RevOps in 2026 · The Year of AI Survey

73% of mid-market B2B SaaS
teams have an AI mandate.
38% have the data to support it.

We surveyed 123 RevOps practitioners — Managers, Directors, VPs, freelancers — at AI-engaged B2B SaaS companies. Boards are mandating it. Vendors are selling it. The plumbing underneath says something different. And this is the leading edge — the broader market is almost certainly worse.

73%
have a board / CXO-level
AI mandate
38%
have a unified
customer data layer
61%
build their own
AI tools
37%
are hiring less
because of AI
N = 123 RevOps practitioners
66% SaaS · 75% mid-market (51–1k employees)
61% Europe · 22% North America · 10% LATAM
62% Manager / Director · Median team size: 4
Sample is the leading edge — broader market likely worse
01 / THE HEADLINE

Mandates fly. Plumbing leaks.

The boardroom narrative — "we're an AI-first GTM company" — is running ahead of operational reality. Three of every four respondents have a top-down mandate. Less than two of every five have the data foundation to make it work.

AI Mandate from CXO/Board
73%
say yes. 27% say no, "we're just experimenting."
Unified Customer Data Layer
38%
have a CDP, warehouse-native or reverse-ETL setup feeding AI tools.
Mandate, but no data layer
56%
of mandated companies are running AI on fragmented data.
A board-level AI mandate without a unified data layer is a press release, not a strategy. The output quality of any model is capped by the input quality of your CRM, billing system and customer database.
Where the gap shows up: how much of your CRM is updated automatically?
Self-reported share of CRM updates happening automatically (e.g. from call transcripts) vs manually by reps. Lower = more rep keyboard work.
Median: 30% 75%+ automated: 13% of respondents
02 / GOVERNANCE

The governance vacuum.

AI touches CRM data, lead routing, customer comms, pricing and pipeline forecasts. Ask 100 RevOps leaders who owns the governance for it and a fifth say "nobody". RevOps owns it in just 19% of companies — even though it sits on top of the systems RevOps already runs.

Who owns AI governance?
Among 123 respondents.
"Nobody owns it" rate
20%
of companies have no defined AI governance owner.
Even with an AI mandate…
16%
of mandated companies still have nobody owning governance.
RevOps is losing the governance fight to IT — even though every meaningful AI use case in commercial ops runs on the systems RevOps already owns. If RevOps doesn't claim it, IT will define it without you.
03 / BUILD vs BUY

Building beats buying.

The "wait for the vendor" era is over. 61% of teams are building custom AI tools internally — point solutions, custom CPQ, deal desks, dedup agents, deal coaches. Salesforce-native, n8n-wired, Clay-fed, Claude-reasoned.

Building custom AI tools
% of GTM tech spend going to AI
Median across 99 respondents who could quantify it.
20%
25%+ of spend on AI: 49% of respondents.
What people are actually building (sample of in-house tools)
Quote to cash system.
Director · Portugal · Mid-Market
Territory planning & mapping, capacity planning, firmographic enrichment.
Director · SaaS · 200-1k
Sales Coach — sales prep, enablement, content proposal, MEDDPICC scoring.
Manager · SaaS
CPQ, RevOps OS, territory tooling, forecast tooling, notifications.
VP RevOps · SaaS
Deduping and conversion agent, transcription analysis agent.
Director · Mid-Market
Lead researcher, GTM toolstack powerhouse, virtual CSM.
Manager · SaaS
04 / HIRING

The hiring curve isn't bending up.

AI was supposed to free RevOps to hire more — to do more, faster, with the same people. The data says the opposite. Only 9% of teams are hiring more because of AI. 37% are hiring less. The mandate makes it worse, not better.

2026 hiring plans — AI impact
Hiring-less rate, with vs without mandate
With mandate
42%
Without mandate
24%
A formal AI mandate is associated with ~1.75× the rate of "hiring less".
The "AI as productivity unlock" pitch gets sold as a force multiplier. The lived experience is a labour replacement story. RevOps practitioners are reading the room: do more with the headcount you've got, because the next req isn't coming.
05 / TALK vs WALK

What people say vs what they do.

78% rate AI 4 or 5 out of 5 for day-to-day importance. Then you look at the actual usage metrics. The outward edge — customer calls, outbound emails — is more AI-touched than the inward plumbing. Internal calls, CRM hygiene: still mostly manual.

The execution gap
Median % across respondents who could quantify it.
"Do you see ROI on AI investments?" — coded responses
Yes-rate looks healthy on the surface. Read the verbatims and most "yes" answers cite time saved, not revenue. Almost nobody is measuring it rigorously.
"Yes — but in personal efficiency, not revenue yet." · Manager

"Hard to say yet." · VP RevOps

"Yes, time savings of 2–8 hours per rep per week." · Director
06 / THE RECORDING ASYMMETRY

We record customers. Not ourselves.

The biggest single execution gap in the dataset isn't CRM hygiene. It's a question of who teams are willing to record. External customer calls: 87% median transcription. Internal calls — pipeline reviews, deal coaching, forecast meetings, the conversations where actual decisions get made: 38%. A 49-point gap, on the same hardware, with the same vendors available.

External calls (with customers)
87%
median % transcribed
DISTRIBUTION
Internal calls (with colleagues)
38%
median % transcribed
DISTRIBUTION
Record both heavily (75%+)
32%
have transcript discipline across the board.
Customers heavily, colleagues lightly
19%
are transcribing >75% of customer calls but <25% of internal ones.
Record neither
2%
transcribe almost nothing. Effectively flying blind.
Customer transcripts are an "enablement tool" — easy to sell, easy to ROI. Internal transcripts are a "performance management tool" — politically harder, GDPR/works-council exposed in Europe, and threatening to anyone who'd rather their forecast call not be searchable. So the easy half gets done. The valuable half gets postponed.
What transcripts get used for (when teams use them)
Multi-select among 120 respondents who use transcripts.
7% say transcripts are "not really used" — recording happens, but the data sits in a vendor archive nobody mines. The recording asymmetry has a sibling: an activation asymmetry.
Why this matters operationally

Forecast accuracy lives in internal calls, not customer calls. Reps tell customers the optimistic story; they tell their managers a slightly less optimistic story; they tell themselves the real story. If you only transcribe the first conversation, your AI is being trained on the most polished, least predictive version of reality.

The teams that have closed this gap aren't necessarily more sophisticated — they've just made the political decision earlier. If you're a RevOps leader sitting on a customer-call library and no internal one, that's your 2026 plumbing project.

07 / THE STACK

The 2026 RevOps stack.

Two clear stories. Claude has taken pole position as the paid LLM of choice — though this sample carries Anthropic bias. And in automation, n8n has closed the gap with Clay, with Zapier slipping into third place. The code-light Zapier era is fading.

Paid LLM in use (multi-select)
% of respondents whose company pays for each LLM.
Automation tooling (multi-select)
% of respondents using each automation platform.
Reading the stack

Clay vs n8n is the underrated story. Clay sits at 54%, n8n at 46%, Zapier at 41%. n8n is a self-hosted, open-source, low-code automation platform — the fact that it's almost on par with Clay suggests RevOps teams are increasingly comfortable owning their own automation infrastructure. Two years ago this question would have been Zapier dominant. The maturity curve has shifted hard.

08 / WHO ANSWERED

The leading edge, not the median.

Transparency note: every survey carries sample bias. This one was distributed through Revenue Wizards' network and Substack — meaning respondents are practitioners who care enough to spend 10 minutes on a 30-question survey about RevOps and AI. That self-selects toward mid-market B2B SaaS companies that are actively trying to be good at this. If the leading edge looks like the numbers in this report, the median B2B SaaS company almost certainly looks worse.

Region
Org size
Seniority
RevOps reports to
09 / WHAT TO DO

If you run RevOps, do this in 2026.

Five moves the data points to. Opinionated. Borrow what helps, ignore what doesn't.

01
Fix the data layer before the AI layer.
A model can't reason on records that don't exist or contradict each other. Audit CRM completeness, then warehouse-or-CDP, then bolt the AI on. Don't do it in reverse.
02
Claim governance before IT does.
If you run the CRM, the engagement platform and the customer data, you should run the rules for how AI uses them. Write the policy, propose the committee, own the seat.
03
Build, don't wait.
61% of teams already are. n8n, Clay, Claude Code and Salesforce-native tooling cover most use cases at a fraction of the price of waiting for a vendor to ship the right feature.
04
Measure revenue, not minutes.
"Saved time" is the wrong KPI. It becomes someone else's headcount cut, not your win. Tie AI initiatives to pipeline conversion, cycle time, retention, ARR per rep. Things the CFO will renew the budget on.
05
Plan for a flat headcount year.
37% are already hiring less. The req isn't coming. Promote internally, automate the work nobody wants, and stop pretending the team will double.