Daash estimates competitor product sales and tracks ingredient trends for beauty brands – turning opaque market data into an actionable intelligence layer.
ENTRY ANGLES
Build vertical-specific competitor intelligence platforms using consumer survey data + web-scraped commercial signals + AI synthesis · Create domain-specific estimation algorithms that calibrate signal detection to each industry's unique patterns · Develop vertical-specific data layers (e.g., product composition databases, marketing copy parsing) that increase product defensibility
VERTICALS
CAPABILITIES
Domain-specific signal calibration and vertical-tailored estimation algorithms, AI synthesis layer to combine consumer surveys + web scraping into actionable insights, Vertical-specific data infrastructure (databases, parsing systems)
Daash built a sales intelligence platform for beauty, personal care, and fragrance brands – one that estimates competitor sales volumes and surfaces the trends driving them.
The platform answers two core questions: which products generate the highest sales in a competitor's line, and which products are growing fastest. But the more interesting layer is ingredient and attribute intelligence. Daash tracks product formulations and the benefit claims brands emphasize – "hydration," "firming," "anti-aging" – and correlates them with sales performance. The result is a view not just of which specific products are winning, but which ingredients and positioning angles are driving consumer preference across the category.
All of this data can be tracked over time, broken out by product category, and split between retail channel and direct-to-consumer sales – giving brands a dynamic view of their competitive position.
Of course, precise competitor sales data isn't available through any legitimate public source. Daash uses estimation methods instead: on one side, regular consumer panels tracking which beauty and personal care products different audience segments are buying; on the other, web-scraped data on search volume, site traffic, brand mentions, and any commercial data findable online. An AI engine synthesizes these signals to estimate sales figures at the SKU level.
The critical refinement: Daash periodically enters data-sharing agreements with individual brands, obtaining their actual sales figures and using them to calibrate the AI model through a feedback loop. That keeps the estimates grounded in real-world data rather than pure inference.
Interestingly, Daash sees financial services – banks and investment funds that want sales data to assess potential borrowers and acquisition targets – as a secondary target audience alongside beauty brands themselves.
The platform is currently in beta but already has paying beauty brand clients. Daash just raised $5.5M in new funding, adding to $2.8M raised in 2023.
The estimation methodology might sound like it couldn't possibly produce useful results.
But SimilarWeb – a widely used service that estimates website traffic using a conceptually similar approach – raised $260M before going public, where it now trades at a $1.38B market cap. That's a reasonable proof point that AI-driven estimation, done carefully, can be accurate enough to support real business decisions.
More interesting than the technical execution, though, is Daash's use of the "1 + 1 = 3" formula.
On one side: consumer survey data. Many companies run these – including startups doing it with modern technology. Cafeteria ([covered previously](/review/podsadi-klienta-na-podpisku)), for example, sells youth brands a subscription to regular audience surveys and returns trend data as charts and tables. That earned Cafeteria $3M in its first round.
Knit ([related review](/review/budut-li-oni-jeto-pokupat)) started as a video survey platform and evolved into a comprehensive qualitative and quantitative research tool powered by AI – raising $14.6M in the process.
On the other: data enrichment services that crawl the web, extract information about specific companies or topics, and sell access to the resulting databases. Openmart ([covered here](/review/jeto-uzhe-dengi-no-mozhno-zarabotat-eshhjo-bolshe)), a recent Y Combinator graduate, does this for local businesses – helping regional and national distributors find potential customers; raised $2.75M. Resquared follows the same model with a focus on peer-to-peer local business outreach, and raised $5M.
Daash's insight is that combining these two inputs – consumer panel data and web-scraped competitive signals – through an AI synthesis layer produces something worth far more than either input alone: estimated competitor sales at the SKU level. That's "1 + 1 = 3": the output commands a meaningfully higher price than the sum of its component parts, because it answers a question neither input could answer individually.
Nichefire ([covered here](/review/kak-uspet-vojti-v-novyj-trend-chtoby-pobolshe-zarabotat)) pulled a similar move, raising $2.6M for a platform that predicts emerging cultural trends in food and dining. Under the hood it's semantic keyword analysis – a technique with deep roots in search engine optimization – but wrapped in an AI layer and repositioned for restaurant operators and food brands. Same underlying technology, different application, different buyer, higher price point.
The direct opportunity is building Daash equivalents for other industries.
The challenge is that the estimation algorithms need to be tailored to each vertical. Generic approaches won't hold up: each industry has its own signal landscape, and accuracy requires domain-specific calibration. Daash, for instance, also maintains a database of product ingredient compositions and parses marketing copy to extract benefit claims – that's vertical-specific work, but it's exactly what makes the product more valuable.
Asking "which industries want to track competitors?" is a meaningless question – the answer is all of them. The more useful question: in which other industries can you combine consumer survey data and web-scraped commercial signals through AI, and get estimates reliable enough that businesses will actually make decisions based on them?
The broader framing: which combinations of existing, well-understood technologies can be connected through an AI synthesis layer to produce a new product that answers a question neither could answer alone – and that can therefore be sold at a meaningfully higher price than the sum of its parts?