# Mark Deacon

Mark Deacon shares practical analysis of B2B SaaS go-to-market strategy, including pricing, sales and marketing, and AI-driven revenue workflows. The LinkedIn posts also promote interviews with business leaders on the GTM Playbook podcast.

- Profile: https://b2binfluencer.marketing/creators/mark-deacon/
- LinkedIn: https://www.linkedin.com/in/markpdeacon
- Topics: artificial-intelligence, sales, revenue-operations, ai-agents
- Based on 17 dated posts, latest 2026-10-07

## Summary

Mark Deacon is most useful for practical B2B SaaS go-to-market analysis, especially pricing and AI-driven revenue workflows.

- Best for: Series B B2B SaaS workflow software teams needing fractional GTM and pricing advice.
- Not for: Paid sponsorship packages with guaranteed reach, media kit, and standard usage rights.
- Check first: Confirm client results, self-reported figures, and audience analytics before contracting.

### Takeaways

- Price on a unit of value, with a usage meter under the seat.
- Defensibility is accumulated context, not the underlying model.
- An agent automates government tender discovery, HubSpot recording, alerts, and draft preparation for human review.

## Main ideas

- **Price on a unit of value, and put a meter under the seat — not the other way round**: Asked how they would build a pricing model from scratch in 2026, Deacon rejects seat-based pricing as the starting point — not because seats are dead, but because they answer the wrong first question. Headcount and logins are poor proxies for value once agents do the work. The sequence they set out is: choose the unit of value before the unit of pricing, instrument it before a single contract is signed because retrofitting telemetry into live billing is a mess, then layer a small seat base on top of a variable meter so the customer gets a predictable floor and the vendor keeps upside. The stated tension is what to meter: actions are easy to count and easy for procurement to argue down, while outcomes are hard to count but much easier to sell. They leave that as "the harder call" rather than resolving it. The rollout recommendation is deliberately narrow — a hybrid model on new logos only, since migrating existing seat contracts is a separate, painful project. The failure mode named is customers quietly renewing the same seat count while doing three times the work through agents: same ARR, half the value capture.
- **Defensibility is accumulated context, not the underlying model**: Reading Anthropic's roughly $965 billion valuation against about $47 billion of run-rate revenue as a 20x multiple, Deacon argues the market is not betting on benchmark performance but on context. Once a company wires a model into procurement, code review, customer operations and finance workflows, that integration learns how the business operates — a point they credit to Sequoia's Alfred Lin — and accumulated context makes removal feel like a re-organisation. They apply the same curve to SaaS generally: products that accumulate workflows, decisions, institutional memory and messy judgment calls get harder to switch every quarter, while products that wrap a model and re-skin a user interface get cheaper to replace every quarter. One side compounds, the other commoditises.
- **AI is pulling twenty years of sales specialisation back into full-cycle roles**: This argument comes from a named guest, not Deacon: Michael Hartman, who leads go-to-market at Anthropic, said in Boston that AI is undoing twenty years of sales specialisation. Deacon relays and expands the point, tracing how one account executive role split into SDRs, BDRs, AEs, solutions engineers, RevOps, CSMs and AMs. They argue that AI removes much of the manual work inside those roles — research, prospecting, account planning, follow-up, CRM administration, proposal preparation, finding information, analysing customer data — while the responsibilities themselves remain. The result they point to is a generation of full-cycle sellers judged on outcomes rather than task completion, with human context and judgment still in the loop. They emphasise a reframed leadership question: instead of asking which jobs AI can replace, ask what a person could become capable of if AI removes half the work from the role.
- **Replace the manual revenue process with an agent, not a better checklist**: Deacon's most concrete build example is government tendering. Eighteen months after expanding into public sector as a vertical, the old approach was someone logging into portal after portal every day, reading every new listing, deciding fit, hand-creating Company, Contact and Deal records in HubSpot and assembling a 20-plus page submission from a blank page. The replacement agent runs daily across Australian, New Zealand, UK, EU and US portals, pulls each new tender at the source, judges relevance, creates HubSpot records with the portal and source URL stamped on them, posts to Slack when something looks promising, and drafts the submission so the team only reviews and submits. The post sets the old and new approaches side by side and asks other people selling into government what they have learned.
- **Outbound list building becomes one command, with the plumbing handled by a data waterfall**: Deacon describes years of assembling prospect lists across Apollo for emails, ZoomInfo or Lusha for mobile numbers the others missed, and Sales Navigator for LinkedIn URLs — enriching, exporting, finding the gaps and enriching again into what the post calls a Frankenstein CSV built over hours or days. The replacement is a Claude skill plus FullEnrich connected through its MCP: a /find-contacts command with a target account list attached waterfalls through more than twenty data providers, and the enriched contacts push into HubSpot and into outreach sequences. On a 400-account list the reported result is roughly half a day down to about twelve minutes, a mobile and email hit rate above 86%, and around double the coverage of any single provider. The post carries a partner hashtag for the enrichment vendor but gives no payment terms. Ask to see the underlying test before relying on the vendor-aligned figures.
- **Platform other operators' experience through interviews and revenue-leader events**: A large share of what Deacon publishes is a named operator's framework, packaged for an audience, rather than their own operating result. Interview posts follow a consistent shape — the guest's backstory, then an arrow-list of takeaways, then a link to the full episode. Varun Bhandarkar, Stripe's head of SMB and mid-market for Southeast Asia, is presented through why outbound is the last pipeline lever to pull and what must be in place first, the standardise-then-localise framework Stripe uses across very different Asian markets, how to spot expansion signals before a customer mentions budget or renewal, an interview question for identifying coachable hires, and the three moments where churn is actually decided rather than at renewal. Nicholas Holland, HubSpot's former SVP of product and head of AI, is presented on scaling Marketing Hub from $400 million to $1.6 billion ARR, Content Hub from zero to $130 million, the free CRM launch, and what HubSpot learned when customers objected to paying for AI they previously received free. Alongside the podcast, Deacon promotes and reports from revenue-leader events, including a Chief Revenue Officer Summit in Sydney and UNBOUND 2026 in Boston, where the recurring theme they identify is AI-first GTM: the companies getting ahead are rethinking how the whole revenue engine operates rather than adding AI tools to old processes.

## Work to read

- [How Mark Deacon would build a pricing model from scratch in 2026](https://www.linkedin.com/posts/markpdeacon_heres-how-i-would-approach-a-pricing-model-activity-7477983429016489984-RW3C)
- [A $965 billion valuation is a bet on context, not on the model](https://www.linkedin.com/posts/markpdeacon_a-965-billion-valuation-on-47-billion-of-activity-7470211213193560064-B8dU)
- [An agent that automates government tender discovery and submission](https://www.linkedin.com/posts/markpdeacon_about-18-months-ago-we-expanded-into-government-activity-7493431966445326336-4sa1)
- [Replacing a multi-provider prospect-list process with one Claude command](https://www.linkedin.com/posts/markpdeacon_gtm-outbound-claude-activity-7495979912009236480-mhe_)
- [Michael Hartman of Anthropic on AI undoing sales specialisation](https://www.linkedin.com/posts/markpdeacon_michael-hartman-leads-gtm-at-anthropic-and-activity-7510503785983610880-UvgB)
- [Stripe's Varun Bhandarkar on the GTM Playbook podcast](https://www.linkedin.com/posts/markpdeacon_gtmplaybook-saas-growth-activity-7442343867577745408-4uYz)

## Brand relationships

- **FullEnrich** (disclosed_partnership)
- **Anthropic (Claude)** (mention)
- **HubSpot** (mention)
- **GTM Playbook Podcast** (mention)
- **Revenue Operations Alliance** (mention)

## Partnerships

Mark Deacon's ten LinkedIn posts show three possible ways for a brand to work with them, though none is presented as a published offer. First, a post walking through a contact-enrichment workflow carries a partner hashtag naming FullEnrich and reports a specific result: a 400-account list built in roughly 12 minutes, with an 86%-plus mobile and email hit rate. Second, the GTM Playbook Podcast, which they promote and use to interview revenue leaders, including Stripe's Varun Bhandarkar and HubSpot's former head of AI Nicholas Holland. The posts don't describe a sponsor slot or host-read ad. Third, event appearances: they promoted the invite-only Chief Revenue Officer Summit in Sydney and pointed readers to a registration link in the comments, suggesting a speaking or attendance role rather than a paid placement. LinkedIn is the only contact route visible in the posts. Ask directly about payment, exclusivity, approval, usage rights, sponsor slots, deliverables, disclosure wording and pricing.

## Alternatives

- Ori Shimon: Choose Ori Shimon when the blocker is an unreliable pipeline view, not uncertainty about pricing. The posts show how to build an internal GTM layer on top of the CRM, call recordings and enrichment tools, giving each rep a view of their open pipeline, surfacing bottlenecks and enriching every deal from the same sources. The second post covers verification: an issue does not close when a fix is pushed; it moves to a checking stage, where a real browser runs the scenario against production. If the check crashes or cannot tell, the outcome is unknown, not fixed. That discipline matters to a company selling agent-assisted workflows: it answers the operational question behind the pricing need—how do we know a shipped agent change actually altered behaviour before we charge for it?
- Sergey Danishevskiy: Choose Sergey Danishevskiy when the live question is build versus buy and the company needs a costed answer and a way to assess a RevOps partner. One post models three-year total cost of ownership for HubSpot versus building and maintaining a custom CRM, assuming AI-assisted development: 8 seats at $1M ARR, $61,000 versus $225,000; 45 seats at $15M ARR, $480,000 versus $1.85M; 250 seats at $100M ARR, $2.8M versus $6M. The conclusion is to buy the core and build only the modules that give the business an edge. The second post offers five tests for selecting a RevOps partner: diagnose before quoting, build rather than advise, plan their own exit so workflows sit in the client’s CRM, cover the full revenue loop, and tie accountability to outcomes rather than hours. Both address the buyer’s task, from the procurement side.
- Suzanne Hill: Consider Suzanne Hill when the company is deciding what category of thing to buy, rather than how to charge for what it sells. The posts frame AI-era procurement as a shift in the buying decision: the choice may be between buying technology, outsourcing a capability, partnering, building internally, procuring a managed service or simply buying the outcome. The questions change accordingly: what is being bought, who owns the data and IP, where risk sits, how dependent the business becomes on the supplier, and what happens if the commercial model changes. The other post argues that scattered qualitative signals across tenders and supplier conversations can now be captured and used rather than lost in someone’s memory. For a Series B company being asked to restructure how its agents are bought and integrated, this framing is useful on the client side of the table.

---
AI-assisted summary of public posts. Audience demographics are not available. Engagement appears only where we checked recent posts. Method: https://b2binfluencer.marketing/methodology/