240,000 retailers identified. Zero reliance on data vendors.

Context

Blingsting is a $7M consumer safety brand. They make personal safety products — pepper spray, safety alarms, car escape tools — designed for retail display and impulse purchase. Not the kind of product that sells through a sales team. The kind that sells through shelf placement in the right stores.

They had 12,000 retail customers across boutiques, jewelry stores, pharmacies, home goods stores, gift shops, beauty shops, and clothing stores. A broad but fragmented customer base, spread across independent retailers nationwide. The kind of market where no single channel covers more than a fraction of the addressable universe.

Faire was their primary marketplace. It was how new retailers discovered Blingsting, placed first orders, and reordered. Faire controlled the buyer relationships, the communications, and the discovery algorithm. Blingsting had customers — but Faire had the infrastructure.

Blingsting didn’t have a system to manage their customers independently. No CRM. No outbound capability. No owned data infrastructure. When a retailer stopped reordering, there was no automated follow-up. When a new category showed promise, there was no way to prospect into it. The operating model was simple: list products on Faire, wait for retailers to find them, fulfill orders.

Then Faire cut them off. Overnight, Blingsting lost their primary sales channel and their access to 12,000 buyer relationships.

The problem

This wasn’t one problem. It was three, compounding simultaneously.

No owned data

12,000 customers, but no centralized system to manage, segment, or reactivate them. Customer relationships existed inside Faire’s platform — order history, buyer contact info, communication threads, reorder patterns. When access was revoked, the relationship data went with it.

Blingsting could reconstruct some of this from their own fulfillment records — shipping addresses, order volumes, product mix. But they had no emails, no phone numbers, no direct line to their own customers. The data that would let them reach out, follow up, or re-engage lived on a platform they could no longer access.

No discovery channel

Faire was how new retailers found Blingsting. The marketplace’s search algorithm, category browsing, and recommendation engine drove discovery. Without it, there was no mechanism for new prospects to find the brand.

Blingsting had no outbound capability, no content engine driving inbound, and no presence in the channels where independent retailers search for new products. Their website was built for consumers, not wholesale buyers. Their social media spoke to end users, not store owners. Every acquisition channel they had was either Faire-dependent or consumer-facing.

No outbound infrastructure

Even if Blingsting could identify target retailers, they had no system to reach them. No scraping capability to build prospect lists. No enrichment pipeline to find contact information. No campaign tooling to send personalized outreach. No scoring system to prioritize who to contact first. Starting from zero on every axis.

This is the dependency problem manifesting all at once. The three problems aren’t independent failures — they’re symptoms of one architectural decision: building a business on infrastructure you don’t own. Faire didn’t fail Blingsting. Blingsting’s pipeline architecture failed Blingsting. Any single-channel dependency — a marketplace, a data vendor, an ad platform — carries this structural risk.

The question wasn’t “how do we get back on Faire?” It was: how do we build a pipeline that doesn’t depend on any single platform, vendor, or data provider?

The approach

Five steps. Each one produced a concrete, auditable deliverable.

Step 1: Define the target universe

We analyzed Blingsting’s existing customer base to identify which retailer categories drove the most success. Not just “stores that bought” — categories where the product had natural shelf placement, strong sell-through, and repeat ordering patterns.

Seven categories emerged: boutiques, jewelry stores, pharmacies, home goods stores, gift shops, beauty shops, and clothing stores. These weren’t guesses. They were the categories where Blingsting’s existing customers concentrated, where the product fit the retail environment, and where reorder rates were highest.

These seven categories became the target universe. Every subsequent step filtered through this lens. If a business wasn’t in one of these categories, it didn’t enter the pipeline. This constraint was deliberate — it’s better to deeply cover seven high-fit categories than shallowly cover fifty.

Step 2: Build the dataset

We scraped Google Maps across all target categories. Not a sample — a comprehensive sweep of independent retailers nationwide. 240,000+ businesses identified across the seven categories.

This is not a purchased list. It’s a proprietary dataset built from public sources, filtered to match Blingsting’s proven customer profile.

Standard B2B data providers don’t cover independent retail. A boutique in Tulsa isn’t in ZoomInfo. A specialty pharmacy in rural Ohio isn’t in Apollo. A gift shop in a college town isn’t in 6sense. These businesses exist on Google Maps, on their own websites, on Instagram — but not in the databases that B2B sales tools rely on. Building the dataset from public sources accessed the market that vendor data structurally misses.

The scale matters. Blingsting’s existing customer base was 12,000. The sourced dataset was 20x that — revealing a target universe they didn’t know existed. Not because the retailers were hidden, but because the standard tools for finding B2B prospects don’t look where independent retailers live.

Step 3: Enrich every record

Multi-source enrichment turned raw business listings into actionable intelligence profiles:

• Website scraping for direct email addresses and phone numbers — not from a vendor API, but from the business’s own contact page
• Social profile discovery for engagement context — Instagram presence, Facebook pages, follower counts that indicate brand activity level
• Search engine results for digital footprint — does this business have a web presence beyond a Google Maps listing? Are they active online?
• Geographic enrichment for regional personalization — nearest metro area, distance from major markets, regional context that enables location-aware outreach

Each enrichment source operated independently. A business with no website might still have a Facebook page with a direct email. A business with no social presence might have a verified phone number on their Google listing. Multi-source cross-validation caught bad data that single-source enrichment would have passed through — bounced emails from one source flagged when another source showed a different address.

Where websites existed, the system achieved a 72%+ email discovery rate. That’s direct extraction, not lookup through an intermediary database with stale records.

Step 4: Score and segment

Raw data isn’t ready for outbound. It needs filtering, scoring, and prioritization.

Category filtering excluded 150+ non-target categories before any lead entered the pipeline — medical facilities, government buildings, restaurants, wholesalers, regulated industries (tobacco, cannabis, liquor), service-only businesses (therapists, consultants), and B2B-only operations (manufacturers, industrial suppliers). This wasn’t a simple keyword filter. Google Maps categories are noisy — a business listed as “gift shop” might actually be a hospital gift shop. Scoring evaluated category fit alongside other signals.

Quality scoring assessed each remaining lead on data completeness, enrichment coverage, and category alignment. A lead with a verified email, active website, Instagram presence, and clear category fit scored higher than a lead with just a phone number and a Maps listing. The scoring system prioritized leads where the system had enough intelligence to support personalized outreach — not just “we have an email.”

The result: a prioritized target list where every entry had been validated against the target universe, enriched from multiple independent sources, and scored for outreach readiness.

Step 5: Personalized outbound

AI-powered name normalization ensured outreach addressed businesses naturally. “Riverside Boutique” — not “Riverside Boutique LLC DBA Riverside Gifts Inc.” This sounds minor, but it’s immediately visible in the subject line and opening of every email. Getting the business name wrong is the fastest way to signal “this is a mass email.”

The normalization system maintained a cache of 274,000+ processed names with a 95%+ hit rate — meaning most names resolved instantly from prior processing, with only genuinely new businesses requiring AI classification.

Outbound copy was tailored by category, geography, and retailer profile. A boutique in Nashville received a different message than a pharmacy in Portland — not because of template branching, but because the copy was generated from each lead’s actual profile data. Category context, regional relevance, and business characteristics all informed the messaging.

Exported directly to Smartlead for campaign execution. The system handled the full loop: sourcing → enrichment → scoring → copy generation → campaign delivery. No manual research. No spreadsheet handoffs. No BDR spending an hour per prospect on LinkedIn.

We subsequently deployed the same sourcing and enrichment pattern against Faire’s marketplace data, identifying and classifying 14,000 wholesale brands by quality tier using AI-powered text and visual analysis — confirming the architecture works across data sources and verticals.

What made the data different

Half of the cold email campaign success was because the data was better than anything you could buy. Here’s why.

Coverage

Standard B2B data providers are optimized for enterprise and mid-market companies with established digital footprints — the kind of businesses that have LinkedIn company pages, Crunchbase profiles, and press releases. Independent retail falls outside their coverage model entirely.

A boutique owner in a strip mall doesn’t have a LinkedIn company page. A gift shop in a college town doesn’t show up in Crunchbase. A specialty pharmacy in a rural county doesn’t generate press releases. These businesses exist — on Google Maps, on Instagram, on their own .com domains — but not in the databases that B2B sales tools query.

Building the dataset from public sources accessed the market that vendor data structurally misses. Not because the vendors are bad at what they do, but because independent retail isn’t what they’re built to cover.

Freshness

Vendor data is aggregated through intermediary layers. A business updates its website, and eventually that change propagates through the data supply chain — scraped by a data vendor, normalized, cached, maybe enriched by a second vendor, then sold through an API or dashboard. By the time a record reaches your outbound tool, it’s been through three or four systems, each with its own update cadence and caching layer.

Direct scraping produces data that reflects the current state of the business. Current website. Current contact page. Current social profiles. No stale caches from an aggregator that hasn’t re-crawled this ZIP code in six months.

Enrichment depth

A vendor record gives you a company name, maybe an email, maybe a phone number. The proprietary pipeline produced full intelligence profiles: website content, social presence, geographic context, category classification, business characteristics, and engagement signals.

This depth enables personalization that template-based outbound can’t match. The system doesn’t just know that a business exists — it knows what category they operate in, where they’re located relative to major markets, whether they have an active online presence, and what channels they’re reachable through. That context flows directly into outbound copy.

Exclusivity

This is the competitive advantage that compounds over time. Your competitors can’t buy this dataset because it doesn’t exist as a product. It was built for this specific engagement, against this specific target universe, from sources that aren’t packaged and resold through data vendor APIs.

Every outbound campaign that runs against proprietary data is reaching prospects that no one else is reaching with this level of profile depth. That’s not a marginal improvement in reply rates — it’s a structural advantage in market coverage.

The compounding effect on cold email

Cold email success is usually attributed to copy quality, send timing, or volume. But the single largest driver is data quality. Are you reaching the right person? At a real address? With enough context to say something relevant?

When the underlying data is proprietary, fresh, deeply enriched, and exclusive, every downstream metric improves. Deliverability improves because addresses are verified through direct scraping, not cached lookups. Open rates improve because subject lines reference real context, not generic variables. Reply rates improve because personalization is genuine — the email references the recipient’s actual business, not a template placeholder.

Most cold email optimization focuses on the copy and the cadence. This system optimizes the layer underneath: the data.

Results

• 240,000+ independent retailers identified across seven target categories
• 20x the size of Blingsting’s existing customer base — a target universe they didn’t know existed
• Multi-source enrichment producing verified contact profiles across website, social, search, and geographic data
• 72%+ email discovery rate where business websites existed — from direct scraping, not vendor lookups
• 150+ non-target categories filtered before any outbound was sent
• AI-powered personalization tailoring outbound copy by category, geography, and business profile
• 95%+ name normalization cache hit rate for natural, human-readable business addressing
• Proprietary dataset built from public sources — no vendor dependency, no recurring data subscription
• Cold email campaigns with strong response rates, driven primarily by data quality over volume

Blingsting went from platform-dependent — no outbound capability, no owned data, no way to contact their own customers — to owning a proprietary dataset 20x their existing customer base, with a system that sources, enriches, scores, and executes outbound independently of any marketplace or data vendor.

The system didn’t just solve the immediate crisis. It put Blingsting in a structurally better position than before they were on Faire. They now own their pipeline infrastructure instead of renting it. The dataset compounds with each deployment. The enrichment pipeline runs independently of any single data source. And the outbound system scales with compute, not headcount.

What this means for your pipeline

The pattern is repeatable. Define your target universe. Build proprietary data from public sources. Enrich through independent channels. Score before you send. Personalize from real context. Execute at scale.

The specific categories change — retailers for Blingsting, something else for your business. The architecture doesn’t.

This approach works best when:

• Your target market is fragmented — many small businesses, not a few large enterprises. The kind of market where no single data vendor has comprehensive coverage.
• Standard data providers have gaps in your vertical — your prospects aren’t in ZoomInfo, Apollo, or 6sense because they don’t fit the profile those platforms are built to cover.
• Your competitors use the same vendor data — everyone’s outbound starts from the same export, producing the same messaging into the same inboxes.
• You’ve been burned by platform dependency — a marketplace, a data vendor, or an ad platform changed terms and your pipeline broke.
• You need to scale outbound without scaling headcount — adding territories or verticals shouldn’t require proportionally more BDRs doing manual research.

It’s not the right approach when your TAM is 200 enterprise accounts that are already well-covered by existing data. Or when your outbound challenge is messaging, not targeting. Or when you don’t have the operational capacity to handle the response volume a system like this produces.

Not everything needs custom data infrastructure. But when the data layer is the bottleneck, no amount of better copy or better tools fixes it.

Read how the system works →

Ready to build your own pipeline? Let’s scope it →