A fractional AI Sales Department for marketplaces, tuned for the two-sided motion.

A marketplace at 30 employees has the acquisition surface of a 60-employee SaaS company and the headcount of a 30-employee SaaS company. The math does not work out. Supply acquisition is the high-touch side. You are reaching out to a vendor, a seller, a host, a driver, or a service provider whose decision to onboard onto your platform is a multi-month conversation about take rate exposure, payout cadence, KYC compliance, and the category density you already have in their region. That conversation is enterprise sales. Average cycle 60 to 120 days. Average research depth 30 minutes per prospect. Two supply-side reps can run 200 quality conversations a quarter at best, and the supply team you need to feed GMV growth is 800 conversations a quarter.

Demand acquisition is the opposite shape. You are running programmatic outreach to end buyers, end riders, end guests, or end users whose decision to transact is a 15-minute browse-to-checkout cycle. The labor problem on the demand side is volume and language coverage. Programmatic SEO across the categories your supply side covers. Partner acquisition where the buyer concentration lives. Email and push retention for the cohort that bought once and bounced. The metric that matters is cost per first transaction and ninety-day repeat rate. Two demand-side reps cannot run the volume the funnel needs, and even if they could, they are not the same reps as the supply team because the message, cadence, and CRM shape are completely different.

The default marketplace answer is to treat both sides as one growth team. That fails for a structural reason. The supply-side reps write generic emails because they are also running the demand-side cadence. The demand-side reps under-invest in performance because they are also doing supply research. The take rate growth that the board model promised in the Series A deck does not materialize because neither motion is fully staffed at the cadence and depth the two-sided network effect requires. We unpack the structural double-counting problem in the broader industry breakdown at AI for Marketplaces. The sales-specific version of the problem is that the labor required is not 2x a SaaS sales team. It is 2x with side-specific specialization, which a 30-person marketplace cannot afford to staff.

The marketplace economics on personalized outbound look better than almost any other vertical once the network effect is compounding. A single onboarded vendor on a goods marketplace at 10 percent take rate doing 200K in annual GMV is worth 20K in platform revenue per year, gross margin north of 80 percent. A single onboarded host on a stays marketplace at 15 percent take rate doing 80K in annual nights is worth 12K per year. A single onboarded driver on a mobility platform doing 40K in annual rides is worth 4K to 8K per year depending on take rate. The personalization tax of 40 dollars in agent compute to research a target seller against their current channel mix, their listing volume, their category density, and their current platform take rate exposure pays back on the first month of GMV. The math against retail-only acquisition cost models is not in the same universe.

The data shapes inside marketplaces also fit the agent model better than most verticals once you have network effects to work against. Public supply data on most marketplaces is queryable through scraping or partner APIs. Vendor reviews, listing volume, category density, and competitive marketplace presence are all queryable. On the demand side, GMV cohort data from Stripe Connect is structured. Buyer behavior data from your own analytics stack is structured. Geographic concentration on both sides is mappable. The agents pull all of those in parallel, build a real picture of which sellers to prioritize in which regions, and write outreach that references the competitor marketplace they currently sell on, the listing volume they ship today, and the take rate delta against your platform.

The output we see across marketplace engagements is consistent. On the supply side, 100 to 150 enterprise-style touches per week with three to five percent reply rates against priority seller targets, 15 to 25 quality vendor conversations a month routed into the supply manager queue. On the demand side, 400 programmatic touches a day against buyer acquisition funnels in every market language, 80 to 150 first transactions per week from net-new demand acquisition. For the full breakdown of where the fractional model fits across both sides of the marketplace, including ops, content, and support, see AI for Marketplaces. The sales motion is where most marketplaces start because the two-sided acquisition problem is what board decks call out first.

Monday morning the agents ship a two-paragraph recap to the head of growth. The supply paragraph covers which priority sellers replied last week, which categories are running ahead of supply density target, and which competitor marketplace lost share in which region. The demand paragraph covers which acquisition channels converted highest, which language markets are tracking ahead, and which cohort retention bucket needs intervention. Ten minutes of reading and a thumbs-up on the angle adjustments for the week.

Tuesday through Friday the two sides run in parallel. 100 supply-side enterprise touches a week against priority seller targets, each one with 30 minutes of research baked in. 400 demand-side programmatic touches a day against buyer acquisition funnels, language-aware per market. The supply manager opens HubSpot or Salesforce and sees a warm-reply queue of 4 to 6 vendor conversations a day with the full enrichment context, the GMV scoring, and the current competitor platform exposure already attached. The demand manager opens the funnel dashboard and sees the first-transaction count climbing against the cost-per-acquisition target.

By Friday the supply pipeline shows 15 to 25 new qualified vendor conversations from the week and the demand pipeline shows 80 to 150 net-new first transactions. The GMV contribution from each motion tracks separately so the board model finally maps cleanly to the unit economics. The two-sided network effect compounds because the supply-side priority targeting is informed by where demand is growing fastest, and the demand-side targeting is informed by where supply density crossed network-effect threshold last week. The 3-rep-plus-manager plan that the board signed off on at the Series A is not on the payroll, and the take rate growth that the model promised is finally materializing. For the integrated view across supply, demand, ops, and support on the marketplace side, see AI for Marketplaces.