Build your target-account list from evidence, not opinion
Derive, score, and tier your ABM accounts from real deal data
This one is for the ABM leads, the growth marketers, and anyone who has built a target-account list from a gut feeling and a LinkedIn scroll, and then watched the sales team quietly ignore every name on it.
What: Using Breeze Assistant to build your ABM target-account list from your own revenue data rather than opinion: mining your closed-won deals for the pattern that actually predicts a win, your losses and churn for the accounts to avoid, and turning both into a queryable definition of a perfect-fit account that Breeze can score every company against and sort into disciplined tiers, mapped straight onto HubSpot's native ABM properties. It is the one decision the whole programme is built on, made with evidence instead of a hunch.
Prompt of the week:
Ask most teams how they chose their target accounts and the honest answer is some blend of gut feeling, a few logos the chief executive admires, and an afternoon on LinkedIn. It feels like strategy, and it is really just a list of companies that look impressive on paper. The trouble shows up downstream: the accounts do not convert any better than the rest, the sales team notices, and within a quarter the carefully built target list is a spreadsheet nobody opens. Every serious piece written on ABM in 2026 lands on the same uncomfortable point, that the whole programme lives or dies on this one decision, and most teams make it on a hunch.
The fix is not a better hunch. It is to stop guessing and let your own history choose. Your closed-won deals are a precise record of exactly which kinds of company actually buy, buy quickly, and stay. Your losses and your churn are an equally honest record of who to avoid. HubSpot's own framing is blunt about the payoff: a strong, evidence-based ideal customer profile is associated with materially higher win rates, and the teams that filter aggressively to perfect-fit accounts consistently outperform the ones chasing an impressive-looking long list. The raw material for a great target list is already sitting in your CRM.
There are two things most teams miss even when they do look at the data. The first is that an ideal customer profile written as a paragraph is useless. “Mid-sized B2B companies in growth mode” cannot be acted on. It has to be a queryable filter, a specific set of firmographic and behavioural criteria you can actually run against your database. The second is the anti-signal: knowing who tends to churn or never close is as valuable as knowing who buys, because it stops you pouring expensive effort into accounts that were always going to leave. A good target list is defined as much by what it excludes as by what it includes.
So this week's prompt turns your revenue history into your target-account list. You point Breeze at your closed-won and closed-lost deals, and it derives the pattern that predicts a win, defines the anti-signal that predicts a loss, expresses both as one concrete queryable filter, scores your existing companies against it, and proposes a disciplined three-tier split with a hard cap on how many accounts earn the expensive top-tier treatment. It writes the result onto HubSpot's own ABM properties, so the list is not a document, it is a working, filterable part of your CRM.
Prompt structure
Paste this into Breeze Assistant and make sure CRM data access is enabled in your AI settings so Breeze can reference your companies, deals, closed-won and closed-lost history, and the properties behind them:
Role: You are a HubSpot ABM strategist who builds target-account
lists from revenue data, not opinion. You know the whole programme
rests on selecting the right accounts, that an ideal customer profile
has to be a queryable filter rather than a paragraph, and that knowing
who churns or never closes matters as much as knowing who buys. You
map everything onto HubSpot's native ABM properties.
Task: Build our target-account list from our own history. Analyse our
closed-won deals for the pattern that predicts a win, our closed-lost
and churned accounts for the anti-signal, express the ideal customer
profile as one concrete queryable filter, score our existing companies
against it, and propose a disciplined three-tier split. Map the result
onto HubSpot's Target Account and ICP Tier properties, and give me a
review cadence.
Context:
- Company: [COMPANY NAME]
- Industry: [INDUSTRY]
- HubSpot tier: [note Sales or Marketing Hub Professional or above,
since the ABM properties, Target Accounts dashboard and account
scoring depend on it]
- What we sell, and typical deal size and sales cycle:
[so value and speed can weight the analysis]
- How many accounts our sales team can realistically work at once:
[so the top tier is capped to reality]
- Data available: [closed-won deals, closed-lost reasons, churn or
renewal data, firmographics, enrichment]
- The ICP we think we have today: [so it can be tested against the
data, or "none agreed"]
- Known good and bad fits: [logos that were great customers, and
ones that churned or never closed]
Design the following:
1. THE WIN PATTERN (from closed-won)
- Analyse our closed-won deals for the traits they share:
industry, company size, revenue band, region, tech stack, the
personas who drove the deal, and any signals that preceded
the win
- Weight by deal quality, not just count: prioritise the accounts
that closed quickly, were large, and stayed (highest lifetime
value, lowest churn)
- State which traits genuinely predict a win, versus which are
simply common across all our deals
2. THE ANTI-SIGNAL (from losses and churn)
- Analyse closed-lost and churned accounts for the traits that
predict a poor fit: the industries, sizes or profiles that
stall, discount heavily, or leave
- Turn these into explicit exclusion criteria, so we stop
targeting accounts that were never going to work
3. THE ICP AS A QUERYABLE FILTER
- Express the ideal customer profile as a specific, runnable
filter, not a paragraph: firmographic criteria plus behavioural
criteria, each concrete enough to filter our database on
- Shape it like this: [industry set] and [employee range] and
[revenue band] and [region] and [a tech or trigger signal],
EXCLUDING [the anti-signal traits]
- Note which criteria are firmographic (from enrichment) and
which are behavioural (from engagement)
4. FIT SCORING & TIERING
- Score our existing companies against the filter, combining a
Fit Score (how well they match the ICP) with an Engagement
Score (how much the account is already engaging)
- Propose a three-tier split: Tier 1 (best fit, highest value,
for one-to-one treatment), Tier 2 (strong fit, scalable
personalisation), Tier 3 (emerging fit, automated nurture)
- Cap Tier 1 hard, to roughly what the sales team can actually
work one-to-one. Fewer, better accounts beat a long list
5. MAP TO HUBSPOT
- Which companies to flag as Target Account = true
- How to set the Ideal Customer Profile Tier property for each
- The saved company views or lists to build per tier, and the Fit
and Engagement score properties to create
6. REVIEW CADENCE
- How often to revisit the list, and the trigger to re-tier an
account as it gains or loses fit or engagement
- The signal that the ICP filter itself needs updating, such as
new won-deal patterns the current filter would miss
Constraints:
- Derive the ICP from our actual closed-won and closed-lost data, not
from generic B2B assumptions or an ideal we wish were true
- Express the ICP as a queryable filter that can be run against our
database, never as a vague paragraph
- Treat the anti-signal as first-class: define who to exclude as
clearly as who to include
- Cap Tier 1 to what sales can realistically work one-to-one. A
shorter, sharper list is the point, not a failing
- Map everything onto HubSpot's native ABM properties (Target
Account, ICP Tier), so the output is a working part of the CRM,
not a separate document
- Do not invent firmographics or deal history. If a figure or pattern
you need is not visible from the current context, state:
"SIGNAL MISSING: [what needs checking manually]"
Output format:
### I. TARGET LIST SUMMARY
{3-sentence overview: the win pattern in one line, the single
strongest anti-signal, and how many accounts land in each tier}
### II. THE WIN PATTERN
| Trait | Seen in Won Deals | Predictive or Just Common? |
### III. THE ANTI-SIGNAL (exclusion criteria)
| Trait | Seen in Losses or Churn | Exclude? |
### IV. THE ICP FILTER
{The queryable filter written out: firmographic and behavioural
criteria, plus the exclusions}
### V. TIERED ACCOUNT LIST
| Company | Fit Score | Engagement Score | Tier | Why |
### VI. HUBSPOT SETUP & CADENCE
{Which properties to set, the lists and views to build per tier, and
the review rhythm}
Why this prompt works, and how to adapt it
Almost every ABM framework starts one step too late, at what to do with your target accounts, and skips the harder question of how you chose them in the first place. This prompt lives at that earlier, more decisive step, because a brilliant campaign aimed at the wrong accounts is just expensive noise. It replaces the hunch that usually drives account selection with the one source of truth that cannot flatter you: your own record of who has actually bought, stayed, and left.
A few things to note about how it is constructed:
It lets the data pick, not the loudest voice in the room. The win pattern is derived from your closed-won deals, which quietly removes the single most common failure in account selection: the impressive logo somebody wants to chase for reasons that have nothing to do with whether it will ever buy. Evidence is harder to argue with than enthusiasm, and it points at accounts that genuinely resemble the ones that already worked.
The anti-signal is half the work. Most target lists are built entirely from who to include, and never from who to exclude, which is why teams keep pouring effort into profiles that reliably churn or stall. By mining losses and churn for exclusion criteria and treating them as first-class, the prompt stops you spending your most expensive attention on accounts your own history already warned you about.
An ICP has to be queryable, or it does nothing. A profile written as a paragraph feels like progress and changes nothing, because you cannot run “ambitious mid-market companies” against a database. The prompt insists the ideal customer profile comes out as a concrete filter, firmographic and behavioural criteria you can actually apply, which is the difference between a strategy slide and a list you can act on tomorrow.
It weights by deal quality, not just count. A trait that is common across a pile of small, discounted, quick-to-churn deals is not your ideal customer, it is just noise dressed as a pattern. The prompt weights the analysis towards the accounts that closed fast, were large, and stayed, so the profile is built from your best revenue rather than simply your most frequent.
Capping Tier 1 is the discipline that makes ABM work. The entire economic case for ABM is spending disproportionate effort on a small number of accounts, so a top tier that quietly swells to fifty names is ABM in name only. The prompt caps Tier 1 to what your sales team can genuinely treat one-to-one, because the constraint is not a limitation, it is the whole point of the approach.
“SIGNAL MISSING” stops a confidently wrong list. A target list built on invented firmographics or half-remembered deal history does not just contain a few errors, it aims the entire programme at the wrong accounts. Breeze can read your deals and your companies, but it cannot always see a churn reason that was never logged, or a firmographic that enrichment never filled in, so where the analysis would rest on a guess, the flag marks the gap for a human instead of manufacturing a pattern that was never really there.
Adapting it for your portal:
Thin on closed-won history? If you are early-stage or light on deals, add: “We have fewer than [N] closed-won deals. Lean more on the strongest anti-signal from our losses and on firmographic fit, be explicit about how much confidence the small sample allows, and tell me what to re-run once we have more wins.” The analysis will be honest about the size of the sample rather than overclaiming a pattern.
Expansion and retention matter as much as new logos? If you grow through your base, add: “Weight the win pattern towards accounts that expanded and renewed, not just closed. Treat high lifetime value and low churn as the strongest ICP signal, and flag the profiles that close but then leave.” The profile will favour the customers who stay, not just the ones who sign.
Selling into several segments? If you serve distinct markets, add: “We sell into distinct segments. Derive a separate ICP filter and tiering for each rather than one blended profile, so we do not average two different ideal customers into a vague middle.” You get sharp per-segment lists instead of one muddy compromise.
Using Breeze Intelligence enrichment? If you enrich your data, add: “We enrich with Breeze Intelligence. Use the enriched firmographics in the filter, and tell me which criteria depend on enrichment so I know where coverage gaps would weaken the scoring.” The filter will make the most of your enriched fields and flag where they are thin.
Sales and marketing working different lists? If the two teams disagree on the accounts, add: “Sales and marketing keep working different account lists. Build the tiering so both teams share one definition, and show me where the data agrees or disagrees with the accounts each team currently favours.” The output becomes a shared, evidence-based settlement rather than two competing opinions.
Want a quarterly cadence? Save the output and re-run it 90 days later with: “Compare against the output from [DATE]. Tell me which accounts changed tier, which new closed-won deals the current ICP filter would have missed, and whether the anti-signal still holds.” That keeps the list matched to a market and a customer base that keep moving.
Beyond the prompt:
The list Breeze produces is the foundation the whole programme stands on. Turning it into results is about applying it with discipline and keeping it honest.
Apply the tiers in HubSpot before you build a single campaign. Set Target Account and ICP Tier on the companies, build the saved views per tier, and create the Fit and Engagement score properties. Until the list lives in the CRM as filterable data, it is just a spreadsheet, and it will drift out of date the moment you look away.
Resist the urge to widen Tier 1. The instinct, once the list is built, is to sneak a few more exciting logos into the top tier because they would be nice to win. That is precisely how ABM quietly turns back into ordinary demand generation. Hold the cap: Tier 1 should never be larger than the number of accounts your team can genuinely treat one-to-one.
Match the effort to the tier, and mean it. Tier 1 earns real one-to-one work, Tier 2 gets scalable personalisation, Tier 3 runs on automated nurture. The whole economic case for ABM rests on spending the expensive effort only where the return justifies it, so a Tier 3 account receiving Tier 1 attention is money quietly wasted.
Re-tier when reality changes, not once a year. Accounts gain and lose fit and engagement constantly: a Tier 3 account that starts engaging hard has earned a promotion, and a Tier 1 account that has gone silent for two quarters probably has not. A quarterly review that actually moves accounts between tiers keeps the programme pointed at the accounts that matter now, rather than the ones that mattered when you first drew up the list.
Everything in ABM, the personalisation, the campaigns, the sales effort, the reporting, is downstream of one decision: which accounts you chose. Get that right, from evidence rather than instinct, and the rest of the programme has a chance to work. Get it wrong, and you are simply doing ordinary marketing to a list of companies that happened to look impressive on paper.