Landbase
Toolkit · Seed-Expansion
← All tools
T2 · Tool

Seed-Expansion

Provide 10–325 known fits. Landbase returns a ranked look-alike audience scored across many attributes at once — description, keywords, size, location, tech stack, hiring signals.

Examples

How customers use it

01
Finding more accounts like your wins
Paste 50 of the best closed-won accounts and ask Landbase for more like them. Seed Expansion returns thousands of look-alikes ranked across description, language, size, and tech stack — clustered by what they actually do.
02
Finding every account where a competitor is incumbent
Drop in 15 wins against the same competitor and ask for every other prospect where that competitor is installed. Seeding the wins surfaces the rest of the look-alike universe without manual research.
03
Expanding into a niche you've cracked
Crack a specific niche — "DTC subscription boxes that ship cold" — and ask for more of that exact shape. Seed Expansion picks up the niche signature from a small seed set and returns the long tail no filter UI can articulate.
04
Going beyond industry codes and headcount
Most platforms find look-alikes on shared industry codes or employee bands. Seed Expansion finds similarity on description, language, customer mix, and signal patterns — the soft signals ZoomInfo and Apollo simply can't see.
Pairs with
T4Match T6AI Qualification T5Enrich
Found in
Seed Expansion TAM Mapping
Under the hood+
How it's used

The strongest path when the customer can name fits but can't articulate the filters that describe them. Often used as its own discrete workflow to find look-alikes; also a core step inside TAM Mapping. When seeds aren't provided, Landbase synthesizes a small seed set from the ICP first, then expands.

Why it matters

Industry taxonomies are unreliable — in one real TAM, ~80% of the right companies were tagged outside the obvious industry (a BPO labeled "Advertising Services"; a 3PL labeled "IT Services"). Filter-based search misses them. Seed expansion finds them by similarity, regardless of how anyone tagged them.

The expansion pipeline
Form Seed Set
20–325 seeds
→
Data Retrieval
pull seed profiles
→
Seed Analysis
per-seed similarity
→
Preview Expansion
validation set
→
Full Expansion
up to 50K results
→
Compare / Contrast
signal keywords
Seed expansion runs on top of the Similar Company Graph (see Semantic Search). The compare/contrast step then converts the look-alikes into reusable signal keywords for downstream fit scoring.
Stage by stage
  • Form seed set. 20–325 customer-provided fits (or synthesized from the ICP). Quality of seeds drives quality of output.
  • Data retrieval. Pulls each seed's full profile from the buying graph — description, keywords, attributes, technographic adoption.
  • Seed analysis. Every candidate company in the 21.6M corpus is scored against each seed. Top 5 seed matches are retained per result.
  • Preview expansion. A small validation set surfaces first — a sanity check that the seeds are producing good look-alikes before the full run commits.
  • Full expansion. Returns up to 50,000 ranked look-alikes, each tagged with which seed(s) it matched. End-to-end under 10 minutes for a 200-seed input.
  • Compare / contrast. A lightweight AI qualification pass splits the look-alikes into yes / no. A keyword-divergence analysis then surfaces words that appear in the "yes" group but not the "no" group — signal keywords — which power downstream fit scoring.
By the numbers
20–325Seed domains in
5K–50KLook-alikes out
Top 5Seed matches per company
< 10 minEnd-to-end at 200 seeds