Happy to share that Stimuler has now crossed $5Mn ( as well as ₹50 crore) ARR in Aug. We crossed $1Mn in June last year. In this period, we doubled our performance ROAS, improved overall CAC and D30 retention doubled too, in all target geos. Here is everything that has worked for us across product and growth in the last 14 months. (While these are specific to Stimuler, i believe most of these would be relevant to most consumer products, esp those with global ambitions) Growth > For 0 to $1Mn ARR, performance marketing for us was highly unprofitable and in a way, subsidized by our organic growth, but if we had to continue to grow, performance needed to become self sustainable. That became one of our priority targets. > W0 ROAS in all performance channels became our north star rather than chasing multiple metrics. We added a new channel (tiktok ads) and invested significantly in building systems around creatives in all our target geos (having a network of creators, diversity in scripts, twice a week iteration frequency & more). Allocation between creatives as well as channels became much more methodical based on north star optimization > As a result, W0 ROAS for us moved from 44% in Oct’25 to 89% in Aug’26 which translates to near 100% Gross recovery by M0 itself, an early target we had set for ourselves. Still a lot of room here. > For organic, the journey was again to translate early wins into a much more scalable and replicable (across geos) playbook, with specific systems around hiring, training creators, script writing, format iterations, and more. We started with a good presence in LATAM, but we were able to replicate into Indonesia last year, making it our single biggest geography. Earlier this year, we were able to replicate this into India, and we always keep experimenting to win new geographies as well. > Being able to replicate the organic playbook in multiple geographies has been the major reason why our blended CAC has stayed consistent (& in fact dropped, to be specific) despite scaling 5x in the last 14 months. As a result, there are now 4 countries where we have crossed $500k ARR, and all of them have grown at least 3x year on year, which are very rare numbers for a company at our stage. Now that we have a huge number of users coming in (>1Mn new app installs every month), making them stay while having an amazing experience becomes the obvious focus, bringing me to product :) Product > After we crossed $1Mn ARR last year, the overall team decided to make retention our P0 across the product because our numbers were very far from how the best companies in the world were doing. > The approach here was very simple. We divided retention into three separate problem statements: Early stages: D1 to D30 Mid-stage: M1 to M6 Late stages: M6 to Y1, Y2, and beyond > There was another choice to make in terms of geography because solving retention across separate geographies had certain differences. We started with Indonesia first here as it was the biggest geo by user-size and hence, allowed faster iterations on experiments in terms of statistical significance. This involved a lot of changes in showing progress early to users in their journey, improving the core product engines via breakthroughs from our AI team (more on this in another post), localization and more. > Once we started seeing the results in Indonesia, we then started replicating the same efforts in LATAM and, to a certain extent, in India. The results show this story well In last 12 months, Indonesia : D7 went from 23.5% to 31.6% , D30 went from 5.9% to 12.6% Latam: D7 went from 21.1% to 31.1% , D30 went from 5.1% to 14.2% India (our most recent market): D7 went from 15.9% to 22.1% , D30 went from 3.8% to 8.7% > Parallely, we started now focusing on midterm retention in Indonesia, which involved going much deeper into user-level personalization, reducing friction across user interactions with the app, and novelty at an experience level. Of course, the core product thesis, which is the ability to show users how much they have moved in their fluency journey forward, remains the core loop to bring people back. As a result, Indonesia M3 (usage) went from 20.6% to 40%+ M6 also has now touched a peak of 28-29% > The road from here is to translate these midterm gains into longer-term results in year 1 retention (targeting 30% Y1) and beyond, while also solving the midterm retention in Latin America and then India. ***** At @StimulerApp, our thesis is that English Fluency is an incredible income mobility lever in emerging economies across the world (India, SEA, Latam, Mena etc). And that we can build a voice AI product for these 1.5Bn English learners globally, right here from India- a product that can take them to a stage where they can actually communicate confidently, not just learn basic words. We're at 1% today relative to that number We started the company right out of our college, went full time in 2023 after graduation and its been an amazing 3 years of super-accelarated learning trajectory, building a consumer AI product with everything happening around us. Been highly privileged to work with an insane talent-dense team which deserves credit for everything we wrote above. Ofc we are very very early in the path to $100Mn ARR and then much beyond that in the next 3-4 years. See you all next in the $10Mn ARR post!

❗️Wordpress is force-updating 100,000+ websites over a critical flaw in Pods, a WordPress plugin, that lets unauthenticated attackers overwrite any account's password, including the site owner's, for full takeover. Due to the critical severity of this vulnerability, the plugin vendor is working with the WordPress plugins team to push a forced update to the patched version (CVE-2026-19598, CVSS 9.8).

the future should look like the future

Aaaaaaand we're live! outbid.lol No ads. No API keys. No revenue sharing. Just outbid your competitors to rank #1 and consider marketing done for today 🚀

how is ai still so bad at design

congrats. we use cal.com heavily at @zappush. it's helped us close all our sales along with @_skris's superblog

A customer's superblog was not scoring great on Core Web Vitals/Lighthouse Audit. But superblog was built with the philosophy to hit 90+ score on average. So I started investigating their superblog. They uploaded flat Figma-export PNGs that were small on disk (~3MB) but enormous in dimensions (up to 28,527×9,162 = 261 MP). Our upload guard only ever checked bytes, never pixels, so they passed. So every visitor's browser was allocating up to 1GB to decode a "512KB" image. Just issued a fix and backfill to this customer's images automatically. Now they're flying with 90+. With zero effort from their side. These are the kind of edge-cases superblog platform is built to handle. You focus on writing content, superblog handles the rest.