The method
How we actually work
Agencies keep their method private because the method is what they sell. Here is ours: what we get access to before we start, the four questions we ask of every tool you own, and the rules we work under, including the ones that cost us money.
Most engagements begin with a discovery call and a proposal built from what people say in that call. That produces a plan shaped by whoever was most articulate in the room.
Ours begins with read access and a set of queries. What people believe about their pipeline and what the pipeline says are different documents, and the gap between them is usually where the work is.
Day zero
Nine systems, before any analysis starts
Read access only. Not every company has all nine and the list adapts to what exists, but the order matters: contracts and renewal dates are collected at the start, not discovered at the end.
[ fig. 01 · the access checklist ]
| System | Why | How |
|---|---|---|
| CRM (Salesforce, HubSpot) | Pipeline attribution, closed-won analysis, field hygiene audit | CLI auth or CSV export |
| Call recording (Gong, Chorus, Fireflies) | Conversation signal extraction, ICP language, objection patterns | API or MCP integration, or transcript exports |
| Meeting transcripts (Granola, Otter, Fathom) | Stakeholder sentiment, stated priorities, political dynamics | MCP tool or direct access to notes |
| Knowledge base (Notion, Confluence, Docs) | Existing strategy docs, frameworks, OKRs, roadmaps | Shared access or exports |
| An agent with institutional context | Cross-referencing assumptions against conversations we were not in | API key plus workspace or agent id |
| Stated priorities in the stakeholder's own words | OAuth | |
| Ad and marketing dashboards | Channel performance: cost per lead, conversion rate, spend trend | Exports or direct platform access |
| Vendor contracts and renewal dates | Renewal dates are sequencing constraints, not admin detail | Request from marketing ops or finance |
| The internal war room or project tracker | What is already in flight, what is blocked, who owns what | Direct file access or a shared link |
read access, never write. we do not change anything we were brought in to assess.
The stack audit
Four questions, asked of every tool you own
In this order. Each one has a reading that is easy to get wrong, and the wrong reading is how good tools get cancelled and bad ones get renewed.
[ fig. 02 · question 1 of 4 ]
Is there attributed pipeline?
Search the CRM for campaigns named after the vendor and total what closed against them.
Zero attributed dollars is not a cancellation signal on its own. If reps use the tool daily and nothing is attributed, the attribution model is broken, not the vendor. Check for informal usage before touching the contract.
[ fig. 03 · question 2 of 4 ]
What does the vendor's own dashboard show?
Twelve months of cost per lead, conversion rate, lead volume and engagement, read for trend rather than level.
A tool can look healthy on absolute numbers while degrading every quarter. The slope matters more than the value.
[ fig. 04 · question 3 of 4 ]
Who actually uses it?
Map every tool to the teams that depend on it: demand generation, BDRs, AEs, ops.
This is the disruption cost of any change, and it is the number that gets left out of consolidation plans that then fail.
[ fig. 05 · question 4 of 4 ]
When does the contract renew?
Collect every renewal date before proposing any sequence.
Renewal dates create the only natural decision points you get. A plan that ignores them asks people to break contracts instead of letting them lapse.
The rules
What we will not do
Some of these cost us revenue. They are published because an analysis that exempts its author from its own standards is marketing wearing an analysis costume.
Never recommend replacing an incumbent agency with us.
Scope overlaps with an incumbent constantly. Describe the capability gap and let the client draw their own conclusion. A document that reads as a competitive takeout loses credibility with everyone in the room, including the people who hired us.
not this
This agency is underperforming and should be replaced.
this
Evaluate scope at renewal against the capability gap described above.
Never compete with an internal framework. Extend it.
If a maturity model or roadmap is already in circulation, our work positions itself as the layer that sits on top. Reference it by name, credit the author, use their terminology. This is the difference between a document that gets adopted and one that gets shelved.
List ourselves in the vendor table like everyone else.
Same format, same neutrality, same scrutiny. An analysis that exempts its author from its own criteria is marketing wearing an analysis costume.
State performance data factually and list causes before assigning any.
Vendors and internal teams both have people attached to them. Naming a cause too early turns an analysis into an accusation and ends the conversation.
not this
This channel is failing.
this
Performance has declined over the period; multiple factors may contribute and root cause analysis is required.
Frame recommendations as observations.
“The data surfaces a pattern worth validating” survives a room with competing interests. “This is broken” does not.
Keep anecdote out of the narrative.
Specific emails, messages and transcript quotes belong in evidence callouts and footnotes, never in running prose. A document that quotes someone's complaint in the body makes an enemy on page three.
[ fig. 05 · the arc ]
- 01Access
Nine systems, read only
CRM, call recording, transcripts, knowledge base, email, ad dashboards, contracts, the internal tracker.
- 02Queries
Six attribution queries
What actually closed, sourced by what, and whether the fields can support a scoring model at all.
- 03Findings
Evidenced, not asserted
Every current-state claim traces to CRM data, a transcript, or a vendor dashboard.
- 04Choice
Four sequencing options
The same end state, four paths, priced by the constraint that actually binds you.
the choice comes last. everything before it exists to make the choice an informed one.
The choice at the end
Four paths to the same destination
The argument in the room is usually about what to build. It should be about what constrains you, because that is the only thing that actually differs between these four.
[ fig. 06 · four paths, one destination ]
positions are editorial, derived from the risk profiles rather than measured. the point is the spread, not the coordinates.
[ fig. 07 · the comparison ]
- optimises for
- Time to end state
- sequence
- Everything in one quarter, disruption accepted
- best when
- Leadership sees the current state as actively damaging and holds the political mandate for rapid change.
Risk: Highest execution risk, highest speed
- optimises for
- Predictability, each phase funding the next
- sequence
- Velocity proof and foundation, then expand, then scale
- best when
- Leadership wants comprehensive coverage but values execution quality over raw speed.
Risk: Moderate risk, moderate speed, highest compounding leverage
- optimises for
- Earliest credibility with minimum commitment
- sequence
- Two or three quick-win sprints, bigger commitments contingent on results
- best when
- Budget approval is incremental, political appetite is fragile, or stakeholders are sceptical of transformation programmes.
Risk: Lowest upfront risk, slowest path to end state
- optimises for
- Long-term compounding leverage
- sequence
- One quarter entirely on tagging, scoring, hygiene and signal capture, then activation
- best when
- Leadership believes the data and signal layer is the binding constraint, and can defer visible wins.
Risk: Lowest execution risk, highest political risk, no visible wins for 60 to 90 days
not four end states. four routes to the same one, each optimised for a different binding constraint.
Why publish this
Because the alternative is asking you to trust a deck. Method is the only thing that distinguishes one agency from another once everyone has the same tools, and a method that cannot survive being read by a competitor was not much of a method.
If you are running this yourself, take it. If you want to know which rung you are on before deciding anything, the assessment takes three minutes, and the verification criteria set out the evidence behind each one.
Ready to build a GTM engine that compounds?
Schedule a free 20-minute call. We'll tell you exactly where AI fits in your GTM, and where it doesn't.
Join 250+ B2B leaders