Same outbound, opposite result: what the 2026 State of B2B GTM Report says about the signal layer

Kyle Poyar and Maja Voje polled 300 B2B software and AI companies for the 2026 State of B2B GTM Report, published 14 October. The headline is pipeline. The finding worth acting on sits across three tables the report never joins up.

Cold outreach lands in the over-saturated quadrant: popular, low odds of impact, very poor ROI under $5k ACV and no better than OK anywhere else. Agentic campaigns, the AI SDR category, are the worst row on both ROI heatmaps, with 7% of companies calling the channel core and adoption down four points. The free-text answers to “what didn’t work” are an outbound post-mortem: spray and pray, AI SDR sequences with near-zero reply rates that hurt sending reputation, cold email with superficial personalisation.

Signal-based outbound sits in the opposite corner. It is in the next big thing quadrant, the second-largest investment bet for 2027 at 48% of respondents, and it earns excellent odds of ROI above $50M ARR. Cold outreach gets 22% planning to invest more. Signal-based gets more than double that.

The activity is identical: a message to someone who did not ask for it. One version is being abandoned and the other is where next year’s budget is going. The difference is the layer that decides who gets the message and when, and the report’s own data shows most teams have barely built it.

The signals teams actually act on

Asked which signals they act on, three clear 40% in any deal-size segment: product usage, content engagement and website visits. Then the table drops off a cliff. Champion movements: 11% to 35%. Hiring activity: 11% to 31%. Leadership changes: 13% to 33%. Funding events: 13% to 19%. Third-party intent: 13% to 24%. One in six sub-$5k companies acts on none of them.

The signals almost everyone uses are the ones a buyer generates by already being in your orbit. They tell you a known account is warming up. They say nothing about an account you have never spoken to entering a buying window. The signals that do that, a champion landing as VP somewhere new, a target account posting three RevOps roles, a new CRO with ninety days to show a plan, a round closing, are the ones in the bottom half of the table.

So for most companies calling it a channel, “signal-based outbound” in 2026 means outbound to website visitors. That is retargeting with a new name, and it is a different motion from the one the >$50M ARR segment rates as excellent.

The staffing data says the same thing

Sumble’s analysis of job posts across 5,990 AI-native B2B companies, included in the report, puts a number on it. AI-native companies run 12.8 GTM engineers per 1,000 GTM people. Other B2B companies run 0.3. They staff twice the RevOps per seller, and GTM engineer ties with account executive as the top priority hire above $10M ARR.

The teams whose outbound works are hiring the people who build the layer underneath: the pipeline that catches a leadership change within a day, the enrichment that confirms the new VP is the right persona, the routing that puts the account in front of a rep with context attached. One respondent summed up the rest: “AI is used for everything right now. But there’s no AI that actually generates SQLs consistently.” An AI SDR pointed at a static list burns a domain faster. The same model pointed at a champion-move feed with enrichment behind it is a different product.

Acting on the four neglected signals

Champion movements. Re-enrich every closed-won contact’s employer and title monthly in Clay. A move to an ICP account is the highest-intent signal you own: the person knows the product and has a reason to call. Route it to the rep who owned the original deal, with the deal summary attached, inside a day. A few hundred credits a month.

Hiring activity. Scrape postings for the target account list weekly. Trigger on roles that imply your problem (a first RevOps hire, three SDR roles at once), score by title match and volume, and fire only when an account crosses a threshold so one generic listing does not launch a sequence.

Leadership changes. New VP or C-level in the function you sell to, inside their first 90 days. The message is different: this person is building a plan and wants inputs, not a demo. The window is short, so this is where signal-to-sequence latency matters most. A webhook from the enrichment tool into n8n that creates the task the same day is the whole build.

Funding events. Filter hard by stage and ICP fit, then combine: a Series B plus two relevant hiring posts plus a site visit in 30 days is a buying window. Any one alone is noise.

The pattern is the same each time. The signal is public. Catching it is a scheduled job. Confirming it is enrichment, where provider order decides the cost. Acting on it is a routed task with context, and latency is the metric. All of it needs someone who can build and maintain the pipeline rather than another seat on the sequencer, which is what the hiring numbers show.

Where this leaves a Q4 plan

AI search is the report’s top 2027 bet at 50%, and it is the same shift from the other side: research moving into assistants removes the website-visit signal most teams lean on. The public signals above matter more as that happens. The authors’ closing frame, a barbell of relationships and automation, is right; network and intros is the top channel at every deal size above $5k. The signal layer decides which 40 accounts get the dinner invitation and which rep makes the intro call this week.

If signal-based outbound is on your 2027 plan, look at the list of signals you act on before you look at tooling. If it is product usage, content and web visits, the sequencer you are about to buy will run on the same list cold outreach ran on, with the results the report describes. Build the champion-move feed first, because it is cheap and produces a pipeline number inside a quarter. Add hiring and leadership changes next. Put a person on the pipeline before a seat on the sequencer. Where signal-based outbound sits alongside the other channels in a 2027 plan, and how to sequence the build, is in The 2027 Demand Generation Execution Guide.