Blog · abm
What proves ABM is working at €25k+? Not win rate in year one
Prove ABM by comparing target accounts with a pre-launch baseline or holdout. At €25k+, win rate needs years of deals; judge year one on opportunity rate.
closed deals, won or lost, in each group to tell a 20% win rate from 30% (Illustrative)
Panelhop, 2026for a team closing 120 deals a year to read a 10-point win-rate lift (Illustrative)
Panelhop, 2026average B2B buying cycle reported by buyers with a median purchase of $200–300k
6sense, 2025Which ABM metrics prove impact, and which describe activity?
A metric proves ABM only when it compares target accounts with what would have happened without the programme. That means their own baseline from before launch, or a holdout of similar accounts the programme leaves alone. Everything else describes activity.
Most metric lists mix 3 kinds of measure:
- Activity: impressions, clicks, MQLs, engagement scores and influenced pipeline show the programme is running.
- Progress: buying-group coverage, account progression and the share of target accounts that open an opportunity show accounts moving.
- Outcomes: pipeline, win rate and revenue from target accounts show the money.
Influenced pipeline sits with activity because any touch counts. The 6sense guide that ChatGPT cites defines it as “whether any of your ABM channel strategies influenced your target accounts”.↗
Progress and outcomes turn into proof once they’re set against a baseline or a holdout. The catch is volume. Each group needs enough events to tell a real change from chance, and at €25k+ the outcome events are rare.
Today’s AI answers stop short of that. All 5 answers we captured on 5 October 2026 set target accounts against other accounts or a baseline somewhere.↗ ChatGPT and Perplexity call for a control or holdout group, and 4 of the 5 name win rate as proof.↗ In the text we captured, none says how many accounts or deals the comparison needs.↗
Win rate can’t prove ABM in year one at €25k+ deal volumes
Win rate needs hundreds of closed deals in each group before a 10-point lift shows, and an ABM list at €25k+ rarely produces that many in a year.↗ By our Illustrative maths, at the usual test standard (two-sided, 5% significance, 80% power), telling a 20% win rate from 30% takes 294 closed deals, won or lost, in each group: about 590 in all.↗ A team closing 120 deals a year, split evenly between target and holdout accounts, adds 60 to each group a year.↗ It needs about 5 years.↗ Even a doubling, from 20% to 40%, needs 82 closed deals in each group, or 16 months of that team’s deals.↗
A 10-point win-rate lift takes 294 closed deals in each group to read: about 5 years at 120 deals a year.
Data behind this chart
| Item | Value |
|---|---|
| Closed in a year | 60 deals |
| Needed for 20% → 40% | 82 deals |
| Needed for 20% → 30% | 294 deals |
Timing makes it worse. Buyers in 6sense’s 2025 survey, with a median purchase of $200–300k, reported buying cycles of 10.1 months on average.↗ So a buying journey the programme starts in March is, on average, decided the following January. Most deals that close in year one began before the programme did, and they dilute whatever it changed.
Win rate still matters: it’s the outcome the programme exists to move. Report it from year two, with the deal count beside it, and treat a small lift as unproven until the count is there.
A holdout needs enough events, not a 20–30% share
Size a holdout by the events it will produce in a year. A share of the list or a round number of accounts tells you nothing about that, and the advice we found skips the step. An ABM provider’s guide published on 2 October 2026 suggests a control group of 20–30 accounts with similar firmographic profiles, tracked on win rate and time to close.↗ Its year-one target includes a 15–20% improvement in win rate.↗ An ad platform’s play keeps at least 20–30% of accounts in the holdout and runs for at least 21 days “for statistically significant results”.↗
Run the numbers. If 10% of accounts open an opportunity in a year, 25 control accounts produce 2 or 3 opportunities.↗ Their win rate rests on 2 or 3 deals.↗ Read as a relative lift, a 20% improvement on a 20% win rate means 24%.↗ Telling 24% from 20% takes about 1,700 closed deals in each group.↗ A 21-day test may be long enough for ad engagement.↗ For pipeline it isn’t: at the same 10% yearly rate, a whole 300-account list opens about 2 opportunities in 21 days.↗
Measures where every account counts read sooner. Take the share of target accounts that open an opportunity: a rise from 10% to 20% in a year can be read with about 200 target and 200 holdout accounts.↗ Buying-group coverage reads sooner still, because it’s measured in every account at any time and the programme acts on it directly.
| Smallest change you can read in a year (Illustrative) | 100 target + 100 holdout accounts | 300 + 300 | 1,000 + 1,000 |
|---|---|---|---|
| Win rate (20% before) | None: about 10 closed deals a group | 20% → 54% | 20% → 38% |
| Accounts opening an opportunity (10% before) | 10% → 25% | 10% → 18% | 10% → 14% |
| Accounts with 3+ buying-group roles engaged (20% before) | 20% → 38% | 20% → 30% | 20% → 25% |
The table assumes a holdout as big as the target list and 10% of accounts opening an opportunity a year.↗ Win rate and the share of accounts with 3 or more buying-group roles engaged both start at 20%, at the same test standard.↗ Below about 200 accounts a group, a holdout catches only very large changes.↗ Lean on the baseline instead, and say in the report that the comparison is before and after, without a control group.
Record the baseline before launch, from the CRM
Take the baseline from the CRM in the weeks before launch, for the target list and any holdout alike. Record each number with its date, definition, data source and the count behind it, then freeze the sheet. Look back at least 12 months, because at €25k+ a handful of deals can swing a quarter.
Draw the holdout at random from accounts that pass the same fit rules, so both groups start alike. A holdout made of the accounts nobody wanted would flatter the programme.
- The target list and the holdout, frozen, with the rule that picked each account
- Opportunities opened per account over the last 12 months, by quarter
- Qualified pipeline created in target accounts, per quarter
- Win rate on closed deals, with the number of deals behind it
- Buying-group coverage: engaged roles against required roles, per account
- Account progression: the share of accounts that moved at least one stage
- Median days from opportunity to close
- Written definitions of an opportunity, an engaged contact and each stage
If a definition changes mid-year, restate the baseline under the new definition instead of comparing across the change. Name who may change a definition, too. A baseline that moves with the results can’t prove anything.
MQLs, clicks and influenced pipeline can’t prove ABM at €25k+
We think none of them can, because each rises with activity whether or not target accounts changed what they do. More ads bring more impressions and clicks. More campaigns bring more MQLs and more influenced pipeline.
At this deal size the buying group decides together. So we don’t read one form fill as a sign that the account will buy. We also think an MQL count pulls marketing towards volume over fit, which is why we’d drop the MQL target above €50k ACV.
One engaged contact isn’t coverage either. We count engaged roles against the roles the decision needs, per account. In our view, waiting for one champion to loop the others in leaves the deal to chance (how many people to map at €25k+).
And we think a results report without a pre-launch baseline is guessing. Without one, almost any number can be framed as a win. With one, the team and any partner it hires can agree on what changed instead of trading opinions. We don’t claim every effect can be isolated from everything else changing in the business. The baseline is the minimum bar for a credible comparison.
What should you report to the board each quarter, and what should you drop?
Report how target accounts moved against their baseline and any holdout, with the count behind every rate, and stop presenting activity counts as proof. A rate without its count invites the board to read noise as news.
Each quarter, report:
- Buying-group coverage: engaged roles against required roles, per target account
- Account progression: the share of target accounts that moved at least one stage
- The share of target accounts that opened an opportunity, against the holdout or baseline
- Qualified pipeline created in target accounts, against the same quarter before launch
- Win rate and revenue from target accounts once enough deals have closed, with the deal count
Stop presenting these as proof:
- MQLs, above all from accounts outside the list
- Clicks, impressions and engagement scores on their own
- Influenced pipeline
- ROI multiples from someone else’s study
Board reporting still leans away from outcomes. In 6sense’s 2025 survey, 39% of ABM programmes reported MQAs (marketing-qualified accounts) to the board, and 13% reported closed-won revenue.↗ According to 6sense, sales usually reports pipeline and revenue from target accounts instead.↗ Nearly half of ABM adopters still measured success by MQLs, even from accounts outside the list.↗ The survey covered 634 B2B marketers, and its dataset adds 100 synthetic responses.↗
Check every multiple before it reaches a slide. The provider’s guide above gives 208% twice: as ITSMA’s figure for ABM’s marketing ROI, and in a table as Marketo and SiriusDecisions’ figure for the revenue of aligned sales and marketing teams.↗
In practice
How we do it at Panelhop
In a Panel Check (GTM audit · 2–3 weeks), we take the baseline from your CRM before anything changes: the 7 metrics on our method page, among them qualified pipeline from target accounts, account progression and buying-group coverage. Each gets its value, date, definition, data source and sample size. The sheet is signed off with the readout and then frozen, and any later change to a definition is logged.
With Panel Ops (we run it monthly), every monthly report sets the after-numbers against that sheet. Read the method.
Questions buyers ask about this
How many closed deals does win rate need before it can prove ABM?
By our Illustrative maths, telling a 20% win rate from 30% takes 294 closed deals, won or lost, in each group, and a doubling to 40% takes 82. Divide by the deals each group closes in a year to see how long your team would need.
Can we prove ABM with fewer than 200 target accounts?
Not through win rate, and a holdout of that size catches only very large changes in the share of accounts opening an opportunity. Compare the target accounts with their own pre-launch baseline on buying-group coverage and account progression, and call the result a before-and-after comparison.
Is influenced pipeline a good way to prove ABM?
Not on its own. It counts opportunities in accounts that any campaign touched, so it rises with activity whether or not the programme changed the outcome. Report qualified pipeline created in target accounts against the same period before launch instead.
How soon should an ABM programme show results at €25k+?
We’d expect buying-group coverage and account progression to move first, because the programme acts on them directly. A rise in the share of accounts opening an opportunity takes a year and a few hundred accounts a group to read; win rate needs 82 closed deals a group even for a doubling (Illustrative).
What should replace MQLs in an ABM board report?
Account measures set against the baseline: buying-group coverage, account progression, the share of target accounts opening an opportunity and qualified pipeline from target accounts. Add win rate once enough deals have closed, with the deal count beside it.
