n\n\n If you’re building something big, reach out to support@sparktoro.com<\/a>. We’ll work with you on pricing and credit allocations.<\/p>\n\n\n\n We can’t wait to see what you build. Seriously: what’s the first workflow you’d wire this into? Leave a comment or send it our way. The best ideas will end up as worked examples in the docs, with credit to you (and a link, and a mention AI tools will like, and all the glory our social channels and newsletter can deliver).<\/p>\n","protected":false},"excerpt":{"rendered":" I have a confession. For years, my least favorite customer question was “Does SparkToro have a public API?” Not because it was a bad question. But, because the answer was always “not yet, we still have other, higher priorities,\u201d which is\u2026 a pretty deflating answer to have to give.” SparkToro was built as a web<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[74,11,14],"tags":[],"class_list":["post-11443","post","type-post","status-publish","format-standard","hentry","category-ai-llms","category-product","category-sparktoro"],"yoast_head":"\nThe API is its own product, not a subscription add-on<\/h2>\n\n\n\n
\n
What you can build with it<\/h2>\n\n\n\n
1. Land the account: automated new-business pitches for agencies<\/h3>\n\n\n\n
![\"\"](\"https:\/\/images.sparktoro.com\/blog\/wp-content\/uploads\/2026\/07\/api-curl-call-1024x504.png\")
<\/figure>\n\n\n\n2. Media planning + influencer shortlists for in-house marketers<\/h3>\n\n\n\n
3. An AI agent that researches every new signup<\/h3>\n\n\n\n
A quick skeptic’s note<\/h2>\n\n\n\n
How pricing works<\/h2>\n\n\n\n
![\"\"](\"https:\/\/images.sparktoro.com\/blog\/wp-content\/uploads\/2026\/07\/api-calculator-1024x505.png\")
<\/figure>\n\n\n\n![\"\"](\"https:\/\/images.sparktoro.com\/blog\/wp-content\/uploads\/2026\/07\/api-dashboard-1024x640.png\")
<\/figure>\n\n\n\nGet started<\/h2>\n\n\n\n
\n