5 Public Buying Signals That Beat Firmographics (2026)

For twenty years, B2B outbound targeting was firmographic. Pick an industry (SIC / NAICS code). Pick a company size band. Pick a country. Buy a list. Load it into a sequencer. Send.

That model produced 2 to 4 percent reply rates when it worked. It never worked well; it just worked at scale, and scale hid the mediocrity. In 2026 it is finally breaking, for two reasons that compound: sending volume tripled between 2022 and 2025, and the same firmographic lists get used by three hundred vendors simultaneously. The buyer’s inbox filters out anything that reads like a list-driven send. What survives is outbound that reads like the sender already knows something specific.

That specific something is a buying signal. The five signal types below are the ones that, in 2026, an AI agent can watch continuously without producing noise, and that produce reply rates 3 to 8× above firmographic-only sequences with identical messaging.

Why firmographics stopped working around 2023

Three things changed simultaneously.

First, sequencer software got cheap. Every sales team can now send 500 messages a day for €200/month. Total sending volume tripled between 2022 and 2025.

Second, LLM-generated first drafts got good enough that mid-tier vendors started producing plausible-looking personalization at scale. The buyer’s cognitive filter, calibrated on 2019-era templated outbound, could no longer separate a real conversation opener from a synthetic one.

Third, the same firmographic list (VP of Sales at 100–500 employee SaaS companies in DACH region) is now sold by six data vendors and used by three hundred outbound teams. If your reason for messaging a prospect is that they match a firmographic filter, three hundred other people had the same reason last month.

Firmographic targeting stops being differentiation and starts being the definition of noise. The only signal that survives is one that most senders cannot see — because it requires reading behaviour rather than buying lists.

Signal #1: Hiring

The strongest public buying signal in 2026, by a wide margin. A job posting is:

• Public — appears on the company’s careers page, on LinkedIn, on job boards
• Timestamped — posted date is exact
• Specific — the role description names the tools, categories, and problems the hire will address
• Self-declared — the company itself said this is what they need

A recruitment agency reads “Senior SDR with HubSpot experience, based in Munich” and sees a company that (a) is expanding sales, (b) uses HubSpot, (c) has budget signed off, and (d) is about to spend €80–€120K/year on a person. Every one of those inferences is stronger than any firmographic filter.

Job signals work across almost every B2B vertical: hiring an ops manager implies logistics spend; hiring a compliance officer implies a regulator conversation; hiring a QA engineer implies a product release. See our industries pages for how we scope this by vertical.

Reply rate benchmark: 15–24% on job-signal sequences for well-scoped ICPs (theagency47 client data, 2026), versus 3–6% for firmographic-only sequences with identical messaging quality.

Signal #2: Content

Publications map to categories. A restaurant that publishes its menu online is telling you which suppliers it uses. A brand that photographs its packaging on Instagram is telling you which format it currently prints. A consultancy that publishes a case study on “our SAP migration project” is telling you it does SAP work.

The 2026 stack can now read these publications reliably: menu images turn into ingredient categories, packaging photos turn into format-and-material specs, case studies turn into service-line inventories. Then you match your product to what the publication already implies the company buys.

This works best when your category is visually or textually obvious to third parties: food and beverage supply, packaging, hospitality equipment, professional services offerings. See /industries/food-beverage-exporters/ and /industries/packaging-suppliers/ for two current builds that use content signals as the primary matcher.

Signal #3: Regulatory

Trade filings, import/export declarations, licensing databases, permit applications, product notifications to regulators, and public procurement portals are all timestamped, public, and specific. They are the least-known signal type in B2B outbound because they require domain knowledge to interpret.

Examples:

• A logistics company sees an exporter file customs declarations for a new route — that exporter is a candidate to switch or add a freight forwarder.
• A medical device supplier sees a clinic file a treatment-authorization application — the clinic will need consumables within 30–60 days.
• A packaging supplier sees a new SKU appear in a food safety authority database — the brand is about to launch and the packaging spec is still open.

Regulatory signals are hard to work with because the databases are fragmented and often unstructured. But when an AI agent has been trained on a specific database’s format, the signal is nearly ground truth about intent. See /industries/medical-dental-suppliers/ and /industries/logistics-freight/ for verticals where regulatory signals dominate.

Signal #4: Stack

Public tech-stack changes are readable from three sources: job postings (already covered above), careers pages (“we use these tools”), and product documentation (integrations, plugins, third-party credentials).

A stack signal is not always a buying signal — but a stack change almost always is. If a company was on Salesforce last quarter and is now hiring a HubSpot admin, someone signed a HubSpot contract in between. The company that displaces incumbents in a stack transition is a company that shows up while the transition is still open.

For B2B SaaS vendors, stack signals are usually the top signal type — see /industries/vertical-saas/.

Signal #5: Directory / procurement

Directories and supplier lists are boring, high-signal data sources. If your ICP appears in a specialised directory (industry association member list, certification registry, approved-vendor list for a major procurement authority), the presence of that entry is a signal about their operating shape.

More interestingly: procurement portals publish upcoming tenders. Public sector, healthcare, and utilities buyers all pre-announce large purchases. The tender description is the buying signal — timestamped, specific, and often disclosing the incumbent supplier.

The 2026 AI agent stack can watch procurement portals across the EU, translate tender titles to your product taxonomy, score by fit, and surface the top 10 per week to your BD team. That is the shape of a serious procurement-driven outbound engine.

The stack that watches five sources without you

The five signal types above look like five separate operations. In an integrated build they are one operation, run by an AI agent that:

1. Reads job boards, careers pages, and company blogs on a schedule
2. Reads open regulatory databases and procurement portals in your target markets
3. Reads company product docs and integration pages
4. Reads directory / association pages
5. Scores every raw event against your ICP definition
6. Deduplicates to the company level, resolves the entity
7. Ranks the top signals of the week
8. Drafts the outbound message with the specific signal cited in the first line
9. Passes to a human for send approval

The build cost for a single-vertical version of that agent is €9,500 (see /services/ai-sales-agent/ for the €9,500 build + €1,400/mo operating economics). The build cost for a firmographic-only sequence is €0 (buy a list), but the reply rate ceiling is 3–5×.

You are not buying software; you are buying a systematic advantage over teams still using firmographic filters. That advantage is compounding, because every month the noise floor of firmographic outbound rises and yours does not.

Two traps first-time buyers walk into

Trap 1: Confusing signal quality with signal quantity. Watching a hundred low-quality signals per week produces a worse pipeline than watching ten high-quality signals per week. If your team cannot personalise the message to the specific signal cited, the signal was too abstract to begin with. Cut the list.

Trap 2: Trying to build signals across too many verticals at once. Signal libraries are vertical-specific because the sources are vertical-specific. Ship one vertical to reply-rate proof, then add the second. A serious signal stack for one vertical outperforms a shallow signal stack across five.

For the legal side of running signal-based outbound in Europe — which countries require opt-in, which require an address, which permit opt-out — see Is Cold Outreach Legal in Europe?. The signal quality does not save you from the legal regime; both matter.

---

Firmographic outbound is not dead. It is a commodity, and commodities eventually lose to differentiated substitutes. Signal-based outbound is the differentiated substitute. The teams that build the signal engine in 2026 lock in a 2–3 year window before their competitors catch up. The teams that wait until 2028 will be replicating what was public advantage in 2026.

If you want to talk through which of the five signal types fits your product, the 30-minute discovery call is scoped for that conversation.

---

Key terms in this post: AI agent · BDR · SDR · ICP · tool use · ePrivacy