What: Using Breeze Assistant to draw an honest map of what your Buyer Intent setup actually sees, what it does not see, and where the rest of your buyer journey is happening - then producing a Dark Funnel Operating Plan that combines HubSpot signals with the proxies you can read from elsewhere, gives sales a defensible Monday-morning rhythm, and gives the CFO a measurement framework that doesn’t pretend to attribute the unattributable.
Prompt of the week:Spend an hour on B2B marketing LinkedIn in mid-2026 and you will see the same claim made twenty different ways: somewhere between seventy and ninety-five per cent of the buyer journey now happens in places your attribution model cannot follow. AI search, peer Slack groups, anonymous review-site browsing, LinkedIn DMs, competitor comparison pages on G2, podcast mentions, the inside-the-buying-committee thread on email nobody outside the company will ever see. By the time a buyer fills out the contact form, the shortlist is already made, and you are either on it or you are not.
The numbers vary depending on who is selling what, but the direction of travel is unambiguous. Forty per cent of B2B marketing budgets now flow into buyer-intent signals of one kind or another. HubSpot Buyer Intent itself is a meaningful slice of that picture - reverse-IP visitor identification, third-party research signals across the partner network, company news, contact-level changes - and the April updates added more depth on top: new event, product-development and client-signing signals, plus unstructured engagement signals that pull Budget and Deal Timing mentions out of calls, emails and notes.
All of which is useful, and none of which solves the underlying problem. HubSpot Buyer Intent sees a specific, well-defined slice of the buyer journey. The rest happens elsewhere. Most teams respond in one of two unhelpful ways: they dismiss the tool because it cannot see everything, or they overweight it and act as if the dashboard is the funnel. Neither is right. The right move is a deliberate operating plan that names what HubSpot sees, names what it doesn’t, layers in the proxies that are actually readable from elsewhere, and gives the sales team a rhythm to work the resulting picture.
Underneath the noise, the community questions have been remarkably consistent. “We turned Buyer Intent on, the dashboard says 200 companies a month are visiting, and we have no idea which five will actually buy.” “Our CMO keeps asking how much of pipeline is influenced by AI search and I genuinely don’t know.” “We can see the company is researching, but not which seven people on the buying committee are doing it.” “How do I prove the marketing programme is working when the deal records all say the source was direct?” The honest answer is that you cannot answer those questions with a single tool. You answer them with a system.
This prompt builds the system.
Prompt structure
Paste this into Breeze Assistant and make sure CRM data access is enabled in your AI settings so Breeze can reference your Buyer Intent configuration, target market settings, signal history, pipeline records, and any connected enrichment or intent sources:
Role: You are a B2B Revenue Operations Strategist who specialises in
dark-funnel-aware go-to-market design. You understand the boundaries
of HubSpot Buyer Intent precisely, you know which other signals are
readable and which are genuinely invisible, and you can design an
operating model that respects both.
Task: Produce a Dark Funnel Operating Plan that (a) maps what our
current Buyer Intent setup sees and does not see, (b) layers in the
complementary signals we can practically read from elsewhere, (c)
specifies the Monday-morning rhythm that turns this picture into
outbound and ABM action, and (d) gives leadership a measurement
framework that is honest about what is attributable and what is
influence-only.
Context:
- Company: [COMPANY NAME]
- Industry and sub-vertical: [INDUSTRY / SUB-VERTICAL]
- HubSpot tier: [Pro/Enterprise; note any add-ons such as G2
Buyer Intent or Breeze Intelligence enrichment]
- Buyer Intent status:
[NOT ENABLED / ENABLED, BASIC CONFIG / ENABLED, FULLY TUNED]
- Average deal value: [NUMBER]
- Typical buying committee size: [NUMBER of stakeholders]
- Typical sales cycle: [WEEKS or MONTHS]
- Geographic mix: [e.g., "60% UK/EU, 30% US, 10% APAC"]
(matters because IP-to-company resolution varies by region)
- Connected third-party intent or enrichment sources:
[e.g., "G2 Buyer Intent add-on, Bombora, ZoomInfo, none"]
- Channels where you suspect dark-funnel research happens:
[e.g., "ChatGPT, Perplexity, LinkedIn, niche Slack communities,
G2/Capterra, industry podcasts, founder newsletters"]
- Current attribution model:
[FIRST-TOUCH / LAST-TOUCH / MULTI-TOUCH / NONE]
- Top three competitors by displacement frequency: [LIST]
Audit and design the following:
1. SIGNAL COVERAGE MAP
Produce an honest inventory of what is visible to us today and
what is not. Categorise each known buyer touchpoint into one of
four buckets:
- FULLY VISIBLE: form fills, identified contact activity, deal
and email engagement, support interactions
- PARTIALLY VISIBLE: HubSpot Buyer Intent (anonymous visitor
identification via reverse-IP and partner research network);
third-party tools we already pay for; review-site behaviour
where integrations exist
- PROXY-READABLE: not directly observable, but inferable from
proxies - AI referral traffic share as a signal of
ChatGPT/Perplexity citation; LinkedIn follower spikes from
target accounts; podcast download geography; branded vs
non-branded search volume trends; community-mention monitoring
- GENUINELY INVISIBLE: peer DMs, private Slack threads, internal
procurement conversations, conversations on competitor sales
calls
For each PARTIALLY VISIBLE and PROXY-READABLE category, name the
specific signal source we should be pulling from and the tool
or method to do it.
2. MULTI-SIGNAL SCORING RUBRIC
Combine the readable signals into a unified scoring model. For
each signal type, recommend:
- Weight (0 to 25) and reasoning
- Decay curve (how quickly the signal goes stale)
- Whether it stacks (multiple instances increase weight) or
saturates (one is enough)
- Whether it triggers an action at the account level, the
contact level, or both
At minimum cover: HubSpot Buyer Intent visitor signals, research
intent on tracked topics, company news signals (the new event,
product-development and client-signing types included),
unstructured engagement signals (Budget and Deal Timing
mentions), G2 or other third-party intent if connected, AI
referral arrival behaviour, branded-search volume changes, and
LinkedIn engagement spikes from named target accounts.
Output a single SCORING TABLE and a worked example for one
hypothetical account scored against the model.
3. DARK-FUNNEL DISCOVERABILITY PROGRAMME
If most of the journey happens off-platform, the only thing we
can do is be
findable, citable, and credible in those off-platform places.
Recommend a content and presence programme covering:
- AI search citability: which content formats and structures
get cited in ChatGPT, Perplexity and Gemini answers, and how
to instrument it (this connects directly to HubSpot AEO if
in use, and to AI referral traffic share as the proxy
measurement)
- Review site presence: where we should hold a current G2 or
equivalent profile, how recent the reviews need to be, and
what categories we should be listed under
- Peer-network presence: which Slack communities, LinkedIn
creators, podcasts, or newsletters are credible voices in
our category, and a feasible cadence of contribution
- Founder or executive content: how visible our leadership is
in the channels our buyers actually read
Keep recommendations realistic for the team size implied by
the Context. No “start a podcast and a Substack” plans for a
two-person marketing team.
4. SALES OPERATING RHYTHM
Translate the scoring model into a working week. Specify:
- MONDAY: the SDR’s first-task-of-the-week ritual. What lists
do they open, which signals are they prioritising, what
research depth before outreach
- DAILY: the trigger-based moments - a score crossing a
threshold, a Budget signal firing, a competitor signal on
a target account
- WEEKLY: the marketing and sales joint review - which
accounts moved into and out of in-market scoring, which
signals are producing meetings, which are noise
- MONTHLY: the calibration review - adjust weights, retire
signal types that aren’t earning their weight, add ones
that have proven themselves
For each cadence, define who owns it and what the success
criterion is.
5. HONEST MEASUREMENT FRAMEWORK
Design a measurement model that treats attributable and
influence-only contributions separately rather than pretending
everything can be cleanly sourced. Specify:
- ATTRIBUTABLE LAYER: pipeline and revenue with a clear
first-party touchpoint trail
- INFLUENCE LAYER: closed-won deals showing dark-funnel signal
activity in the 90 days preceding conversion (intent
research, AI referral arrival, review-site behaviour, target
account engagement spikes) even when the deal record
attributes the source elsewhere
- LEADING-INDICATOR LAYER: AI referral traffic share over time,
branded-search lift, target account in-market percentage,
review-site share-of-voice - the early signs that the
dark-funnel programme is or is not working
- HONESTY LINE: an explicit statement of what we cannot
measure at all, accompanied by the qualitative methods
(win/loss interviews, post-deal stakeholder surveys) that
substitute for it
Recommend the specific dashboards and views to build in
HubSpot for each layer.
6. ROLE AND HEADCOUNT IMPLICATIONS
Be honest about what running this programme actually requires.
Flag:
- Skills the team has versus skills it needs
- Whether the existing martech stack can support the plan,
or whether one or two additions are needed (and which
ones can be skipped)
- Whether the sales team is structurally ready (named-account
SDRs versus inbound-only, dedicated ABM ownership, marketing
and sales joint accountability)
Constraints:
- Do NOT recommend adding more than two net-new tools. The
failure mode of dark-funnel thinking is buying ten point
solutions and calling that a strategy
- Every recommendation must be implementable inside HubSpot
natively or with an integration that already exists in the
HubSpot ecosystem unless explicitly flagged as a build
- Distinguish clearly between signals that can fire a workflow
and signals that can only inform human judgement. Do not
pretend the second is the first
- For any signal source where coverage drops below 50% for our
geographic mix (e.g., reverse-IP in mobile-heavy APAC
markets), state it explicitly and adjust the rubric
- If the data needed to calibrate weights, decay curves, or
baseline rates is not visible from the current context,
state: "SIGNAL MISSING: [what needs checking manually]"
- Avoid attribution claims that the data cannot support.
Influence-layer reporting must be labelled as such
Output format:
### I. EXECUTIVE SUMMARY
{3-sentence overview: where the current Buyer Intent setup is
strong, where it is blind, and the single highest-leverage
addition we should make this quarter}
### II. SIGNAL COVERAGE MAP
| Touchpoint | Visibility | Source | Tool/Method | Coverage % |
### III. MULTI-SIGNAL SCORING TABLE
| Signal | Weight | Decay | Stacks? | Account/Contact | Action |
{Plus one worked example: hypothetical account scored across the
model with explanation of each contributing signal}
### IV. DARK-FUNNEL DISCOVERABILITY PROGRAMME
{AI citability, review-site presence, peer-network presence,
executive content - with realistic cadences for our team size}
### V. SALES OPERATING RHYTHM
| Cadence | Owner | Ritual | Success Criterion |
### VI. MEASUREMENT FRAMEWORK
{Attributable, Influence, Leading-Indicator and Honesty layers,
with specific HubSpot dashboards to build for each}
### VII. STACK AND HEADCOUNT GAPS
| Gap | Risk if Unaddressed | Proposed Action | Priority |
### VIII. 30 / 60 / 90 DAY ROLLOUT PLAN
{Sequenced actions with owners across RevOps, Demand Gen,
Content, Sales Leadership, and Executive Sponsor roles}
Why this prompt works - and how to adapt it
Most dark-funnel writing online sits at one of two extremes. The first is a strategy essay that names the problem with great eloquence and provides no implementable action.
The second is a vendor pitch that names the problem and proposes that the answer is, conveniently, the vendor’s product. This prompt does neither. It forces a structured map of where you actually stand today, a rubric for combining the signals you can read, an honest list of the ones you cannot, and a working rhythm for the team that has to act on the picture.
A few things to note about how it is constructed:
Honesty about visibility is the foundation. The four-bucket coverage map (Fully Visible, Partially Visible, Proxy-Readable, Genuinely Invisible) does the unglamorous work most plans skip: it names what your tools can actually see and refuses to pretend otherwise. Once that is on the page, the rest of the plan stops being theoretical - it is grounded in what you can do something about, not what you wish you could.
Proxies are doing the heavy lifting. AI referral traffic share is not a perfect measure of ChatGPT citation, but it is a defensible one. LinkedIn follower-spike from a target account is not a confirmed buying signal, but it correlates well enough to be worth weighting. Branded vs non-branded search volume is not attribution, but it tells you whether your brand is moving in the dark funnel or stagnating in it. The plan leans into proxy reading because proxy reading is what most of the actually-useful dark-funnel work looks like in practice.
The two-tool ceiling is a forcing function. The failure mode of intent-data thinking is buying every point solution on the market and calling that coverage. The cap forces a hierarchy - what is the single highest-leverage addition this quarter, and what is the second - instead of producing a wish list that nobody ever buys. It also keeps the plan executable inside an SMB or mid-market budget rather than only making sense for enterprise.
The influence layer is where the honest measurement lives. The hardest conversation in B2B marketing right now is the one with the CFO about which spend produced which revenue. The framework deliberately separates attributable revenue from influence-only revenue and from leading-indicator movement, because conflating them is how marketing budgets get cut at the next downturn. A piece of pipeline that closed direct but had three dark-funnel signals in the 90 days before is influence, not attribution - and labelling it correctly is what keeps the marketing team’s credibility intact when the next finance review lands.
The Monday-morning rhythm is what makes the plan survive contact with reality. A scoring model nobody opens is a spreadsheet. A scoring model the SDR consults at 9am on a Monday before they touch their first sequence is operational. The cadence specification - Monday ritual, daily triggers, weekly review, monthly calibration - is the bit that turns the document from a strategy memo into a working system.
“SIGNAL MISSING” keeps the model honest. Breeze can see your HubSpot Buyer Intent configuration, your target market criteria, and your historical signal data. It cannot see the conversations your reps are having in their heads about which accounts are real, the conversations your CFO is having about budget pressure, or the bits of the martech stack you are not yet logged into. Where the agent runs out of visible evidence to anchor a recommendation, the flag forces it to admit the gap so a human can fill it \u2014 which is the alternative to fluent guesswork dressed up as analysis.
Adapting it for your portal:
Enterprise sales motion? If your typical deal value is in six or seven figures and your buying committees run to seven-plus stakeholders, weight the rubric heavily towards account-level signals and add this line: “Treat individual contact signals as supporting evidence only. Score and route at the account level, and require multi-stakeholder engagement before promoting an account to in-market status.” The output shifts towards an ABM-shaped operating rhythm.
Heavy reliance on AI search traffic? If you have started seeing meaningful inbound traffic from ChatGPT, Perplexity or Gemini, add: “AI referral share is rising and we want a measurement and content programme that compounds it. Recommend the specific HubSpot AEO setup and the content cadence that maximises citation likelihood.” The discoverability programme will lean into AEO specifics rather than generic content advice.
APAC-heavy mix? Reverse-IP resolution drops sharply in mobile-first markets and where corporate VPN use is heavy. Add: “Our addressable market is significantly APAC. Adjust the coverage map and rubric to reflect lower IP-to-company resolution in this geography and propose the proxies that compensate.” Expect more weight on form-based behaviour, social engagement, and direct response channels.
Already running ABM tooling? If you have 6sense, Demandbase, or similar in place, name them in Context and add: “We already operate a third-party intent platform. Recommend the cleanest division of labour between that platform and HubSpot Buyer Intent so we are not paying twice for overlapping signal.” The plan will produce a clearer hand-off between the systems rather than a redundant stack.
Small team? If you have a marketing team of three or fewer, add: “Headcount is constrained. Prioritise the plan around what one or two people can realistically operate, and explicitly flag anything that requires more capacity than we have so we know what we are deferring.” The output becomes a realistic minimum-viable dark-funnel programme rather than an idealised enterprise version.
Want a quarterly cadence? Save the output and re-run 90 days later with: “Compare against the output from [DATE] and report on which signal weights have proven accurate, which dark-funnel proxies have started producing readable trends, which dashboards are being used by leadership, and which parts of the operating rhythm have stuck versus quietly lapsed.” That turns the plan into a rolling discipline rather than a one-off document.
Beyond the prompt:The Dark Funnel Operating Plan is a starting point, not a finish line. The sequence below is the order that has the best chance of producing a visible win inside one quarter without breaking anything that already works.
Start with the coverage map. Walk the map round the marketing, sales and RevOps leads in one workshop. The conversation alone is usually worth more than any individual signal you will end up tracking, because it forces a shared mental model of where the buyer actually lives. The bickering during the workshop is the work - it is where the team’s implicit assumptions about pipeline get said out loud and corrected.
Then ship the scoring model inside HubSpot as a single scoring property on the company object. Resist the urge to build five overlapping scores. One score, with the contributing signals visible as supporting properties beneath it, will be argued about and trusted; five competing scores will be ignored. Run it in shadow mode for a fortnight against current pipeline - not driving any automation - to sense-check the weights against deals you already know are real.
Once the score is trusted, wire the Monday-morning rhythm into the SDR workflow. A saved view of in-market accounts that opens automatically. A daily digest that surfaces score crossings and Budget mentions. A weekly fifteen-minute marketing-and-sales review of what crossed the threshold and what closed. The rhythm is the bit that makes the system survive the executive sponsor changing their priorities, which they will.
Then, and only then, look at the discoverability programme. The temptation is to start here, because content is fun and signal architecture is not. Resist. A team that has not yet sharpened how it reads signals will produce content with no measurable hook back to pipeline, and the programme will be cut at the next budget cycle. A team that already operates a scored signal model can produce content with leading indicators baked in from the start - AI referral traffic share, branded-search lift, review-site share-of-voice - and defend it when the question “is this working” arrives.
Throughout, hold the honesty line. Influence is influence. Attribution is attribution. There is a slice of revenue you genuinely will not be able to source, and the right move is to label it honestly rather than create a misleadingly precise number. The marketing leader who walks into a board meeting with “thirty per cent of pipeline is attributable, forty per cent is influence-only with these signals behind it, and we are not pretending to source the remaining thirty” survives finance scrutiny in a way the one with a single confident pie chart does not.
The dark funnel is not a measurement problem to be solved with another pixel. It is the new shape of how buyers actually buy. The teams that win in the next twelve months are not the ones with the most signals in their dashboards - they are the ones with the clearest read on what those signals do and do not mean, and the operating rhythm to act on them anyway.
