Teardown · December 2023 · Sales automation
Built as a case study exercise using public information. Not a client engagement.
PhantomBuster: should a LinkedIn tool chase customers who don't sell on LinkedIn?
The situation
PhantomBuster grew product-led, no sales team, selling LinkedIn automation to tech, advertising and consulting firms. The expansion question was whether it could serve businesses selling to brick-and-mortar buyers instead.
The appeal is obvious in a spreadsheet. The existing market is 3 million US small businesses actively using LinkedIn for sales. The proposed one is 6 million engaged in brick-and-mortar B2B sales. Twice the size.
The problem is that the entire product is built on one platform those buyers don't use.
6M
Proposed market
3M
Current market
1
Platform dependency
The analysis in three numbers.
What I built
Existing and proposed ICP sets side by side, with the differences made explicit rather than implied: digital versus physical market focus, B2B services versus B2B and B2C dynamics, tech-savvy versus mixed technical proficiency, professional networking versus direct sales and distribution.
Three new persona groups covering local service providers, manufacturers and wholesalers, and supply and distribution companies. A competitive landscape for the new segment. The product implications: aggregating data from Google Maps, Yellowpages, Facebook and industry directories rather than LinkedIn alone. And a GTM roadmap with next steps split across product, customer success, content and sales.
The judgment call
Answering the TAM argument with a product argument.
A market twice the size is the kind of number that ends discussions. But the honest reading is that PhantomBuster's advantage is LinkedIn-shaped (the automation, the integrations, the onboarding assumptions), and the new segment's buyers are discoverable on entirely different surfaces.
So the recommendation wasn't yes or no. It was: this expansion is a product investment wearing marketing clothes, and if the data-source work doesn't happen first, the segment can't be served regardless of how good the messaging is.
The most useful thing a market analysis can do is tell you which function actually owns the decision.
What I'd do differently
I'd have pressure-tested the 20% assumption behind the 6 million figure. It's a reasonable estimate, and it's doing an enormous amount of work in the argument. Any number carrying that much weight deserves a sensitivity range rather than a point estimate.