❯

    cogny blog

    Insights & updates.

    AI-powered analytics, growth strategies, and the future of data-driven decision making.

    Blog

    Bing Ads Is Live in Cogny: Audit and Run Microsoft Advertising From an Agent

    Microsoft Advertising now has an MCP in Cogny. Sign in with Microsoft, pick the ad accounts, and your agent can list campaigns, ad groups, ads and keywords, run campaign, keyword, search-term, ad and geographic reports with impression share and Quality Score, and push negatives, bids, budgets and status changes — 17 tools over the Bing Ads API v13, callable from chat, the Cogny CLI, or any MCP client. No SOAP client and no developer-token application on your side.

    Sep 9, 20266 min read
    Blog

    Wish for an MCP: Name the Tool, Cogny's Agent Starts Building the Integration

    There is a new box at the bottom of Settings → MCPs For Cogny. Paste a link to a tool's API docs, or just type its name. Cogny reads the docs, checks the catalog, and if the tool has a public API a customer can authorize on their own, a coding agent starts building the MCP integration right away — with a pull request, CI and a human review before it ships. You get the verdict in the same box, and the team follows up when it's live.

    Sep 9, 20263 min read
    Blog

    Abicart Is Live in Cogny: Orders, Revenue and Catalog for Your Agent

    Abicart joins Shopify, WooCommerce and Wix as the fourth e-commerce platform in Cogny — and the first built for the Swedish webshop market. Paste one admin API token and Claude can sum revenue per week in your base currency, rank best sellers, read orders with line items, audit the catalog, and pull customers and discounts — through app.cogny.com/mcp and the Cogny CLI.

    Sep 8, 20264 min read
    Blog

    Litium Is Live in Cogny: Orders, Revenue, Stock and Customers for Your Agent

    Litium Commerce Cloud joins Shopify, WooCommerce, Wix and Abicart as the fifth e-commerce platform in Cogny — and the first built for the Nordic B2B/B2C mid-market. Connect a back-office service account and Claude can sum revenue per week in your base currency, split it by channel and country, rank best sellers, read orders with line items, check stock and price lists, and pull customers — through app.cogny.com/mcp and the Cogny CLI.

    Sep 8, 20266 min read
    Blog

    Epiminds Lucy: What's Publicly Known, Pricing, and Alternatives (2026)

    Epiminds' Lucy is an AI marketing manager coordinating 20+ agents, backed by a $6.6M Lightspeed seed. There are no user reviews on G2 or Capterra yet and pricing is unpublished. Here is what the public record says, what to ask on a demo, and the alternatives worth shortlisting — including where Cogny fits.

    Sep 7, 20267 min read
    Blog

    Meta Says 40 Leads, HubSpot Says 65: How to Calculate Real Cost Per Lead

    When visitors decline tracking consent, Meta's pixel never sees the form they fill in — but your CRM does. Meta's reported cost per lead is then too high, sometimes by half. Here is the method for reconciling Meta spend with HubSpot leads on campaign and UTM, and how to run it as one query in Cogny.

    Sep 7, 20266 min read
    Blog

    ChatGPT Ads: What They Are, What They Cost, and How to Run Them From an Agent

    Everyone searches for "ChatGPT ads" — OpenAI calls the product OpenAI Ads. Same auction: sponsored chat cards under a model response, targeted by context hints instead of keywords. Here is how the channel is actually built (ad groups, hints, card character bands, CPM-derived bids), what it costs, and how to run and audit the whole thing from an agent through Cogny's OpenAI Ads MCP — plus a hands-on build walkthrough from our partner agency.

    Sep 5, 20265 min read
    Blog

    B2B Marketing Attribution Models Explained: First-Touch, Multi-Touch, CRM-Linked

    First-touch, last-touch, linear, W-shaped, data-driven — attribution models have multiplied while the underlying question stayed the same: which marketing effort actually drove this deal? Here's what each model tells you, where each lies to you, and which one fits B2B sales cycles.

    Sep 3, 202611 min read
    Blog

    Marketing Attribution for B2B: Which Campaigns Actually Drive Sales Calls

    Most B2B marketing teams can tell you their ad spend. Far fewer can tell you which campaigns actually booked the calls that became pipeline. This guide closes that gap — covering attribution models, UTM hygiene, CRM linkage, and how to know which LinkedIn campaigns warm up prospects before they pick up the phone.

    Sep 3, 202613 min read
    Blog

    Connecting Ad Spend to Closed Deals: A Practical Attribution Playbook

    Your ads generate leads. Your sales team closes deals. Somewhere between those two facts, the connection breaks down. This playbook walks through the exact steps to link every closed deal back to the campaigns that drove it — from UTM setup to CRM reporting to board-ready ROI numbers.

    Sep 3, 202611 min read
    Blog

    Does LinkedIn Ad Warm-Up Increase Call Booking Rates? Data & Framework

    Running LinkedIn ads straight to a 'Book a Demo' CTA rarely works in B2B. Here's why warm-up sequences — content before conversion — increase call booking rates by 20-45%, and how to measure it for your own campaigns.

    Sep 3, 20268 min read
    Blog

    OpenAI Ads Is Live in Cogny: Run and Audit Your ChatGPT Ads From an Agent

    Ads inside ChatGPT now have an MCP in Cogny. Paste an Ads API key from ads.openai.com and your agent can list campaigns, ad groups and chat cards, pull insights by day, country and device, upload creative images, create and pause objects, resolve geo targets, and check that the conversion pixel is firing — 32 tools over the OpenAI Ads Advertiser API, callable from chat, the Cogny CLI, or any MCP client.

    Sep 3, 20267 min read
    Blog

    How to Track Which Marketing Channel Books Your Sales Calls

    If you can't answer 'which channel booked this call?', you're flying blind on budget allocation. Here's the exact UTM and CRM setup to close the tracking gap between your ad campaigns and your calendar.

    Sep 3, 20268 min read
    Blog

    Agents Can Now Push Back: Ticket Feedback Over MCP

    Cogny generates growth tickets from your marketing data. Until now, an agent that found one wrong — already shipped, bad numbers, duplicate — had nowhere to say so, and the same ticket came back next week. The new add_ticket_feedback MCP tool writes an agent's verdict into the field the ticket generator actually reads, so a rejected idea stays rejected.

    Aug 31, 20265 min read
    Blog

    New Plans: Cogny AI ($18) and Cogny Pro ($90) — Plus a 5x Bigger Free Gift

    Two new rungs land between Solo and Cloud: Cogny AI ($18/mo with 150 monthly credits) and Cogny Pro ($90/mo with 1,500). Both keep Solo's unmetered MCP for your own agent and add a monthly credit pool for the in-app agent. The free signup gift grows from 10 to 50 chat credits — sized so one full loop, question to shipped result, completes before the wall. Cloud grows from 2,000 to 5,000 included credits. Here's the new ladder and the data that shaped it.

    Aug 31, 20263 min read
    Blog

    Two New Solo Skills: SEO & GEO Content Creator + Structured Data & Schema Audit

    Solo just got two new Search Console-powered skills. The SEO & GEO Content Creator mines your queries — plus Google Ads search terms when connected — for non-brand intent you should own organically and in AI answers, then drafts the content straight into your CMS. The Structured Data & Schema Audit fetches your highest-traffic pages, validates their JSON-LD against Google's requirements, and adds the missing markup for you. Both end with a draft to review, not a to-do list.

    Aug 27, 20264 min read
    Blog

    Solo Guides: a library of what an AI growth agent can ship for you

    A new open section: step-by-step example workflows Cogny Solo runs for solo businesses — the analysis, the exact prompts to use, and the change that ships.

    Aug 27, 20262 min read
    Blog

    The Solo Tasks Board: Cogny Now Measures Every Change You Ship

    Ship a campaign, a blog post, or a tracking fix from the Cogny chat and it now lands on a Tasks board — where Cogny automatically comes back days later, measures the before/after impact with your connected data, and hands you a shareable result card. The analysis→implementation loop finally closes with the part everyone skips: checking whether it worked.

    Aug 27, 20263 min read
    Blog

    How to Prepare Your Website for AI Search Engines: An 8-Step GEO Checklist for Solopreneurs

    ChatGPT, Perplexity, and Google AI Overviews are now the first page of the internet for millions of queries. Here is a concrete 8-step checklist a solopreneur can run in an afternoon to get their site cited in AI answers — no agency, no expensive tools required.

    Aug 24, 202611 min read
    Blog

    What AI Assistants Actually Cite When Someone Asks 'Is Cogny Any Good?'

    ChatGPT and Claude already cite Cogny — but not the pages you'd expect. The Aug 2026 AI Search Audit found a clear pattern: original technical documentation gets cited, thin marketing pages get ignored. Here's what that means for buyers doing AI-assisted research.

    Aug 21, 20265 min read
    Blog

    How to Get Your Shopify Store Cited by ChatGPT and Perplexity

    A Scandinavian activewear store started receiving orders from Perplexity six months after launch — without doing anything AI-specific. The content that ranks on Google is the content AI engines cite. Here's the GEO playbook for e-commerce: what to publish, how to structure it, and how to measure AI citations with real data.

    Aug 20, 20264 min read
    Blog

    How to Grow a Shopify Store with AI: The Organic Search Playbook

    A six-month-old Scandinavian activewear store took organic search from zero to half of all orders — 25× impressions, 13–17× monthly orders, zero ad spend. This is the exact four-phase playbook an AI growth agent ran to get there, and how to run it on your own Shopify store.

    Aug 20, 20264 min read
    Blog

    5 Things Cogny Fixed in My Google Ads Account That No Dashboard Would Have Caught

    Google Ads dashboards show you what happened. They don't tell you why, or what to do about it. Here are five real findings from Cogny's AI agent — the kind of issues that hide in plain sight until an agent goes looking.

    Aug 17, 202611 min read
    Blog

    AI Marketing Tool vs AI Growth Agent: Why the Difference Matters

    Every marketing platform calls itself an AI tool now. But there's a fundamental difference between a tool that shows you data and an agent that decides what to do with it. Here's why the distinction matters for your team.

    Aug 17, 20269 min read
    Blog

    How to Grow Your Vibe-Coded App: The Exact Analytics Setup That Works

    You shipped with Cursor, Lovable, or Claude Code. Now what? Here's the analytics setup that tells you what's actually working — and where users are dropping off.

    Aug 17, 20269 min read
    Blog

    Kimi K3 and Opus 5 on Cogny Bench: A Bigger Open Model Isn't a Better Buy

    Berget added Kimi K3, so we ran it through Cogny Bench. It's frontier on eight of nine problems but lands at 90.2 — statistically tied with GLM-5.2's 90.0 while costing 2.7x more per run. In the same sweep, Claude Opus 5 took the top of the board at 94.0. Here's the ranking, both Berget models side by side, and an honest note on a grader fix Opus 5 surfaced.

    Jul 30, 20265 min read
    Blog

    Framer Is Live in Cogny: CMS, Redirects, Analytics & Publishing for Your Agent

    Framer joins Webflow, WordPress and Ghost as the fourth CMS in Cogny — and it's the first integration built on Framer's official Server API. Connect a project API key + project URL and Claude can read and write CMS collections, manage redirects, query site analytics, and publish — with drafts excluded from publishing by default.

    Jul 20, 20264 min read
    Blog

    Same Task, Same Data, Half the Score: Measuring Context Robustness on Cogny Bench

    We wrapped a Cogny Bench task in a production-shaped envelope — a thousand words of workspace context plus an organizational memory that confidently asserts the wrong explanation. Claude Sonnet 5 held 99 → 99. GPT-5.6 Luna fell 99 → 47, GPT-5.6 Terra 98 → 43, and Claude Haiku 4.5 96 → 49 — after computing the disproving evidence themselves. And the failing runs were cheaper than the passing ones.

    Jul 16, 20266 min read
    Blog

    Webflow MCP: SEO Metadata, Page Copy, and Redirects — Now Agent-Operable

    The Cogny Webflow MCP grew from CMS reads and item writes to the full structural-fix loop we promise on the integration page: update a page's SEO title and description, rewrite static page copy, manage 301 redirects, and evolve collection schemas — 28 tools total, every write gated behind ask-first approval. Here's what shipped and how an agent runs the loop.

    Jul 15, 20263 min read
    Blog

    Wix Is Live in Cogny: Site Analytics, Orders & Contacts for Your Agent

    Wix joins Shopify and WooCommerce as the third e-commerce platform in Cogny. Connect one API key + your Site ID and Claude can read daily site analytics, search store orders, query products (Catalog V1 and V3), contacts and coupons — read-only by design, through the same app.cogny.com/mcp endpoint and the Cogny CLI.

    Jul 15, 20263 min read
    Blog

    GPT-5.6 on Cogny Bench: Terra Matches Opus 4.8 at $0.05 a Run

    OpenAI shipped the GPT-5.6 family — Sol, Terra, Luna — this week. We ran all three through Cogny Bench: Terra ties Opus 4.8 at 93.1 for a thirteenth of the cost, the flagship Sol adds nothing over Terra on our problems, and $1/$6 Luna lands one point behind. Plus the API change that broke our first run.

    Jul 10, 20264 min read
    Blog

    Skills That Implement: the Cogny Solo Audit, Implement, Routine Loop

    Most marketing audits end as a PDF nobody actions. Cogny Solo's skills now end inside your accounts: the SEO audit stages the title fixes in your CMS, the Meta audit pauses the wasted spend, the GA4 monitor repairs the broken tag in GTM. 19 skills across 6 channels, every write approval-gated, and a hub that nudges you back when your weekly routine is due.

    Jul 5, 20265 min read
    Blog

    Steer Your TV Campaign Like a Trader: the CPM Simulator and Multi-Strategy Bidding

    Cogny TV campaigns now run several bid strategies at once — a general tier plus premium audience tiers at higher CPMs — and the campaign card includes a delivery simulator calibrated on live TV4 flights. Move a strategy's CPM slider, watch the predicted impressions curve respond, and Apply sends exactly what you simulated to the ad platform.

    Jul 5, 20264 min read
    Blog

    Claude Fable 5 on Cogny Bench: When the Flagship Isn't the Value Pick

    We ran Anthropic's most capable model, Claude Fable 5, through Cogny Bench. On nearly every problem it's frontier-perfect — but one reproducible stumble on a multi-step action task drags it to 86, below Sonnet 5's 93.6 at four times the cost. A useful reminder that bigger isn't automatically better for marketing-analytics work.

    Jul 2, 20263 min read
    Blog

    Claude Sonnet 5 on Cogny Bench: Frontier Accuracy at a Sonnet Price

    Anthropic shipped Claude Sonnet 5 today. We ran it through Cogny Bench the same afternoon: it tops our entire model stack at 93.6 — beating Opus 4.8 — while costing a fifth as much. Here's the run, both price points, and the one trap it didn't fully crack.

    Jun 30, 20264 min read
    Blog

    The Capability Arc: Running Cogny Bench Across Six OpenAI Models, 2023 → 2025

    Cogny has been building agentic SQL analysis since December 2022 — for years, the models simply couldn't do it. We put that claim to the test by running our internal benchmark across a span of OpenAI models from GPT-3.5 Turbo to GPT-4.1. The failures are half the story: one model is retired and 404s on every call, the 2023–2024 chat models can't be trusted with a trap, and the score only clears 70 when the reasoning generation arrives. Here's the arc, in real numbers.

    Jun 29, 20268 min read
    Blog

    Cogny Bench: Why We Built Our Own Frontier-Model Eval (And Why It Paid Off This Week)

    Generic benchmarks tell you nothing about whether a model will catch a Simpson's paradox in your paid-social funnel or notice that a 'best-ROAS' channel is actually an attribution leak. So we built Cogny Bench. This week — when a new Swedish-sovereign frontier model dropped — we had a ranked recommendation in ten minutes for a few dollars.

    Jun 24, 202610 min read
    Blog

    Meta Ads Is Live in Cogny: Facebook & Instagram, Finally

    The most-requested channel on our roadmap just shipped. Meta Ads — Facebook and Instagram — is now a live, OAuth-connected MCP integration in Cogny, joining LinkedIn, Reddit, TikTok and X as the fifth paid-social channel. Connect it in two clicks and Claude can audit campaigns, read ad-set and ad-level performance, inspect audiences and pixels, and draft changes — all through the same app.cogny.com/mcp endpoint.

    Jun 24, 20263 min read
    Blog

    Review Before Live: Publish-Gated Tickets for Lovable and WordPress

    When a ticket agent built a page in Lovable, it deployed straight to the production site — no preview, no human in the loop. For a lot of work, shipping unreviewed is a non-starter. So we gave Lovable and WordPress tickets the same review gate our coding-agent tickets already have: build on a preview, look at it, then click Make it live.

    Jun 17, 20263 min read
    Blog

    Cogny Solo Now Has a Built-In Chat — Marketing Skills, No Setup

    Solo used to mean one thing: connect Cogny to your own coding agent over MCP. Now there's a second way in. Open Cogny Solo and use the built-in chat — preloaded with marketing skills for SEO audits, competitor research and ad copy, with nothing to install. Technical users still get unmetered MCP at $9/mo; everyone else can just sign up and start asking.

    Jun 16, 20263 min read
    News

    Claude Fable 5 Is Live in Cogny — And Why a Multi-Tier Model Strategy Beats Betting on One Model

    Anthropic's newest frontier model, Claude Fable 5, now powers the deepest analysis work in Cogny — live today for all Cloud customers. Here's why we route different jobs to different models, and why that's the architecture every serious AI product will converge on.

    Jun 10, 20265 min read
    Blog

    MCP Observability: The Write Ledger That Turns Agent Actions Into an Audit Trail

    When an AI agent creates a campaign, changes a budget, or sends an email through MCP, who recorded that it happened? Cogny's new MCP Write Ledger streams every mutation — successes and failures — into your own BigQuery as an immutable audit row. Here's why that's the difference between 'MCPs in the wild' and MCP you can put in front of an enterprise.

    Jun 10, 20269 min read
    Blog

    The Cogny CLI: Your Marketing MCP, Scriptable From the Command Line

    The Cogny CLI exposes every Cogny MCP tool as a plain command-line subcommand — Google Ads, GA4, Search Console, BigQuery, Meta, and more — callable from Bash with no MCP wiring, no OAuth dance, and no restart. It's built for agents: one command signs the user up, the next runs a real ad audit. Here's why a CLI beats a native MCP connector for scripting, runbooks, and agentic onboarding.

    Jun 6, 202611 min read
    Blog

    How to Set Up MCP for Marketing Analytics in 15 Minutes

    Step-by-step guide to connecting your marketing stack via MCP — GA4, Google Ads, Stripe, and more. Get AI-powered growth insights in your first 24 hours.

    Jun 5, 20269 min read
    Blog

    The 10 Best MCP Marketing Tools in 2026 (Ranked by What They Actually Do)

    Most MCP tools just connect your data. We ranked the 10 best MCP marketing tools by what they actually do with it — including the only platform that generates verified growth actions.

    Jun 5, 202612 min read
    Blog

    MCP vs. Marketing Automation: What's Actually Different (2026)

    MCP marketing and marketing automation solve different problems. One executes campaigns you've built. The other figures out what campaigns to build — and whether they worked.

    Jun 5, 20269 min read
    Blog

    Search Console Finally Shows AI Overviews & AI Mode Data — Here's What's In It (and What's Not)

    Google just launched a Generative AI performance report in Search Console, surfacing impressions inside AI Overviews and AI Mode. It's a real step forward for GEO measurement — but it's impressions-only, UI-only, and not yet in the API. Here's the honest breakdown, and exactly when it lands in our MCP.

    Jun 3, 20266 min read
    Blog

    Introducing the Cogny Free Tier: 15 Calls on Signup, Then Solo or Cloud

    Every new Cogny workspace now starts with 15 free AI calls — no credit card, no sales call, no human in the loop. When you outgrow the gift you pick Solo ($9/mo) — unmetered MCP tools for your own agent, with Cogny's in-app chat on pay-as-you-go top-up credits — or Cloud ($499/mo) with 2000 included credits that cover everything. Here's how the freemium ladder works, why we meter our own inference instead of seats, and how an agent can sign itself up in one command.

    Jun 2, 20266 min read
    Blog

    Self-Driving Pull Requests: How Cogny Sites, Cogny Code, and Visual Proof Removed Us From the Merge Loop

    PRs were sitting for weeks. We rebuilt our own dev loop on four Cogny features — automatic preview environments, agent-driven build autofix, AI-scored deployment risk, and a computer-use visual reviewer — and shipped a self-driving PR-to-prod pipeline that caught six real bugs the day we turned it on.

    May 28, 202612 min read
    Blog

    AI Marketing Harness: Why Free Skills and Raw Claude Aren't Enough

    Anyone can open Claude and ask it to optimize their marketing. The reason most teams quit after two weeks isn't the model — it's the missing harness. Here's what an AI marketing harness actually is, why a chat window and a skills folder won't get you to production, and how Cogny's 25 MCPs, 39 templates, and falsifiable tickets close the loop.

    May 19, 202621 min read
    Blog

    AI Marketing Is the New SEO: One Strategy for Google and the AI Engines

    SEO got you ranked. GEO gets you cited. In 2026, the smart play is to stop treating them as separate disciplines and run them as one strategy — powered by AI agents reading your real first-party data. Here's how the unified loop works, and why Cogny's harness is built for both.

    May 19, 202618 min read
    Blog

    Claude for Marketing: A Practical Guide to Anthropic's Claude for Marketers (2026)

    Claude is the model most serious marketing teams are quietly betting on in 2026. Here's what Claude is good at, what it isn't, and how to actually use it for paid media, SEO, GEO, email, and content — without the chat-window-and-pasted-CSV theatre.

    May 19, 202618 min read
    Blog

    Falsifiable Marketing Experiments: Why AI Tickets Beat AI Suggestions

    Most AI marketing tools generate suggestions. Suggestions aren't experiments — they have no hypothesis, no expected outcome, and no measurement. Here's why falsifiable experiments are the only way AI marketing actually compounds, and how Cogny's Growth Tickets enforce it at the schema level.

    May 19, 202614 min read
    Product

    The Board-Meeting Use Case We Never Pitched — And the Support Ping That Revealed It

    Our chat agent crashed mid-session. Seconds later, a CEO pinged me directly: 'You're disrupting my board prep.' He was running user cohort analysis across CRM, marketing, email, search, and GA in Cogny — a use case we'd never pitched, never documented, never designed for. Here's what that one support ping taught us about positioning.

    Apr 24, 20267 min read
    Blog

    What Is an AI Growth Hacker? The 2026 Guide

    Growth hacking was invented by humans with spreadsheets. The 2026 version runs on AI agents that analyse 100% of your campaigns 24/7. Here's what an AI growth hacker actually is — and why the title is no longer a metaphor.

    Apr 24, 20269 min read
    Blog

    What Is an AI Marketing Agent? (And Why It's Better Than a Dashboard)

    An AI marketing agent doesn't just show you the data — it tells you what to do. Here's what an AI marketing agent actually is, how it differs from a BI dashboard, and what to look for in 2026.

    Apr 24, 20269 min read
    Blog

    Best AI SEO Tools in 2026: From Dashboards to Conversational AI

    Most AI SEO tools bolt AI onto old dashboards or generate content from generic training data. The third approach — connecting AI directly to your Search Console via MCP — changes everything.

    Apr 23, 20268 min read
    Blog

    Email Marketing MCP: What Happens When AI Can Read Your Email Performance Data

    Email marketers spend hours exporting CSVs and comparing open rates manually. MCP changes that by letting AI query your email data directly. Here's what becomes possible when Claude can read your HubSpot campaigns in real time.

    Apr 23, 20268 min read
    Blog

    GEO Meets MCP: How to Monitor Your AI Search Visibility With Real Data

    GEO monitoring without real data is guesswork. By connecting Search Console and Bing Webmaster Tools to Claude via MCP, you can track AI citations, detect zero-click erosion, and optimize for generative engines using your own first-party data.

    Apr 23, 20269 min read
    Blog

    What Is Marketing MCP? How Model Context Protocol Is Changing AI Marketing

    MCP is the open protocol that lets AI read your real marketing data — Search Console, LinkedIn Ads, GA4, and more. No exports, no dashboards, no middlemen. Here's what it means for marketers and why it matters.

    Apr 23, 20268 min read
    Blog

    What Is Vibe Coding? And Why the Tools You Give Your AI Matter More Than the Prompt

    Vibe coding means describing what you want in plain English and letting AI write the code. But the real unlock isn't better prompts — it's giving your AI the right tools. Here's what that means for developers, marketers, and anyone building with AI.

    Apr 10, 20267 min read
    Blog

    What Is Vibe Marketing? The Marketer's Guide to AI-First Growth

    Vibe marketing means describing the growth outcome you want and letting AI do the technical execution — queries, analysis, optimization. It's vibe coding for marketers. Here's what it means, how it works, and how to start.

    Apr 10, 202610 min read
    News

    Anthropic's Claude Mythos Leak: A Step Change in AI — And Why Your Growth Data Strategy Matters More Than Ever

    Anthropic just confirmed they're testing their most capable model ever. Here's what the Claude Mythos leak tells us about the future of AI-powered growth, and why teams collecting structured experiment data today will have a massive advantage when it ships.

    Mar 27, 20266 min read
    Blog

    Over 100 Million Search Impressions. Every Month. Across Our Customer Base.

    We recently ran the numbers on how much search data flows through Cogny's infrastructure. The result surprised even us — over 100 million Google Search Console impressions monitored monthly, across dozens of companies, from startups to household names.

    Mar 27, 20262 min read
    Blog

    How We're Approaching GEO Analysis at Cogny

    A look at how we're using Search Console data and AI-powered reporting to help marketers understand and prepare for the shift toward AI-powered search.

    Feb 21, 20268 min read
    Blog

    5 Signs Your Content Strategy Needs AI

    Declining organic traffic, rising CPCs, content that ranks but doesn't convert -- these are symptoms of a content strategy that's hit its ceiling. Here are 5 clear signals it's time to bring AI into your content operations.

    Feb 17, 20269 min read

    AI-agenten som faktiskt driver tillväxt – inte bara skapar rapporter

    De flesta AI-verktyg är bara dashboards med en chatbot klistrad på toppen. En riktig AI-agent gör jobbet – skapar kampanjer, optimerar budgets och tar action. Så här stänger du gapet mellan insikt och tillväxt.

    Feb 13, 20263 min read
    Blog

    The Real Cost of NOT Having a Unified Search Strategy

    Siloed SEO, SEM, and content teams are quietly burning budget. Duplicated spend, missed synergies, conflicting strategies -- here's what it actually costs and how unification saves money.

    Feb 12, 20269 min read
    Blog

    Search Console Metrics That Predict AI Citations

    Not all Search Console data is created equal. Certain metrics -- CTR patterns, position stability, query types -- strongly correlate with whether AI search engines cite your content. Here's what to watch.

    Feb 10, 202610 min read
    Blog

    Why AI Content Built on First-Party Data Wins

    Generic AI content is flooding the internet and failing to convert. The winners are using first-party data from GA4, Google Ads, and Search Console to create AI-powered content that actually drives results.

    Feb 7, 20268 min read
    Blog

    How Google Ads Data Can 10x Your SEO

    Your paid search account is sitting on a goldmine of SEO intelligence. Keyword performance, Quality Scores, search term reports -- here's how to use Google Ads data to supercharge your organic strategy.

    Feb 5, 20268 min read
    Blog

    What Is GEO and Why It Matters in 2026

    Generative Engine Optimization is the new frontier in search. As ChatGPT, Perplexity, and Gemini consume Google's traffic, brands that fail to optimize for AI-generated answers will become invisible.

    Feb 3, 20267 min read
    Blog

    Architecture of AI-Powered Marketing Automation

    How do you build a production AI system that can analyze marketing data, generate insights, and make recommendations in real-time? Here's the complete technical architecture behind Cogny, from data pipeline to AI agents to streaming responses.

    Jan 21, 202516 min read
    Blog

    Real-Time BigQuery Analysis: Lessons from Production

    Running AI agents that query BigQuery in real-time sounds simple in theory. In production with real users and real data, we learned a dozen hard lessons about performance, cost, reliability, and what 'real-time' actually means.

    Jan 19, 202516 min read
    Blog

    Why We Chose Claude Over GPT-4 for Marketing AI

    When building Cogny, we had to make a critical decision: which AI model should power our platform? After extensive testing with GPT-4, Claude, and others, we bet our company on Claude. Here's the technical reasoning behind that choice.

    Jan 17, 202515 min read
    Blog

    11 Years of Growth Hacking: Why AI Changes Everything

    I've spent over a decade in growth hacking—from manual spreadsheet campaigns to AI-powered automation. Looking back, I realize the game has fundamentally changed. Here's what that means for everyone in growth.

    Jan 15, 202514 min read
    Blog

    What We Learned Building AI Platforms for Netflix and Zalando

    Five years ago, we set out to build AI-powered marketing optimization for some of Europe's biggest brands. Here are the hard lessons we learned along the way—and why most of them were wrong for what comes next.

    Jan 13, 202514 min read
    Blog

    From BI Tools to AI Agents: The Next Evolution in MarTech

    After spending a decade building marketing technology, I've watched the industry evolve from basic dashboards to AI agents that don't just show you data—they act on it. Here's what that transformation really means.

    Jan 11, 20259 min read
    Blog

    How We Built the First AI Agent That Generates Growth Tickets

    Most AI marketing tools are chatbots with analytics access. Real AI agents are different—they monitor continuously, identify opportunities autonomously, and generate actionable tickets. Here's how we built one.

    Jan 8, 20259 min read
    Blog

    Why Every Marketing Team Needs an AI Growth Hacker in 2025

    After 20 years optimizing marketing campaigns for Netflix, Zalando, and hundreds of startups, I've seen every evolution in digital marketing. But AI agents are fundamentally different—here's why every team needs one.

    Jan 6, 20257 min read