Why are B2B buying committees expanding to 18 members in 2027 despite AI tools?

A $600K renewal that started with four people and ended with nineteen

Picture a mid-market manufacturer consolidating its revenue stack. The initial conversation involves four people: a VP of Sales who owns the pain, a RevOps manager who inherited seven overlapping tools, a director of demand gen who wants attribution that finally reconciles, and the account executive on the vendor side. Everyone agrees on the problem in week two. On a 2019 timeline, that deal closes in ninety days with a signature from the VP of Sales and a rubber stamp from finance.

In 2027 it closes in eleven months with nineteen names on the approval chain, and every single addition has a defensible reason behind it.

Here is roughly how that roster fills in. The RevOps manager runs an AI-assisted evaluation in week three and the output flags a data residency question, because the platform stores conversation intelligence transcripts in a region the company's own customer contracts restrict. That pulls in the data protection lead. The data protection lead reads the vendor's subprocessor list and asks whether the embedded model provider counts as a subprocessor under the company's existing customer commitments — a question nobody on the original four had the standing to answer — so outside counsel joins for two sessions.

Meanwhile the VP of Sales shares the proposal in a Slack channel. A regional sales director in EMEA sees it, runs the vendor's own ROI calculator against her segment's numbers, and concludes the payback math does not hold outside North America. She is not on the committee. She raises it anyway, and now she effectively is. The AE has to build a second business case for her region, which requires the vendor's solutions consultant, which requires the buyer's own analytics lead to validate the baseline numbers the case rests on.

Around month five, the company's CIO institutes a standing rule that any platform touching customer records requires a security questionnaire and a SOC 2 review before contract redlines begin. That is two more people — an IT security analyst and the vendor management specialist inside procurement who owns the questionnaire workflow. Month seven, the original VP of Sales is promoted into a CRO role and backfills her old seat with an external hire who, reasonably, wants to re-litigate the shortlist. Two vendors that had been eliminated come back onto the comparison grid. The new VP brings her own trusted analyst.

By month nine the roster reads: CRO, VP Sales, new VP Sales hire, RevOps manager, RevOps analyst, demand gen director, EMEA sales director, marketing ops lead, CFO, FP&A analyst, controller, CIO, IT security analyst, enterprise architect, data protection lead, outside counsel, procurement vendor management specialist, procurement category manager, and a customer success leader who will own adoption. Nineteen. Not one of them is padding. Each represents a genuine approval gate or a genuine veto, and each was added in response to a specific piece of information that surfaced during the evaluation rather than at kickoff.

That is the shape of the problem. The committee did not start at eighteen or nineteen. It *grew* there, additively, one justified name at a time, and the information that justified each addition was itself produced or accelerated by the tooling that was supposed to make the process leaner.

The mechanism: information access decoupled from decision authority

The core dynamic is a decoupling. Historically, the cost of evaluating a vendor was high enough to act as a natural gate. Reading an analyst report, building a comparison matrix, modeling total cost of ownership, and interpreting a security posture each required either specialized skill or a meaningful investment of hours. That cost meant only people with a mandate bothered. The mandate and the evaluation were bundled together.

AI unbundled them. A director-level manager can now paste a vendor's documentation into an assistant and get a competent risk summary in four minutes. She can generate a TCO model without finance. She can compare two contracts' termination clauses without legal. What she cannot do is approve the purchase. The evaluation capability spread across the organization; the signing authority did not move at all. The predictable result is a large population of people who hold informed opinions and no authority, and the only way for them to convert an opinion into influence is to get onto the committee — or to influence someone who already is.

There is a second, subtler mechanism running alongside it. AI-assisted evaluation is very good at surfacing dependencies and risks that used to stay invisible. That is genuinely valuable. But every surfaced risk maps to an organizational owner, and organizations resolve risk by routing it to the person accountable for that category. Surface a data residency question and you have summoned the data protection lead. Surface an integration dependency on a system of record and you have summoned the enterprise architect. Surface a renewal-term concentration risk and you have summoned procurement's category manager. The tooling is functioning correctly. Correct function means more risks identified, and more risks identified means more owners pulled in.

Notice the loop in the middle of that diagram. It is the part practitioners underestimate. A new committee member does not arrive as a passive approver. They arrive with the same evaluation tooling everyone else has, they apply it through the lens of their own function, and they surface a new class of risk that the previous members never had the domain knowledge to look for. Security finds things sales never would. Legal finds things security never would. Each addition has a non-trivial probability of generating the next addition.

The loop terminates when the marginal new member stops surfacing novel risks — when the functional coverage of the committee is broad enough that a new perspective finds only issues already assigned to someone. In practice that saturation point sits somewhere in the mid-to-high teens for enterprise platform purchases, which is precisely why the observed averages land where they do. Committees are not expanding toward eighteen because eighteen is a target. Eighteen is roughly where the risk-discovery loop runs out of new functions to summon.

A third mechanism is temporal. Long evaluations intersect with ordinary organizational churn. If a purchase decision takes ten months and the buying organization has normal turnover and reorganization rates in its management ranks, some fraction of the original committee will not be in the same seat at signature. Replacements do not inherit context, so they either re-open questions or bring their own advisors — and the committee ratchets upward rather than swapping cleanly, because the departed member's institutional memory often lives in a peer who now gets consulted informally.

What the numbers actually look like, and how they scale with deal size

The eighteen-member figure is an average across enterprise-scale purchases, and treating it as a universal constant is the fastest way to misplan a deal. The distribution matters more than the mean, and committee size correlates most strongly with three variables: annual contract value, the number of internal functions whose daily workflow changes, and whether the purchase touches regulated data.

Contract value is the crudest predictor but the most reliable. Purchases under roughly $50K in annual value frequently close with three to six people involved and often no formal procurement process at all, because they fall below the threshold that triggers the company's own review policies. Between $50K and $150K you typically see six to ten, with finance and IT appearing as reviewers rather than blockers. Between $150K and $500K the committee usually lands between ten and fifteen, and this is the band where security review becomes mandatory at most enterprises. Above $500K, and especially on multi-year platform commitments, the mid-to-high teens is the norm and the low twenties is unremarkable.

The functional-footprint variable is less obvious and more actionable. A tool used by one team, integrating with nothing, will hold a small committee even at a large dollar figure — a data warehouse credit commitment can be expensive and still be a two-function decision. Conversely, a moderately priced platform that changes how sales, marketing, service, and finance each do their jobs will pull a large committee even at $200K, because each affected function insists on representation. When you are forecasting committee size, count the functions whose process changes and multiply by roughly two to three people per function, then add the fixed overhead of procurement, legal, and security.

That fixed compliance overhead is worth isolating because it is the segment that grew most sharply. Purchases involving personal data, financial records, health information, or AI-driven decisioning now routinely require independent sign-off from legal, information security, data privacy, and procurement's vendor risk function. That is four seats before anyone has discussed whether the product is good. In heavily regulated industries add a model governance or AI oversight reviewer and you are at five or six. These reviewers hold genuine veto power and are structurally unmotivated to move fast — their downside from approving something bad vastly exceeds their downside from a slow review.

Cycle length scales with committee size but not linearly, and the relationship is worth understanding because it drives how you should sequence work. The cost is not primarily the extra approvals themselves; it is the scheduling. Each additional stakeholder who requires a synchronous meeting adds calendar friction that compounds, because coordinating five calendars is materially harder than coordinating three. Reviews that can run in parallel cost almost nothing in elapsed time. Reviews that must run sequentially — legal cannot redline until security clears, security cannot clear until architecture confirms the integration pattern — are where months disappear. Two deals with identical eighteen-person rosters can differ by four months purely on whether their review gates were parallel or serial.

For RevOps teams building forecast models, the practical takeaway is to stop treating committee size as a single number in your CRM and start capturing three fields: count of distinct functions represented, count of hard veto holders, and whether compliance review has started. Those three predict close date far better than a headcount does. A deal with nineteen names but only three veto holders and compliance already cleared is in far better shape than a deal with eleven names where legal has not yet been engaged.

Working with the committee instead of trying to shrink it

The instinctive response to committee expansion is to fight it — find the real decision maker, go over the noise, get to the economic buyer. That approach still occasionally works and is increasingly a losing strategy, because the additional members are not noise. They are veto holders whose objections surface late and kill deals in the final third of the cycle, which is the most expensive place to lose.

The alternative strategies each carry real trade-offs.

Executive sponsorship as a shortcut. Securing a senior executive who commits to driving the decision does compress cycles when it works, primarily by converting sequential reviews into parallel ones and by giving reviewers cover to approve quickly. The trade-off is fragility. Executive-sponsored deals are disproportionately exposed to the sponsor leaving, changing priorities, or getting reorganized — and when the sponsor evaporates, the deal often has no independent support because nobody else was cultivated. It also tends to breed resentment in the functions that got steamrolled, which surfaces during implementation as passive non-adoption.

Parallel workstream engagement. Rather than sequencing stakeholders, engage each function's requirements simultaneously from early in the cycle — send the security questionnaire in week two rather than month six, get the data processing agreement into legal's queue before pricing is settled. This front-loads effort onto deals that may not close, which is a genuine cost, and it requires the vendor to have documentation ready that many do not. But it converts the dominant cost driver — serial gating — into parallel work, and it is the single highest-leverage change most teams can make.

Radical qualification. Simply declining to pursue deals that will require large committees, and focusing on smaller-footprint purchases with short approval chains. This is a real strategy with real economics behind it, and it caps your deal size. It works well for products with genuine departmental value and poorly for platforms.

Procurement-led standardization. From the buying side, some organizations respond by pre-approving vendor categories, running periodic bake-offs, and letting individual teams buy from an approved list without reconvening a full committee. This moves the committee work upstream into a once-a-year process rather than repeating it per purchase. The trade-off is reduced flexibility and a real risk that the approved list ossifies, but for organizations making many similar purchases it is the most structurally sound answer available.

There is a hybrid worth naming, because it is what the strongest teams actually run. Use parallel engagement as the default operating mode, cultivate a coalition of two to four advocates across different functions rather than a single champion, and reserve executive sponsorship for unblocking specific stalled gates rather than as the primary driver. The coalition structure is what makes the deal survive personnel changes — when one advocate leaves, the others carry context forward, which is exactly the failure mode that turns a ten-month cycle into a fourteen-month one.

Pitfalls that turn a manageable committee into a stalled deal

Discovering the committee late. The most common and most expensive failure is mapping the buying group at the proposal stage rather than during discovery. By then the deal has an assumed timeline that the actual approval chain cannot support, the forecast has been committed, and every newly surfaced stakeholder reads as a slip rather than as expected process. Ask about the approval path in the first substantive conversation, specifically: who has to sign, who can say no, and what reviews are mandatory at this dollar amount. Buyers answer this readily when asked early and become evasive when asked late, because by then the question sounds like pressure.

Treating silent members as absent. Committee members who never speak on a call are frequently the ones who kill deals, because their function's objection is a formal review output rather than a conversational one. Security does not argue in a meeting; security returns a questionnaire with unacceptable findings. Track every named participant's function and confirm what their specific approval requires, even — especially — the ones who have said nothing.

Letting the champion carry information alone. When a single internal advocate is the sole conduit to the rest of the committee, you have no visibility into how your product is being represented and no recourse when that person leaves. The failure is usually silent: the deal simply stops progressing while the champion, overloaded, fails to move it internally. Insist on direct access to at least two other functions, and give the champion material designed for internal forwarding — a one-page business case a CFO can read, a security summary IT can file — rather than material designed for the champion.

Re-litigating instead of re-onboarding. When a new member joins mid-cycle, the reflex is to defend the decisions already made. That reads as defensive and invites a full reopening. The better move is a deliberate onboarding: a short written summary of what was evaluated, what was eliminated and why, and what remains open. New members who receive that summary tend to accept prior work and focus on their own gate. New members who receive resistance tend to reopen everything.

Optimizing for consensus rather than for permission. Not every committee member needs to be enthusiastic. Most need only to not object. Spending equal effort on all eighteen is a misallocation — the practical segmentation is a small number who genuinely decide, a larger number who can veto, and a remainder who need to be informed enough not to raise late questions. Effort should be concentrated on the veto holders, who are usually the most neglected group precisely because they are the least pleasant to engage.

Ignoring the compliance clock. Security and legal reviews frequently have their own queue times independent of your deal urgency. A questionnaire submitted in month six does not get reviewed faster because your quarter ends. Submit early, ask directly what the current queue time is, and build the answer into your forecast rather than discovering it.

Assuming AI tooling will reverse the trend. It will not, at least not through better evaluation features. Better tooling makes evaluation cheaper, which spreads it wider, which surfaces more risks, which summons more owners. Committees will shrink when approval authority is restructured — through pre-approved vendor catalogs, delegated spending authority, or standardized procurement — not when the analysis gets smarter. RevOps teams planning around a coming contraction are planning around something that is not arriving.

Related questions

Does a bigger committee mean a lower win rate?

Not necessarily. Larger committees correlate with longer cycles and higher no-decision rates, but deals that do close from large committees often show stronger retention because more objections were resolved before signature rather than after. The risk is stalling, not losing head-to-head.

Which committee member should a seller engage first?

Whoever owns the most likely veto, not the most enthusiastic advocate. Enthusiasm is cheap to maintain and vetoes are expensive to discover late. In practice that usually means security or procurement, engaged in parallel with the business case rather than after it.

How should RevOps model committee size in the CRM?

Capture distinct functions represented, count of hard veto holders, and compliance review status as separate fields. Raw headcount is a weak predictor of close date; those three fields correlate far more tightly and are usable in stage-gate criteria.

Do smaller companies show the same expansion?

Directionally yes, at smaller absolute numbers. A 200-person company making a $100K platform commitment might involve six to nine people rather than eighteen, but the same mechanism applies — broader evaluation access pulls in functional owners who previously would have deferred.

Is the trend different for renewals versus new purchases?

Renewals above meaningful dollar thresholds increasingly resemble new purchases, particularly where procurement runs periodic vendor reviews. The committee is often smaller because context exists, but compliance re-review is frequently mandatory, so the veto-holder count stays similar.

FAQ

Why hasn't AI shrunk buying committees the way it was expected to?

Because AI reduced the cost of evaluation, not the distribution of approval authority. Cheap evaluation means more people form informed opinions, and more informed opinions mean more people who want a seat. Meanwhile the risks that AI-assisted analysis surfaces each map to a functional owner who must be brought in to resolve them. The technology addressed the analysis bottleneck while leaving the governance bottleneck untouched.

Is eighteen members a hard number I should plan around?

No. Treat it as a central tendency for enterprise-scale purchases, not a rule. Small departmental purchases still close with three to six people. The variables that actually drive size are annual contract value, how many functions change their daily workflow, and whether regulated data is involved. Forecast from those inputs rather than from an average.

How do I keep an eleven-month cycle from becoming fourteen?

Convert sequential reviews into parallel ones and build a coalition rather than relying on one champion. Most of the extra time in long cycles comes from serial gating — legal waiting on security, security waiting on architecture — and from re-onboarding replacements after turnover. Both are addressable. The approvals themselves are rarely the slow part.

Do the extra members actually have veto power, or are most of them observers?

It varies, and distinguishing them is the highest-value thing a seller can do. Typically a minority genuinely decide, a larger group can block within their functional domain, and the rest are informed parties. Compliance, security, and procurement roles almost always hold real vetoes despite rarely participating in sales conversations.

Will committees start contracting again as tooling matures?

Not through better analysis features. Contraction requires structural change to who can approve — pre-approved vendor catalogs, higher delegated spending limits, or standing procurement frameworks that do the committee work once annually instead of per deal. Some large organizations are moving that direction, but it is a governance change, not a technology one.

What's the single highest-leverage change a RevOps team can make?

Instrument the buying group properly in the CRM and enforce it as stage-gate criteria. Most forecast misses on large deals trace back to an approval requirement nobody documented — a security review, a data processing agreement, a procurement threshold. If your pipeline data cannot answer "who still has to say yes," your forecast is guessing on exactly the deals that matter most.

Sources

• Gartner — B2B Buying Journey research
• Harvard Business Review — The New Sales Imperative
• McKinsey — The new B2B growth equation
• Forrester Research — B2B marketing and sales insights
• Gong Labs — sales research and data
• HubSpot Sales Blog
• SaaStr
• Salesforce — sales resources
• Bain & Company — B2B customer insights

Related on PULSE

• Why are B2B buying committees expanding to 14+ members in the current 2027 market?
• How does the expanding size of B2B buying committees increase the risk of vendor consolidation paralysis?
• Why are buying committees expanding to include AI ethics officers in 2027?
• Why are 2027 renewal cycles expanding by 30% when AI adoption is supposed to accelerate retention?
• Should I Hire a Fractional CRO If I Am Expanding Internationally Next Year?