Skip to content

How to Use Account-Based Marketing Segmentation From Your Own Deals

Build ABM segments from your won, lost and no-decision deals, with the questions to ask buyers, a scoring template and how the work changes with budget.

Written by Sabih AhmedPublished 16 min read

Account-based marketing segmentation usually starts with a filter on industry, headcount and region. A segment built that way can hold your best customers and accounts that stalled for a year, side by side.

Firmographics describe who could buy, not who has a reason to. Your CRM records who bought, who chose someone else and who never decided, but a stalled deal can end up filed as closed lost, next to deals another vendor won. This guide covers how to sort those deals, what to ask buyers and how to score segments from the pattern. We'll also show how the work changes from one marketer on $5k a month to a dedicated team.

How to segment accounts for ABM

To segment accounts for ABM, start from your own closed deals, not a firmographic filter. Sort the last 12 to 24 months into won, lost and no decision, find what the wins share that the other 2 groups don't, and turn that into criteria. Then score each segment on how often you win it and how often it stalls, and keep fit and timing on separate lists.

ABM means account-based marketing, where sales and marketing work a named list of accounts together, with plays built for each account or group instead of a broad audience. Segmentation is the part of your ABM strategy that decides which groups of accounts make that list and what each group gets. If it's wrong, every later play reaches accounts with no reason to buy.

Your own deals beat a market filter because they hold 3 different answers. Won deals show who actually buys from you, and they become your include rules. Lost deals show where another vendor wins, and the reasons you can't change become exclude rules. No-decision deals show accounts that looked right and never moved, and they tell you which of those accounts to avoid, which to flag and which to wait out.

This is the method we use, in 5 steps.

  1. Pull 12 to 24 months of closed deals and label each one won, lost or no decision.
  2. Call 3 to 5 buyers from each outcome, because a reason picked by the rep is the rep's view, not the buyer's.

Illustration: 2 lists drawn as cards of grey rows. The tall one on the left is the fit list, with 1 empty row marked by a black point. A thin line runs from that point up into the short list on the right, the core list. The row it reaches there is lime.

  1. Compare the 3 groups through SPICED (opens in a new tab), Winning by Design's framework of situation, pain, impact, critical event and decision, and write down what the wins share that the others don't.
  2. Draw segment lines where the wins differ in SPICED terms, such as their situation or who decides, and score each segment on fit.
  3. Put the top segments on the fit list, then pick the core list with the sales lead from the accounts showing a critical event.

The fit list holds every account that could buy, and the core list holds the ones showing a reason to buy now. Keeping them apart stops a good-fit segment from being treated as a ready one.

The method changes in 2 cases. With few deals, the calls carry more of the weight than the CRM. When you're entering a new market or launching a new product, your deals describe the old one. There, treat the criteria as a guess to test on the first 10 or so deals, not as a filter.

Why firmographic filters alone fall short

Firmographic filters fall short because industry, headcount, region and tech stack narrow the list, but none of them shows which accounts have a reason to change.

The common view is that firmographics plus intent data are enough to pick accounts. We start account selection from psychographic fit instead, which asks whether a company has the pain you solve and can put a cost on it. Whether it has a reason to fix it now is a separate question, and it decides the core list.

SPICED, from Winning by Design, is our lens because it reads what a filter can't see: the pain, its impact, the critical event and who decides. An account with a clear pain and a critical event is ready to change. A perfect firmographic match with neither goes quiet.

Segments named after an industry group accounts by what they are. Segments drawn on SPICED group them by why they buy, so one play can speak to the whole group.

Intent data doesn't close that gap. It shows research activity, not readiness, and readiness comes from pain and a critical event. So intent belongs on the timing side, as one more signal to check against what your buyers told you.

Firmographics still have a job, and it comes last. Once you know what the wins share, size and industry help you find more accounts that look like them.

  1. What we argue against: Firmographics plus intent data are enough to pick accounts

    Our view: Start from psychographic fit: whether a company has the pain you solve and can put a cost on it

  2. What we argue against: Segments named after an industry group accounts by what they are

    Our view: Segments drawn on SPICED group accounts by why they buy, so one play can speak to the whole group

  3. What we argue against: Intent data read as readiness

    Our view: Intent data read as research activity, one more signal on the timing side

  4. What we argue against: Firmographics pick the list

    Our view: Firmographics come last, to find more accounts that look like your wins

Our view, read through SPICED (opens in a new tab), Winning by Design's framework.

How to pull won, lost and no-decision deals from your CRM

Export every deal your CRM closed in the last 12 to 24 months, then label each one won, lost or no decision.

  1. Pick the window. 12 to 24 months keeps the deals close to today's product and price. If your pricing changed inside that window, start from the change.
  2. Export the closed deals with the account's industry, headcount, region and tech stack, plus deal size, cycle length, the last stage reached, the loss reason and the rep's notes. Leave open deals out, since a deal with no outcome yet blurs a slow win with a stall.
  3. Relabel the losses. A deal where the buyer chose nobody, went dark or pushed the project to next year is a no decision. Read the notes and last emails, not only the reason field.
  4. Count each group. Aim for at least 20 deals in each outcome, counted across your whole CRM before you split anything into segments.
  5. Make no decision its own loss reason, so the next export doesn't need relabeling. In HubSpot, the default deal properties (opens in a new tab) include a closed lost reason and a closed won reason, and the lost reason options can be customized.

Below 20 deals in an outcome, a pattern can come from one rep or one quarter. It goes to the calls as a question rather than into the criteria.

The relabeling step is the slowest, because it means reading notes and emails, and it's the last one to cut when time runs short. If no-decisions stay filed as losses, you blame the product for deals no vendor won, and you keep targeting accounts that rarely decide. With every deal labeled, the CRM tells you who to call.

How to run a win and loss analysis for ABM

A win and loss analysis for ABM means calling 3 to 5 buyers from each outcome and asking what made them move, choose someone else or stop. The calls add what the CRM can't show, which is each buyer's reason in their own words.

In each outcome, pick recent deals across the account types you expect to segment on, so every likely segment is heard from. Call the person who led the evaluation while the details are fresh. Someone other than the rep on the deal should make the call, because a buyer who turned a rep down has every reason to give that rep the polite answer.

Ask one question for each part of SPICED, and write the answers into one sheet with a row per part, so the 3 groups can be compared side by side.

  • For the situation, ask what was changing in the business when the search started.
  • For the pain, ask what was breaking and who felt it first.
  • For the impact, ask what leaving it alone would have cost in time, money or people.
  • For the critical event, ask what made that quarter the one to fix it, and what would have happened if the date slipped.
  • For the decision, ask who had to agree and who could have stopped it.

With a no-decision buyer, add one more question about what happened to the project after the last call. Listen for the part that went missing, whether a date that never came or an owner who left.

Treat 3 to 5 as a starting point, and stop when the answers start repeating, which can happen after 3 calls in one group and take 6 in another.

Once the sheet holds answers from all 3 groups, our buying moments skill can read it with the deals and sales notes. It turns them into the ideal customer, the buying group and the moments that make buyers start looking.

What no-decision deals tell you about accounts to avoid

No-decision deals tell you which accounts to wait on, which to avoid and which to flag. It depends on whether the deal lacked a deadline, a priced impact or an owner.

A missing critical event is a timing problem. The pain was real, nothing set a date, and the project slid to next year. That account can still be a good fit, so it stays on the fit list until a critical event shows.

A missing impact is a fit problem when it repeats. If nobody in a segment could say what the problem cost, the pain is too small there to win a budget fight, and the segment drops in the score.

A missing owner is a fit problem too, though sometimes one you can fix. Forrester's State of Business Buying, 2026 (opens in a new tab) found the typical buying decision now includes 13 internal stakeholders and 9 external influencers. A group that size needs one person who owns the problem and carries the rest to a yes, and some accounts don't have one.

So avoid a segment where the same blocker repeats for a reason you can't change, such as a pain nobody in it could price. When the blocker is a missing owner, such as a security review that no business team owns, flag those accounts and find an owner before they count as core. Fit and timing are separate questions, and good-fit accounts with no trigger have nothing to act on. So accounts that only lacked a date wait on the fit list and join the core list when an event shows.

How to turn closed deals into ICP criteria

Closed deals become ICP criteria when you compare the 3 groups part by part through SPICED and keep only what separates the wins from the other 2. The criteria are your ideal customer profile, written as rules a list can be built from.

Take a hypothetical case. A company selling onboarding software to HR teams pulls 18 months of closed deals and relabels the stalled ones. It sorts them into won, lost to another vendor and no decision, then reads each group through SPICED.

SPICEDWonLostNo decision
Situation150 to 800 employees, hiring 40 or more people a year, often after funding or a new officeUnder 100 employees, or running a payroll system the product doesn't connect toThe same size and industry as the wins
PainHR ops runs onboarding in spreadsheets, and new hires wait days for laptops and loginsOnboarding is light enough to run in emailThe same pain, described as annoying rather than urgent
ImpactThe head of HR could name what a slow first week costToo small to outweigh the priceNobody put a cost on it
Critical eventA new office, a hiring plan after funding, or a new head of HRNot what decided it, since a cheaper tool or a payroll link didNone, and "next year" came up again and again
DecisionAn HR ops leader owned the budget, with IT as a reviewerWent to a vendor with a native link to their payroll systemIT security ran the review, with no HR owner

On size and industry alone, the no-decision accounts sit right beside the wins. SPICED separates them on the impact, the critical event and the decision, which is where the criteria come from.

  • Include companies of 150 to 800 employees that plan to hire 40 or more people a year, where HR ops runs onboarding manually and can price a slow first week.
  • Exclude companies under 100 employees, and companies on the payroll system the product doesn't connect to, until it does.
  • Flag companies where IT security runs software reviews, and find an HR owner before they count as core.
  • Watch for a new office, a funding round with a hiring plan or a new head of HR, which are the timing signals for the core list.

Some of these rules can be checked from outside and some only on a first call. Hiring volume shows in job posts, and a new head of HR or a new office shows in the news and on LinkedIn. Whether HR ops runs onboarding manually and can price a slow first week gets confirmed on the first sales call. Until then, the account sits on the fit list as likely, not certain.

How to score and compare segments

Score every segment on the same 6 rows from your deals and calls, and compare them by the total. Use your overall rates as the baseline, so each segment is measured against your own pipeline rather than a benchmark. These bands are our starting scoring, not a standard. Adjust the cut-offs to your deal counts.

RowWhere it comes from012
Win rate, won ÷ (won + lost)CRMMore than 5 points below your overall rateWithin 5 points of itMore than 5 points above it
Stalls with a fit cause, as a share of closed dealsCRM notes and callsMore than 5 points above your overall shareWithin 5 points of itMore than 5 points below it
Pain with a priced impactCallsNot heardHeard, but priced by only some buyersHeard and priced by nearly every buyer called
Critical events you can seeCalls and public sourcesNone visible from outsideVisible for some accountsVisible and repeating
Owner for the problemCRM and callsNo one owned itAn owner, but the budget sat with another teamOne role owned the problem and the budget
Accounts availableYour fit listToo few to run a playEnough for sales outreachEnough for ads as well as outreach, with a core list inside (300 or more if you advertise on LinkedIn)

Win rate leaves no-decisions out on purpose, so it answers only whether you beat the alternatives. The stall row answers whether the accounts decide at all. It counts only stalls with a fit cause, no priced impact or no owner, because a stall that only lacked a date is a timing problem.

A missing owner costs points on this row and on the owner row on purpose. The stall row counts the deals it stalled, and the owner row asks whether the segment has an owner at all.

Because the score measures fit, the critical-event row only asks whether a segment's triggers can be seen from outside, not whether they've fired. A core list can't be built from triggers nobody can see.

In the hypothetical case, the accounts that meet the criteria split along 2 SPICED lines, who owns the decision and how fast they hire. Where IT security leads the review, the review shapes the deal more than hiring does, so those accounts form one segment. The HR-led accounts split at 100 hires a year, which gives 3 segments, scored like this.

RowHigh-volume hirers, HR ownerSteady hirers, HR ownerIT security leads the review
Win rate212
Stalls with a fit cause210
Pain with a priced impact211
Critical events you can see221
Owner for the problem221
Accounts available211
Total1286

Steady hirers win less often, and fewer of their buyers priced the problem. Companies where IT security leads the review win when they decide, but their stalls couldn't price the problem and had no owner. So high-volume hirers with an HR owner are the segment to run first, and the IT-led accounts stay flagged until someone finds an HR owner.

Inside one segment, fewer than 10 closed deals is too few for its rates to mean much, so it gets no score on the 2 rate rows. Below 20 deals, one deal moves a rate by 5 points or more, so treat a score near a band's edge as a tie. Score the call rows only for segments you've called, since a pain you never asked about can't score 0. Compare an untested segment on the rows it has, and check it against the next quarter's deals so one thin quarter doesn't decide it.

Score again every quarter with the new closes. A segment is working when its new deals close at or above the rate that put it on top. When its stall share climbs, call 2 or 3 of the stalled buyers before you change the criteria.

How account-based marketing segmentation changes with budget and team size

Account-based marketing segmentation changes with budget in how many segments you can run, what data sits on top of your deals and who does the work. The deal analysis underneath doesn't change, though who makes the calls does. A $5k program runs one segment well instead of 5 segments thinly.

1:many means existing programs, such as ads and email, aimed at the whole target list rather than tailored to each account. 1:few means one play built for a group of 5 to 15 accounts that share a trigger or a problem. 1:1 means plays and content built for a single account, kept for the few deals where a small lift in the win rate covers the cost.

The monthly budget is program spend on media, content, tools and mail, and the people who run it are on top. These tiers are our starting point, so adjust them to your deal size and sales cycle.

Monthly budgetSegments it can carryData on top of your dealsWho runs the deal review and callsLeave out for now
$5kOne tightly chosen segment, run as 1:manyNone bought, with accounts researched manually against your criteriaOne marketer, with an hour of the sales lead's time each weekIntent platforms and bought lists
$10kThe top segment as 1:many, plus 1:few plays for its best-fit groupFirmographic and technographic data to find more accounts like your wins2 marketers, or 1 with sales running the 1:few outreach, and sales choosing which buyers to callPlatform contracts
$20kTiered, with 1:few for the top segment and 1:many for the nextA paid intent data source on top of fit, read as research activity2 or 3 marketers with ops support, with ops running the export each quarter1:1 for more than a handful of accounts
$50k1:1 for a small top tier inside the top segmentAn ABM platform that joins fit, intent and site visitsA small team with an ops owner, and an outside interviewer for lost and no-decision calls1:1 beyond the top tier
$100k and upAll 3 tiers across several segments, with a dedicated ownerPlatform data, plus a standing win and loss programA dedicated ABM team with reps assigned to accounts, an owner, ops and contentSpreading 1:1 wider than sales can follow up

Below $5k a month, ABM isn't smaller, it's narrower. It runs on data you already own, your closed deals and your site visits, so the budget goes into acting on those signals rather than buying them.

At $5k, the one segment runs as 1:many, so it needs a list big enough to advertise to. LinkedIn's help page on company list targeting (opens in a new tab) says a list must have at least 300 rows to upload. Below that, the segment gets sales outreach instead of its own ad audience. From $10k a month, the groups inside it that share a trigger also get 1:few plays.

Budget doesn't change how many deals you have to learn from. A company spending $50k a month with 15 wins still has 15 wins. The calls carry the analysis there, and bought data only finds more accounts that match a guess.

Team size sets its own limit, because the calls and the relabeling take hours, not money. One marketer with a large budget should still start with one segment. An outside interviewer saves the calls, but someone still has to relabel the deals, read the answers and agree on the criteria with sales.

Where to start

Build segments from the deals you've already won, lost and watched stall, and keep timing on its own list. If you want the deal pull and the quarterly scoring to run in your CRM rather than a spreadsheet, our MarkOps service builds that workflow. Whichever route you take, start by exporting the last 12 to 24 months of closed deals. Relabel every stalled one as no decision before you read any pattern.

Sources

Every figure in this post links to where it comes from. These are the pages we checked.

  1. 01SPICED · winningbydesign.com
  2. 02Default deal properties · knowledge.hubspot.com
  3. 03Forrester's State of Business Buying, 2026 · forrester.com
  4. 04LinkedIn's help page on company list targeting · linkedin.com

About the author

Sabih Ahmed

Founder

Sabih founded SHAKEUP to run strategy, story and channels as one blend for B2B SaaS teams, with pipeline as the measure. He has sat on every side of the table, building campaigns, managing multi-million euro budgets and owning revenue targets. That is why his advice starts with the number a marketing leader answers for. Today he leads our strategy and messaging work, and wrote our B2B Google Ads playbook.

LinkedIn

Newsletter

One note a month on what's working.

One email a month from the work: what we tested, what moved pipeline and what we'd do differently.

  • One email a month, never more
  • Written by the people running the campaigns
  • Leave in one click

Sign-up opens with the first issue.

Prefer tools to reading? See our skills