How do I evaluate whether a new vertical is worth the GTM investment in 2027?

The two ways companies actually enter a vertical

Every vertical-expansion conversation eventually collapses into two competing paths, and the entire evaluation is really an argument about which one you are choosing. The first path is the opportunistic overlay: keep the horizontal product, point existing reps at the vertical, add a landing page and a case study, and see what sticks. The second path is the committed vertical build: dedicated engineering to construct the vertical data model and integrations, specialist sellers who speak the industry's language, vertical marketing with its own conference calendar, and an executive sponsor who owns the P&L. These are not two points on a spectrum with a comfortable middle. They have different cost structures, different failure modes, and different time-to-evidence, and pretending you can hover between them is how companies spend committed-build money for overlay results.

The overlay path is cheap and fast. You can stand it up in a quarter for the cost of some marketing content and a reallocated rep or two — call it $100K to $300K of incremental spend for the first two quarters. Its virtue is that it produces real market signal without a roadmap commitment. Its limitation is equally real: an overlay cannot win a vertical where the buyer's must-have requirements sit outside your product. You will generate meetings, lose the deals at the technical evaluation, and conclude — wrongly — that the vertical does not want you. What the overlay actually tests is whether the vertical will *talk* to you, not whether it will *buy* from you.

The committed build is the opposite trade. It costs $2M to $5M a year once a dedicated pod exists, plus six to twelve engineering-quarters of net-new work before you have a credible v1 to demo. It can genuinely win a vertical where requirements diverge from your core. But it commits capital and roadmap before you have evidence, and it is functionally irreversible inside eighteen months — you cannot un-hire specialist reps or un-build a vertical data model without writing off the whole investment and, usually, the executive who championed it.

There is a third option that most evaluations skip and that usually beats both: the wedge pilot. You pick one narrow, painful, well-bounded use case the vertical cares about, build only what that use case requires, and sign three to five paying design partners. This costs $300K to $800K over two or three quarters — an order of magnitude less than the committed build, an order of magnitude more informative than the overlay. It answers the only questions that matter: will real buyers pay, adopt, expand, and refer? The wedge pilot is not a compromise between the other two paths. It is the correct sequencing of them: overlay-level cost, build-level evidence.

The practical decision, then, is rarely "vertical yes or no." It is "which of these three commitments does the evidence currently justify?" A vertical with strong organic pull and a bounded engineering gap may deserve a wedge pilot immediately. A vertical with a huge TAM slide and no named accounts deserves nothing but a research memo. And a vertical where you already have paying design partners begging for more may genuinely deserve the pod. Matching the commitment level to the evidence level is the whole discipline.

Where each path breaks down

The overlay breaks on the engineering tax — the net-new engineering required before a vertical buyer considers you credible. Enumerate it in engineering-quarters, where one engineer working one quarter equals one eng-quarter. A vertical data model built around the industry's core nouns (encounters and claims in healthcare, RFIs and submittals in construction, positions and trades in financial services) typically runs two to six eng-quarters. Each dominant integration runs one to three. Compliance and security features — audit logs, data residency, role-based access, retention policies — add one to four. Vertical workflow logic, meaning the approval chains and business rules and edge cases the industry takes for granted, adds two to five. A "real but bounded" vertical build lands around six to twelve eng-quarters total, which is three to five engineers for six to nine months.

The overlay path assumes that number is near zero. It almost never is. The tell is when a product leader calls net-new engineering "configuration." A requirement is only configurable if a customer success engineer can stand it up today, in production, with zero engineering tickets. Everything else is engineering, with a cost and a calendar and an opportunity cost. Run a real requirement audit: pull the top twenty to thirty requirements a vertical buyer will evaluate you on, sourced from RFPs, competitor feature lists, and your own customer interviews, and classify each as already met, genuinely configurable, or net-new engineering. If you meet more than 80% with configuration, the overlay may actually work and you can pilot within a quarter. Between 60% and 80%, you are looking at two to four eng-quarters to a credible v1 — wedge pilot territory. Under 60%, you are not evaluating a vertical; you are evaluating a new product, and it deserves a new-product bar.

The committed build breaks on two things: regulatory timeline and focus dilution. Regulated verticals gate entry behind certifications that cost real money but, more importantly, consume calendar you cannot buy back. SOC 2 Type II — the baseline for almost any serious B2B buyer — requires an observation period measured in months, not a check you write. HIPAA compliance in healthcare requires business associate agreements, technical safeguards under the HHS Security Rule, and the engineering to survive a security review on every single deal. Financial services layers on books-and-records and communications-retention obligations under FINRA and SEC rules, whose compliance review reliably becomes the longest pole in the sales cycle. FERPA governs student data in education. FedRAMP, for public sector, is the most expensive and slowest of all and should never be undertaken as a side bet. When you total the regulatory cost, treat the timeline as the binding constraint: a vertical that costs $500K to enter but requires eighteen months of certification before you can legally sell has an eighteen-month hole in its payback model that the dollar figure alone hides completely.

Focus dilution is the failure mode nobody scores and everybody suffers. A vertical bet pulls your best engineers, your strongest reps, and — most scarce of all — executive attention. The visible failure is the vertical underperforming. The invisible and more expensive failure is the core business stalling while leadership's attention was elsewhere. This cost never appears on a budget line, which is exactly why it must be forced onto the scorecard deliberately: write down the specific core initiative that will not ship, by name, and put it in the memo.

The wedge pilot's own failure mode is subtler: wedge drift. Design partners ask for adjacent features, the team says yes to keep them happy, and within two quarters you are building the full committed product without ever having made the committed-build decision. Guard against it by writing the wedge boundary down at the start and requiring an explicit gate review — not a Slack thread — to expand it.

How to decide between the paths

The decision mechanism that survives contact with reality is a weighted six-factor scorecard, scored one to ten on gathered evidence, multiplied by weight, summed into a single composite. Its purpose is not to produce a magic number. It is to force a written, comparable, evidence-backed conversation so the decision turns on evidence rather than on the enthusiasm of the loudest executive in the room.

Weight product fit heaviest, at roughly 25%, because the engineering tax is the most underestimated line item in vertical expansion and the most common cause of quiet failure. Market size takes about 20%, measured as bottoms-up five-year SOM rather than any top-down number. GTM cost takes another 20% — the fully-loaded spend to reach the first $1M to $3M of vertical ARR. Competitive density takes 15%, measuring incumbent strength and, critically, switching costs. Regulatory complexity takes 10%, scored on cost and calendar to legally sell. Expansion and retention takes the final 10%, measuring structural stickiness once you land.

Above a 7.0 composite, fund the research gate. Between 5.5 and 7.0, fund only the cheap research gate with the yellow factors as the explicit research focus. Below 5.5, say no out loud and redeploy. Then apply the override that makes the scorecard honest: any single factor scoring one or two triggers a mandatory fatal-flaw review regardless of composite. A vertical scoring eight on five factors and two on regulatory complexity computes to a respectable 7.4 and may still be un-enterable. Weighted averages flatter fatal flaws; the override is what stops them.

Before any of that arithmetic, run the cheapest and most predictive test available: can you name ten logo-quality target accounts in this vertical, and have you actually talked to five of them? The naming test forces specificity — ten real companies that fit the ICP and would be credible logos, not "mid-size hospitals." If the team can only name three, the targetable universe is far smaller than the TAM slide claims, or the team does not understand the vertical's segmentation. The conversation test forces evidence: five real discovery calls with buyers in those accounts, before a dollar of build. Those calls answer everything the scorecard depends on — actual must-have requirements, the real buying center, what they run today and why, what would make them switch, what they would pay. Five honest customer conversations routinely kill a vertical that looked excellent on paper, and that is the test working. Learning "we will never rip out the incumbent" in a forty-five-minute call is dramatically cheaper than learning it in an eighteen-month, multi-million-dollar expansion.

One more check before the score is trusted: pull the organic-customer signal. Look at every current account already operating in the candidate vertical — accounts that found you and bought you with zero vertical-specific GTM effort. Organic vertical customers are revealed preference rather than stated preference, which makes them the strongest single validation signal available. For each, understand how they use the product, what they configured or worked around, what they asked for and never got, and how their retention and expansion compare to your blended book. Healthy, expanding organic customers raise your confidence in both product fit and retention, and they hand you ready-made references for the pilot. Zero organic customers after years of horizontal selling has an optimistic read ("we never marketed there") and a realistic one ("the product does not solve their problem") — and the realistic read should pull the product-fit score down. Organic customers who are churning or stuck at a low usage tier are the loudest warning of all.

The numbers behind each path

Build the market size bottoms-up or do not build it at all. The classic failure is a top-down number that is simultaneously true and useless: "there are 250,000 practices in this industry and they spend billions on software." That tells you nothing about how many accounts you can win, at what price, on what timeline. The disciplined SOM multiplies three components: genuinely targetable accounts, realistic vertical ACV, and a credible five-year penetration rate.

Subtract relentlessly on the account count. From a 250,000-practice universe, ask how many are multi-location groups large enough to need your product and able to pay your price — perhaps 18,000. How many sit in geographies and segments your GTM motion can actually reach — perhaps 12,000. How many are not locked into a long-term incumbent contract in any given year — perhaps 3,000 "in window" annually. That in-window number, not the headline, feeds the model.

Vertical ACV is rarely your core ACV. It runs higher when the vertical has acute pain and real budget, and structurally lower when the vertical is fragmented and price-sensitive. Build it from the buyer's actual budget reality and from a per-unit metric the vertical already uses to think about its business — per-bed and per-provider in healthcare, per-location in restaurants and retail, per-project in construction, per-account or per-transaction in financial services. A pricing metric that matches the buyer's mental model raises willingness to pay and makes the value story self-evident. The same product can support 2x ACV in one vertical and 0.5x in another, and that spread flows straight through the SOM.

Penetration follows an S-curve, and honesty here is the difference between a model and a fantasy. A realistic path runs roughly 0.5% of the targetable base in year one, 2% by year two, 5% by year three, 9% by year four, and 14% by year five. Fourteen percent of a well-defined targetable base is an aggressive, well-executed outcome, not a conservative one. If five-year SOM does not reach at least $15M to $30M of ARR — enough to be a genuine second act rather than a rounding error — the vertical probably cannot justify pulling engineering and executive focus off the core. Clear $30M with a credible curve and factor one scores eight to ten; $15M to $30M scores five to seven; under $15M, or top-down only, scores one to four.

On the GTM side, model the fully-loaded cost to the first $1M to $3M of vertical ARR, and do not fool yourself with "the existing team will also sell this." A new vertical means an unfamiliar buying center, unfamiliar objections, unfamiliar competitors, zero reference customers, and no brand permission. That is a different motion. Specialist AE capacity typically runs $500K to $1.2M a year for one to three loaded reps, and those reps take six to nine months to ramp. Vertical marketing — positioning, case studies, a content engine credible to an insider — runs $200K to $500K. Conference and event presence at the two to four must-attend industry shows runs $100K to $300K. Thought leadership and benchmark content runs $50K to $150K. Partnership and channel development runs $100K to $300K. Total fully loaded: $1.5M to $3M per year, sustained for two to three years.

Then compute the implied vertical CAC honestly. In years one and two it will run two to four times your blended core CAC, because you are paying the new-motion tax across every line: specialist hiring, long ramps, longer cycles, lower close rates without references, and marketing that has not yet compounded. Sales cycles run 1.5x to 2x your steady state for the same reason, and in slow verticals — hospital systems, banks, government agencies — nine to eighteen months is normal once security review, compliance review, procurement, and committee dynamics each add their weeks. A healthy steady-state target is CAC payback in eighteen to twenty-four months. If the model cannot trend there within roughly eighteen to twenty-four months of dedicated investment, the GTM factor scores low, and volume will not rescue it. The honest board framing: vertical economics look bad before they look good, and the plan must explicitly fund the look-bad period. Pretending a new vertical will show core-business CAC from quarter one is the single most common way a vertical financial model lies.

Competitive density needs its own numbers, but they are structural rather than dollar-denominated. Map the vertical specialists who built their entire business around the industry, the horizontal platforms that bolted on a vertical SKU, and the legacy or homegrown systems still running. Then price the switching cost, which is the real moat — not the incumbent's product. Switching cost comes from four sources: data gravity, meaning years of historical data living in the incumbent's system; workflow embedding, meaning the incumbent is wired into daily operations and adjacent tools; integration centrality, meaning the incumbent is the hub other vertical tools connect to; and contractual lock-in, including the political cost to the buyer of a failed migration. High switching costs do not make a vertical unwinnable, but they turn every deal into a rip-and-replace with a long cycle, and they should depress the score. Note the counterintuitive case: a vertical with *zero* credible incumbents is usually a warning sign rather than an opening. It often means the industry does not buy software the way you assume, or the pain is not acute enough to move budget. Some competition validates that a market exists.

Finally, make opportunity cost an explicit line rather than an implicit assumption. Six to twelve eng-quarters spent on a vertical are six to twelve eng-quarters not spent on core retention, core expansion, or platform reliability. One and a half to three million GTM dollars a year are dollars not deployed against the core pipeline. If the vertical's expected return does not clearly beat the core roadmap's return on the same resources, the honest answer is to keep the engineers on core. Write the sentence down: "Funding this vertical means the platform-reliability initiative slips two quarters and we do not pursue the enterprise-tier expansion this year." That sentence, debated openly, is worth more than another TAM slide — and it is frequently what turns a respectable 6.5 into a correct no.

Sequencing the commitment through gates

The mechanism that makes any of this safe is staged investment with real gates: capital committed in tranches, each unlocked only by evidence from the prior stage, so the downside is always capped and no team can quietly slide from idea to dedicated pod.

Gate 1 is research and discovery: $50K to $150K over six to ten weeks. Scope: build the bottoms-up SOM, score the six factors with each score justified by evidence, run the beachhead test to completion, enumerate the engineering tax in eng-quarters, scope regulatory cost and timeline with actual compliance counsel, pull the organic-customer analysis, and talk to two or three potential channel partners the same way you talk to potential customers. Exit criteria: composite above threshold, beachhead passed and documented, and a specific credible wedge identified.

Gate 2 is the wedge pilot: $300K to $800K over two to three quarters. Build only the minimum wedge product, sign three to five paying design partners — paying, not free, because free pilots test politeness rather than value — and run long enough to observe adoption and expansion signal, not just signature. Exit criteria: design partners pay, actually adopt, expand or credibly commit to expand, and agree to serve as references; and the unit economics still clear the bar after real-world contact.

Gate 3 is the dedicated vertical pod: $2M to $5M a year, multi-year, with dedicated vertical product, engineering, sales, and marketing under an executive sponsor. It is unlocked only by Gate 2 evidence, never by enthusiasm and never by a quarter that came in light on the core.

The cardinal rule is that you never skip to Gate 3. Funding a dedicated team before a pilot has proven the wedge is the single most expensive mistake in vertical expansion — it is how organizations burn millions and eighteen months to learn what ten customer interviews would have surfaced at Gate 1 for a hundred thousand dollars. Most vertical ideas should die at Gate 1, and most that earn a Gate 1 should never reach a Gate 3. The gates are not bureaucracy; they convert an unbounded strategic gamble into a sequence of bounded, evidence-gated bets.

Choose the wedge carefully, because it determines whether Gate 2 produces signal or noise. A good wedge is narrow enough that you can be unambiguously the best at it without an eighteen-month build, painful enough that the buyer will pay and adopt without a heroic change-management effort, and expandable enough that winning it earns the relationship, data, and credibility to sell the next use case. Wedges that fail one of those three tests produce pilots that succeed technically and teach you nothing commercially.

Acquisition strategy sequences alongside the gates. Building is right when the engineering tax is bounded at six to twelve eng-quarters, organic signal is strong, and the timeline is tolerable — you get full control and margin at the cost of speed and core resources. Buying is right when the window is closing, the engineering tax is large (product fit under 60%), and fairly-priced targets exist — you get speed and de-risk product and GTM simultaneously, at the cost of price and integration risk. Partnering is right when the vertical is strategic but not core enough for build or buy, when you want to test pull cheaply, or when the partner channel is genuinely the fastest path — cheapest to start and easiest to unwind, at the cost of margin and control. Many of the strongest vertical entries sequence these: partner to test pull, then build or buy once the thesis holds.

Run one more audit before Gate 3 releases the money: the internal capability audit. Ask honestly whether anyone in product and engineering genuinely understands the vertical's workflows and data model; whether you have or can hire reps with real credibility and a network in the industry; whether marketing can produce content an insider finds credible rather than embarrassing; and whether there is an executive sponsor with access to advisors who have deep vertical experience. The point is not to require existing expertise. It is to make the capability gap explicit and budget for closing it. The unforgivable version is the unaudited assumption that "we're smart, we'll figure the vertical out as we go." Verticals punish tourists, and the buyers can tell within one discovery call.

Finally, sequence the communication. Frame the bet to your board as TAM expansion with a capped, gated downside — you have identified a market that expands your addressable surface, you have evidence it is real, and you pursue it through staged gates so committed downside never exceeds the current tranche. Pre-commit each gate's cost, its unlocking evidence, and its kill criteria; a board that agreed in advance to the kill criteria will support a disciplined no later. Address focus dilution head-on rather than pretending the tension does not exist, because it is the first thing a sophisticated investor worries about. And report against the gates — design-partner adoption, refined CAC payback, requirement-audit movement — rather than against a pipeline number that photographs well in a deck. This is where RevOps earns its seat: owning the model, the gate metrics, and the honest reporting cadence that keeps a vertical bet from drifting into an unbounded adventure.

The uncomfortable base rate is that the right answer is "no" or "not yet" far more often than founders and ambitious GTM leaders want to hear. Most verticals that get evaluated look attractive — that is precisely why they got evaluated — and most should still die at a cheap research gate. The framework's value is not that it helps you say yes. It is that it helps you say no early, cheaply, and with evidence, so the rare vertical that clears every gate gets the focus and capital it needs to become a real second act. A disciplined no costs a slide deck. An undisciplined yes costs millions of dollars, several years, and sometimes the momentum of the core business.

Related questions

Should we expand into a new vertical or go upmarket instead?

Compare both against the same engineering-quarters and GTM dollars. Upmarket usually reuses your existing product and brand permission; a vertical usually does not. If the enterprise motion clears a similar SOM with a smaller engineering tax, it wins on opportunity cost alone.

How many customer interviews are actually enough before committing?

Five completed discovery conversations with buyers at named target accounts is the minimum gate, ten is the working standard for Gate 1. Beyond fifteen, marginal learning drops sharply. Quality of access matters more than count — five real buyers beat twenty analysts.

What if we already have organic customers in the vertical?

That is the strongest signal available and it raises confidence in both product fit and retention. Interview every one: what they configured, what they worked around, what they asked for and never got. Use them as Gate 2 references and as a free requirements source.

Does an AI-native wedge change the evaluation?

It changes what the wedge looks like, not the gates. An AI agent automating a painful, high-volume vertical workflow can be a legitimate wedge an incumbent's older architecture cannot match. But regulated buyers apply extra scrutiny to AI, which adds its own engineering tax.

How do we kill a vertical bet without destroying morale?

Pre-commit the kill criteria at each gate, in writing, before work starts. A stop that fires against agreed criteria reads as the process working, not as a failure. Redeploy the team to core work with a public accounting of what the gate taught you.

FAQ

Which factor should carry the most weight in the scorecard?

Product fit versus net-new engineering, at roughly 25%. The engineering tax is the most consistently underestimated line item in vertical expansion and the most common cause of quiet failure — a two-quarter build that becomes an eighteen-month slog. Weighting it heaviest is a deliberate forcing function that makes the team score the most dangerous factor most rigorously.

What are the fastest disqualifiers?

You cannot name ten logo-quality target accounts, or you have not interviewed five of them. The product gap exceeds twelve engineering-quarters before a credible v1. Regulatory entry consumes more than roughly a million dollars and eighteen months before you can legally sell. Or you have zero organic customers in the vertical after years of horizontal selling.

How much should the first gate cost?

Roughly $50K to $150K over six to ten weeks. That covers the bottoms-up SOM, the scorecard with evidence behind each score, ten customer interviews, an engineering-tax estimate in eng-quarters, a regulatory scoping conversation with counsel, and two or three channel-partner conversations. Most verticals should die here, and that is the gate working correctly.

Why is zero competition a warning sign rather than an opportunity?

Because it usually means the vertical does not buy software the way you assume, the budget does not exist, or the pain is not acute enough to trigger a purchase. Some incumbent presence validates that a real market with real budget exists. An empty field is more often an empty market than an unclaimed one.

How should vertical CAC be modeled in years one and two?

At two to four times your blended core CAC, with sales cycles running 1.5x to 2x steady state. You are paying the new-motion tax: specialist hiring, six-to-nine-month ramps, no reference customers, and marketing that has not compounded. Target a trend toward eighteen-to-twenty-four-month payback at scale, and fund the look-bad period explicitly.

What is the most expensive mistake in vertical expansion?

Skipping to a dedicated team before a wedge pilot has proven anything. It is how organizations spend millions and eighteen months learning what Gate 1's customer interviews would have surfaced for a fraction of the cost. The second most expensive is focus dilution — the core stalling while leadership's attention sat elsewhere.

Sources

• https://www.bvp.com/atlas — Bessemer Venture Partners Atlas, vertical software and cloud benchmarking research
• https://www.sequoiacap.com/article/pmf-framework/ — Sequoia Capital on product-market fit archetypes
• https://a16z.com/the-new-business-of-ai-and-how-its-different-from-traditional-software/ — Andreessen Horowitz on vertical AI and software economics
• https://www.aicpa-cima.com/topic/audit-assurance/audit-and-assurance-greater-than-soc-2 — AICPA SOC 2 Trust Services Criteria overview
• https://www.hhs.gov/hipaa/for-professionals/security/index.html — HHS HIPAA Security Rule guidance
• https://www.finra.org/rules-guidance/rulebooks/finra-rules/4511 — FINRA Rule 4511, books and records requirements
• https://studentprivacy.ed.gov/ — U.S. Department of Education Student Privacy Policy Office, FERPA guidance
• https://www.fedramp.gov/ — FedRAMP program requirements and authorization timelines
• https://openviewpartners.com/blog/ — OpenView SaaS benchmarks and go-to-market research
• https://hbr.org/2016/03/know-your-customers-jobs-to-be-done — Harvard Business Review, Jobs to Be Done framing for buyer requirements

Related on PULSE

• How do I build a bottoms-up TAM model my board will actually believe?
• Should we expand into a new vertical or move upmarket first?
• How do I run a design-partner pilot that genuinely de-risks a strategic bet?
• What does it actually cost to enter a regulated industry?
• How do I model CAC and CAC payback for a brand-new GTM motion?
• When should we build, buy, or partner for a new capability?