Every platform says it can match your account list. The word "matched" appears in the deck, in the proposal, in the post-campaign report. It sounds like identification. It feels like precision. In most cases, it is neither.
Matched audiences B2B is one of the most overloaded phrases in media buying right now. What it describes can range from a deterministic email match against a first-party database to a probabilistic inference built from job-title targeting and company-size filters. Both are described as "matched". The gap between them is the gap between knowing the person is in the room and guessing they might be.
That distinction matters more in B2B than almost anywhere else, because the room you are trying to reach is small. A buying committee for an enterprise mandate might be twelve people across three functions. Reach eleven of them with reasonable frequency and you have done something commercially useful. Reach a modelled lookalike of twelve people who share some behavioural characteristics with your target but sit in different companies entirely, and you have produced a very clean-looking report with nothing behind it.
Why this confusion costs money
The marketing lead goes to the board with an impression count and a reach figure. Both look large. The sales team walks into the first meeting with the account and finds no one has heard of them. The board asks what the campaign did for the named accounts. Marketing cannot say.
This is not a technology failure. It is a vocabulary failure that becomes a budget failure. When a platform says "matched", the buyer rarely asks what method produced the match, what the match rate was, or what happened to the unmatched records. Those three questions determine whether the audience reached is the buying committee or a statistical approximation of people who are vaguely similar to the buying committee.
The cost is not just the wasted media spend. It is the false confidence. A campaign that reports 87% match rate but does not define what "match" means has given the marketing lead a number that feels like accountability but provides none. When sales asks what the committee already knows about the firm, the honest answer is: we do not know, because we do not know exactly who we reached.
This is the core pain in finite-account B2B. The accounts are known. The roles are knowable. And yet the media plan routinely buys a broader, softer version of those people and calls it precision. Starting the media plan from your CRM rather than from a platform audience is the beginning of the fix, but only if you understand what the platform can and cannot confirm about who it actually reached.
What does "matched" actually mean in practice?
There are broadly two things happening when a platform claims to match your account list, and they work very differently.
The first is deterministic matching. A record in your CRM, most commonly an email address or a device identifier, is matched against a record in the platform's own database. The person who logged into LinkedIn with that email address sees your ad. The match is a confirmed identity. The platform can tell you, with reasonable confidence, that a specific individual was served your creative. Match rates here are typically 30% to 60% of uploaded records, depending on data quality and platform coverage. The records that do not match drop out. They are not replaced with approximations.
The second is modelled or probabilistic matching. Your account list or a set of targeting parameters is used to build an audience that resembles your targets. The platform finds people who work at similar companies, hold similar job titles, show similar browsing behaviour, or have engaged with similar content. These people are not on your original list. They are the platform's inference about who else might be relevant. The match rate can appear very high because the model is designed to find enough people to deliver the campaign, not to stay faithful to your original account list.
Both produce audiences. Only one produces matched audiences B2B in the sense that most buyers think the phrase means.
Why platforms are not always clear about which method they use
The incentive structure does not reward clarity here. A deterministic match rate of 40% on a 200-account list is a small, expensive audience to serve. A modelled expansion of that same list can produce a much larger, cheaper audience that delivers a better-looking CPM and a more impressive reach figure. Platforms are measured on delivery. Delivery rewards scale. Scale rewards modelling.
This is not deception, exactly. The tools work as described if you read the documentation carefully. The problem is that most media briefs do not specify which method is required, and most post-campaign reports do not distinguish which method was used. "Matched audience" appears in both columns.
The result is that the marketing lead who asked for account-level precision often receives something closer to job-title targeting with better branding. The audience has been modelled to resemble the accounts. The accounts themselves may or may not have been reached. There is often no way to tell from the report.
In regulated financial services, where a buying committee might be three trustees, a consultant and an investment officer, that distinction is the entire campaign. A modelled expansion of those five people is not five people. It is five hundred people who are statistically similar to five people. The committee may never have seen the creative.
The three questions to ask before you accept a match claim
If a platform or agency tells you the campaign reached your matched audiences B2B, ask three things before accepting the figure.
First: what was the match method? Deterministic means a confirmed identity in the platform's own database. Probabilistic or modelled means an inferred lookalike. If the answer is vague, the method is almost certainly modelled.
Second: what was the match rate, and what happened to the unmatched records? A 40% deterministic match rate on 200 accounts means 80 accounts were reachable through that method. The other 120 were either not reachable or were replaced by modelled expansion. If the campaign still claimed to reach 200 accounts, ask how.
Third: at what level was the match applied? Account-level matching means the platform identified the company domain and served ads to people at that domain, which is a reasonable proxy but not individual identification. Role-level matching means the platform identified the specific function or seniority within the account. Individual matching means a specific person in the database was confirmed. These are materially different claims, and the difference changes what the reach figure actually means.
These questions are not adversarial. They are the minimum needed to understand what the campaign actually did. A platform or partner confident in its method will answer them directly. Vagueness in the answer is information.
The reframe: addressability is a spectrum, not a binary
The more useful way to think about this is not matched or unmatched, reached or not reached, but rather: how confident are we, and at what level?
Some accounts on a target list will be reachable with high confidence. A key decision-maker at a named asset manager who is active on LinkedIn and registered for a recent industry event has left multiple addressable signals. A mid-level procurement officer at a company with no digital footprint and no event attendance may be reachable only through modelled approximation.
A serious campaign does not pretend the second case is the same as the first. It builds around the accounts where confidence is highest, uses every available method to extend reach into the harder accounts, and reports what method produced what result. The report should separate confirmed account reach from modelled expansion from directional inference, because each carries a different weight in the sales conversation.
This is what honest account-level reporting looks like. Not one reach figure. A layered picture that says: these accounts we reached with high confidence through deterministic methods; these accounts we reached through domain-level targeting; these accounts we attempted but cannot confirm; these accounts were outside our addressable universe for this campaign period.
That report is less impressive-looking than a single large reach number. It is also the only version of the report that tells the sales team anything useful. Reaching five of the twelve honestly is more valuable than claiming twelve and delivering guesses.
What this means for the campaign design
If you accept that addressability is a spectrum, campaign design changes. You stop trying to reach everyone on the list through one method and start building a structured approach to each addressability tier.
The most addressable accounts, where deterministic matching is possible and role-level targeting is available, get deep, frequent, coordinated exposure. Multiple touchpoints across the places the buying committee actually moves through: professional networks, trade publisher inventory, relevant content channels. Repeated enough that the firm becomes familiar before the sales call, not a stranger introduced by a cold email.
The accounts where only domain-level or modelled matching is available get a different treatment. Broader reach within the account. More reliance on contextual relevance. Less frequency because confidence in who is seeing the creative is lower. The goal is still familiarity, but the claim is softer: we reached people at these companies, not we reached these specific buyers.
The accounts that are genuinely outside the addressable universe for this period are noted. They may become a signal that a different channel or a different moment in the buying cycle is needed. That is useful information. It is not a failure.
This is what cheap reach misses. A broad campaign that claims to reach all 200 accounts through modelled expansion has not actually resolved the addressability question. It has papered over it with a large number and a vague methodology. When the account does not move, there is no way to know whether the problem was the creative, the message, the timing, or the fact that the buying committee never saw the campaign at all.
What honest reporting looks like
The reporting problem and the matching problem are the same problem wearing different clothes. Both come from the same unwillingness to say clearly what happened and what the evidence supports.
Honest reporting on matched audiences B2B separates three things. What the campaign delivered: accounts reached, by what method, at what confidence level, with what frequency. What the evidence suggests: changes in account behaviour, owned-site activity, engagement signals, meeting requests. What the evidence does not prove: whether the campaign caused the account to move, whether the committee's familiarity increased, whether the next mandate will be won.
That third category is the one most reports omit. Saying what the evidence does not prove is not a sign of weakness. It is a sign that the reporting can be trusted. A report that claims to prove everything proves nothing, because the reader knows the claim is too strong and discounts the whole thing.
For the marketing lead who has to face the board, this is actually a better position. A clear account of what happened, what it suggests and what remains unknown is a defensible position. A reach figure built on a methodology no one has interrogated is not.
Showing the board what the budget did for named accounts requires this kind of layered honesty. The board does not need to know the technical difference between deterministic and probabilistic matching. It does need to know that the answer is grounded in something real.
What we have seen in practice
A top-tier South African asset manager ran a campaign targeting a finite set of institutional stakeholders: trustees, investment consultants and asset-allocation decision-makers at a defined list of named accounts. The campaign was built around accounts where addressability was genuinely high, with role-level targeting available through professional networks and trade publisher inventory.
Over the campaign period, landing-page views increased 554.1% versus the prior period, sessions increased 498.3% and unique users increased 432.3%. Unique reach across the campaign was 61,080, with an average daily frequency of 9.27 exposures per reached user. A separate campaign against 890 target stakeholders recorded that 93% showed engagement, with goal completion rising from 1% to 35%.
These figures are contribution, not sole attribution. Other factors were in play. But the point is not the headline number. The point is that the campaign was designed around accounts where addressability was real, not claimed. The reach figure meant something because the matching methodology was specific enough to make it mean something. The report could say what happened because the campaign was built to produce that answer.
That is what an honest matched-audiences B2B campaign looks like in practice. Not the largest reach number. The most useful one.
Ready to understand what your account list can actually reach?
If you already have the accounts you need to win, the first question is what proportion of them are genuinely addressable, at what confidence level, and through which methods. That question has a real answer, and it changes the shape of the campaign before you spend a rand.
Brief Dot to start with your named account list and understand what is actually reachable.
Or see how it works to understand Find them, Surround them, Move them.
Frequently Asked Questions
What is the difference between deterministic and probabilistic matching in B2B?
Deterministic matching confirms a specific identity by linking a record in your data, usually an email address, to a confirmed identity in the platform's own database. Probabilistic matching builds an inferred audience that resembles your targets based on shared characteristics like job title, company size or browsing behaviour. Both are described as matched audiences B2B, but only deterministic matching tells you a specific person was reached.
Why do match rates look high even when the audience is largely modelled?
Platforms are incentivised to deliver campaigns at scale, and modelled expansion allows them to reach a much larger audience than a strict deterministic match would produce. A high match rate may reflect the platform's ability to find lookalike users rather than confirming it reached the individuals on your original account list. Always ask whether the reported match rate refers to confirmed identities or modelled expansion.
How many accounts on a typical B2B list are genuinely addressable?
This varies significantly by sector, market and the quality of the underlying account data. In practice, deterministic match rates on a named account list typically fall between 30% and 60% of uploaded records, depending on platform coverage and data completeness. The remaining accounts may be reachable through domain-level targeting or contextual methods, but at lower confidence, and that distinction should be reflected in the campaign design and reporting.
Does matched audiences B2B work for very small buying committees?
It can, but the methodology needs to be more rigorous, not less, precisely because the committee is small. Reaching five of twelve people with real frequency is more useful than reaching a modelled expansion of twelve people with low confidence. Small committees reward depth of coverage over breadth, which means the match method and confirmation level matter more, not less, as the target audience shrinks.
What should an honest post-campaign report say about audience matching?
It should separate confirmed account reach from modelled expansion from directional inference, and state the method used for each. It should say what the evidence suggests about account behaviour changes and what it does not prove about causation or committee familiarity. A report that blends all three into a single reach figure without explanation is not giving you the information you need to defend the spend or improve the next campaign.
Is domain-level targeting a reasonable substitute for individual matching?
Domain-level targeting, where the platform serves ads to people at a named company domain, is a useful method when individual matching is not possible, but it is a proxy, not identification. It increases the probability that someone within the account saw the creative, but it cannot confirm which role or individual was reached. Used alongside deterministic matching where available, it adds coverage; used as the primary method and labelled as matched audiences B2B, it overstates what the campaign confirmed.