How do I segment ICP for a $10M ARR mid-market SaaS in 2027?

What ICP segmentation actually is at the $10M ARR mark

At $10M ARR you are past the point where a founder's gut can allocate go-to-market capital, and not yet at the point where you can afford to be wrong about it. ICP segmentation is the system that decides, for every one of the tens of thousands of companies in your addressable universe, how much of your finite selling capacity each one deserves. It is a capital-allocation mechanism dressed up as a marketing artifact, and treating it as the latter is why most one-pagers rot in a Notion doc.

The first discipline is vocabulary, because four terms get used interchangeably and each conflation costs money. Total Addressable Market is the entire universe of companies that could theoretically buy if you had infinite resources — for a mid-market SaaS product this typically runs from tens of thousands to a few hundred thousand accounts globally, and it belongs in board decks, not territory plans. Serviceable Addressable Market is the slice you can actually reach with your current motion, geography, and packaging — usually single-digit thousands to low tens of thousands. Ideal Customer Profile is the much narrower subset of SAM where your product creates outsized value, your motion converts efficiently, and the unit economics work — for most mid-market SaaS companies this lands somewhere between 800 and 8,000 accounts. Buyer persona is the set of individual humans inside an ICP account: typically three to eight people across a VP of the function, a director, a hands-on operator, a finance approver, and an IT or security reviewer.

You can hear the conflation in the hallway. "Our ICP is VPs of Sales at software companies" is a persona, not an ICP. "Our ICP is anyone on Salesforce" is a technographic filter. "Our ICP is the Fortune 5000" is a TAM slice. A complete definition stacks all four: the firmographic shape of the account, the technographic signals in their stack, the accessibility of the buying committee, and the behavioral evidence that they are in-market now. If your one-pager is missing any of the four, it is not finished.

The reason $10M ARR is the specific inflection point is arithmetic. At $2M ARR you have one to three AEs and a founder closing the top twenty deals personally; intuition scales fine. At $10M ARR you typically carry eight to fifteen AEs, ten to twenty SDRs, four to eight marketing channels, six to twelve CS people, and your first or second RevOps hire — with inbound, outbound, product-led, and channel all running at once. Point that many people at "TAM-as-ICP" and a large fraction of go-to-market spend lands on accounts that will never convert at an acceptable customer acquisition cost. Worse, the waste compounds: a wasted $10K in paid spend is not just $10K, it is the AE hours, the SDR hours, the onboarding cost when that account churns at month twelve, and the negative word-of-mouth inside their peer network.

There is a second, less obvious forcing function. Series C diligence stress-tests segmentation discipline much harder than Series B did. Investors will ask for win rate by segment, CAC payback by cohort, and net revenue retention by tier. A company that answers "we sell to mid-market" without a defensible tier structure gets pattern-matched as pre-product-market-fit and marked down accordingly. The segmentation work is therefore both an operating system and a fundraising artifact.

The step-by-step process for building the three-layer model

Build the layers in order — firmographic first, because it is the most stable and the most enrichable; technographic second; behavioral last, because it decays fastest and is worthless without the other two to anchor it.

Step one: define the firmographic spine. Five dimensions do the work. Employee count is the single highest-correlation predictor of deal size and cycle length for mid-market SaaS; the usable band is roughly 50 to 2,000 employees, and inside it there are three meaningfully different sub-segments. Early mid-market (50-200 employees) is founder-led buying with short cycles and annual contract values in the $15K-$40K range. Core mid-market (201-1,000) buys through directors and VPs at $40K-$120K. Upper mid-market (1,001-2,000) behaves nearly like enterprise: six-to-nine-month cycles, security review, $120K-$300K contracts. Pick one primary and one secondary sub-segment. Running all three fragments your messaging, your pricing, and your AE ramp simultaneously.

Revenue band is the second dimension — roughly $5M to $1B for mid-market — but treat it as directional rather than a hard filter, because revenue lags headcount by twelve to eighteen months in fast-growing companies. Cross-reference it against funding stage instead. Industry is third: pick three to seven verticals where your product creates disproportionate value, because at $10M ARR you cannot resource vertical playbooks for twenty. Use two-digit NAICS codes for filtering and four-digit for sub-vertical plays. Geography is fourth: US-only is the right default for most US-based companies at this stage, with international expansion typically waiting until $25M-$40M ARR unless product-led signup data surfaced organic demand abroad; inside the US, concentrating on four to six metros makes field events, dinners, and AE travel economical. Funding stage is fifth, and the sweet spot is usually Series B through Series D — companies with real budget, growing fast enough to buy new tooling, and not yet running enterprise procurement.

A finished firmographic definition should read like a query, not a sentiment: US-based software companies, 201-1,000 employees, $25M-$150M revenue, Series B through D, headquartered in the top eight metros, growing headcount 40%+ year over year, with a director of the relevant function hired in the last eighteen months. That is specific, enrichable, and scoreable — and it typically resolves to a couple thousand accounts, which a fifteen-person go-to-market team can genuinely work.

Step two: layer the technographic signals. What a prospect already runs tells you four things: whether you integrate, how mature their operation is, what their procurement DNA looks like, and who you are displacing. Five axes carry most of the signal. CRM is the most predictive for B2B SaaS — Salesforce shops skew larger, carry dedicated RevOps teams, run procurement gates, and close slower at higher contract values; HubSpot shops sit squarely in the mid-market band, run marketing-led motions, and often close in 30-60 days; Pipedrive, Close, and Copper shops are early mid-market with a practical ceiling around $40K; no CRM at all usually means disqualify unless you are the CRM. Cloud infrastructure is second: AWS-native companies tend to be the easiest sale for API-first products, Azure shops tend to want SOC 2 Type II documentation and marketplace billing, and multi-cloud usually signals you have crossed into enterprise complexity. Data warehouse is third and is a strong budget proxy — a company running Snowflake or Databricks has a data team and has already made a commitment to buying modern data tooling at premium prices. Payments and billing infrastructure is fourth. Collaboration tooling is fifth and matters more than people expect: a Slack-native company adopts tools through Slack discovery and integrations, while a Teams-and-Azure company generally has a slower, more centralized path.

For each account, ask three questions and convert the answers into a 0-25 point score: does their stack mean you integrate or that you compete with something they already paid for; does the stack signal budget for your category; and does the stack signal procurement velocity. A high-fit profile for a RevOps product might be Salesforce plus AWS plus Snowflake plus Slack, scoring 22-25. A low-fit profile with the same headcount — Pipedrive plus Azure plus Teams — might score 8-12. Same firmographic shape, entirely different buying behavior.

Step three: add behavioral readiness. Firmographics and technographics tell you who is plausible. Behavior tells you who is ready now, and this is where most companies at this revenue underinvest. Six signal classes matter. Third-party intent networks measure whether an account is consuming content in your category above their own baseline. Review-site intent — someone researching your category on a comparison site — tends to be the strongest near-purchase signal available and converts to meetings at a large multiple of cold baseline. Product-led signals matter enormously if you have a free tier: the pattern worth routing to sales is a free user from a target account who invites three or more teammates, connects a core integration, and builds something real inside the first two weeks. Website behavior — repeat visits from a deanonymized account, pricing-page views, calculator usage — is weaker but essentially free if you already run reverse-IP resolution. Hiring signals are underrated: a target account posting a VP or director role in your functional area within the last 90 days opens a window where new tooling gets bought. Funding and executive-change triggers open similar windows, typically 60-180 days after a raise and 90-180 days after a new CRO or VP starts; the strongest variant of all is when someone who used your product at a previous employer takes a senior role at a target account, which is close to a guaranteed conversation if you reach out inside 30 days.

Compose those into a 35-point rubric with explicit weights — for example, ten points for a high intent surge, eight for category research on a review site, seven for repeated pricing-page visits inside 30 days, five for product activation milestones, three for a relevant hire in 90 days, two for a recent funding announcement — capped at 35. Then decay the behavioral component by three to five points per month when no new signal arrives, because an intent spike from last quarter is not evidence of anything today.

Step four: assemble the composite score and cut the tiers. The 40/25/35 split is the workable default for mid-market. Inside the 40 firmographic points, weight employee count heaviest (roughly 12), then industry (10), revenue (8), geography (6), and funding stage (4). Inside the 25 technographic points, weight CRM heaviest (8), then cloud (6), warehouse (5), and the remaining tools at 3 each. Cut tiers at roughly 85+ for Ideal-Plus, 65-84 for Ideal, 45-64 for Stretch, 25-44 for Soft-No, and below 25 — or any hard disqualifier — for Hard-No. Recompute weekly. Refresh the enrichment inputs monthly, since headcount, funding, and stack all move.

One important distinction: lead scoring and account scoring are different systems, and at this stage account scoring matters more. Buying committees run three to eight people, and account-level signal aggregates far better than any individual contact's email opens. Score the company; route the humans.

What the five tiers get, and what the whole system costs

A tier that does not change anyone's behavior is a label. The point of segmentation is that each tier receives a materially different allocation of the three scarce resources: AE and SDR hours, marketing dollars, and CS attention.

Tier 1, Ideal-Plus — roughly 50 to 150 accounts. These are the accounts where your product creates the most value and that also unlock peer buying through reference value. They get a dedicated AE, executive sponsorship at the CRO level, custom content, one-to-one account-based plays, pricing flexibility, white-glove onboarding, and a rigorous qualification framework applied to every opportunity. You should tolerate 24-30 month CAC payback here because the retention and expansion profile justifies it. Expect win rates in the 40-55% range and cycles of 60-120 days.

Tier 2, Ideal — roughly 800 to 1,500 accounts. This is the volume core of the business. Full-funnel marketing, AE and SDR pod coverage at roughly 80-150 accounts per AE, one-to-few account-based plays organized by sub-segment, standard pricing, standard onboarding. Target 12-18 month CAC payback, 25-35% win rates, and 30-60 day cycles.

Tier 3, Stretch — roughly 3,000 to 8,000 accounts. These fit the firmographic shape but lack technographic strength or current behavioral readiness. Inbound only. No outbound spend, no assigned AE unless the account self-qualifies through product usage or an inbound request. Nurture sequences and self-serve entry. When they do surface on their own, win rates land in the 12-20% range.

Tier 4, Soft-No — 10,000+ accounts. They match some criteria but fall outside the sweet spot: too small, too large, wrong vertical, wrong stack. Product-led only, no human touch, automated nurture, and a referral path to a partner where one exists. They convert at 5-10% when they convert at all, and they churn materially faster than Ideal cohorts.

Tier 5, Hard-No. Actively disqualified: sub-scale, prohibited industries under your acceptable-use policy, competitor subsidiaries. Decline politely, spend nothing, and route to a partner if a referral fee exists.

The economic case for enforcing this is the spread between the top and bottom of the range: roughly 4-8x on win rate, 3-4x on cycle compression, 2-4x on first-year retention, and 1.5-2x on expansion. Compounded, a dollar spent against Tier 1 returns dramatically more lifetime value than the same dollar against Tier 4. The system is not primarily a prediction engine — it is a rationing engine.

The tooling budget. A realistic all-in stack at $10M ARR runs roughly $95K-$240K per year across six categories, bought in a specific order. First, enrichment and contact data ($30K-$60K) — this is foundational and everything else depends on it. Second, lead-to-account routing ($15K-$30K), which is what actually turns the tier field into behavior by matching inbound leads to accounts and enforcing assignment rules and SLAs. Third, a base intent layer ($30K-$50K). Fourth, native account scoring inside your CRM, which on modern platforms is either included in higher tiers or a per-seat add-on. Fifth, partner co-sell account mapping ($10K-$30K), which surfaces which of your partners' customers match your profile and reliably lifts conversion on warm introductions. Sixth — and this one waits until roughly $15M-$20M ARR — a full account-based platform with predictive scoring and a demand-side ad network ($60K-$120K), which only pays back once you have a real account-based motion to run through it.

The cost line most teams forget. A $120K annual stack requires roughly half to one full-time RevOps person to operate well: routing rule maintenance, enrichment hygiene, intent category curation, score-model tuning, tier cutoff calibration. At fully loaded mid-market RevOps compensation that adds another $90K-$180K, which makes the honest first-year cost of the segmentation system closer to $185K-$340K. Budget it explicitly in the Series B or C plan or you will own expensive shelfware.

What not to buy yet. A customer data platform adds a large annual line and rarely pays back before you have multi-channel personalization at scale — wait. Revenue intelligence tools are genuinely valuable but they are sales-execution tools, not segmentation tools; budget them separately so they do not eat this line. A custom in-house scoring model built on your warehouse with dbt and reverse-ETL is the right end state, but at $10M ARR the marginal lift over vendor scoring is small and the build is two to four months of data engineering; revisit it at $25M+ ARR when proprietary in-product signals actually differentiate the model. And do not run three intent vendors at once — pick one heavy and one light.

Timelines. Expect four to six weeks to write and validate the firmographic and technographic definitions from your closed-won data, two to four weeks to stand up enrichment and get field completeness above threshold, two to three weeks to build and test routing rules in a sandbox, and one full quarter of running the score in shadow mode — computing it, not acting on it — before you let it govern territory and spend. Skipping shadow mode is how you hand your best AE a territory built on a bad model.

Where teams get segmentation wrong

Fitting the profile to past wins instead of forward-looking signals. This is the single most common failure at this revenue. Your existing customer base reflects the motion that got you to $2M ARR — heavily founder-network biased, geographically concentrated, and skewed toward whoever was easiest to reach in year one. That is not the motion that gets you to $50M. Use closed-won analysis as one input, weighted against forward-looking signals: where is category demand growing, which sub-segments show the strongest expansion, which stacks are gaining share. A profile reverse-engineered purely from history is a profile optimized for a market that no longer exists.

Building on firmographics alone. A 500-person company running legacy on-premise software may be a hard no; a 200-person company on a modern cloud stack that just hired three people into the relevant function may be your best account this quarter. Headcount and revenue are necessary but not sufficient. Teams that skip the technographic and behavioral layers end up chasing companies that look right on paper and are not buying anything.

AEs poaching Stretch accounts that show product signal. This one is quiet and expensive. A Tier 3 account activates in your free tier, an AE sees the signal, and reaches out — collapsing a self-serve conversion that would have happened at near-zero acquisition cost into an expensive assisted sale. Write the routing rule explicitly: product-surfaced Tier 3 stays product-led unless projected contract value clears a stated threshold, typically around $40K.

Marketing buying Tier 4 inventory because the lookalike matches. Ad platform audience-building tools will happily build a lookalike from your firmographic criteria and serve it to companies you have explicitly deprioritized. Exclude Tier 4 and Tier 5 from paid targeting at the audience level, not just in reporting.

Spreading CS across every tier equally. A six-to-twelve-person CS team cannot run white-glove motions on 3,000 Tier 2 accounts. Ratio the attention deliberately — roughly 5x to Tier 1, 1x to Tier 2, 0.2x to Tier 3 — and put that ratio in writing, because without it the loudest account wins regardless of tier.

Letting the quarterly tier review get eaten by forecast. The review is always the first thing cut when the number is at risk. Pre-block thirty minutes in the quarterly cadence specifically for tier promotion and demotion, with the CRO and RevOps in the room, and treat it as non-negotiable.

Drifting upmarket without forking the motion. The most expensive version of getting this wrong is stretching the existing profile upward to chase larger deals — adding 1,000+ employee organizations across unfamiliar verticals — without building a separate go-to-market motion to serve them. The buyer changes from a team lead to a CIO and procurement, the product often is not built for top-down deployment, and cycles lengthen while win rates hold flat, which reads as "we just need more pipeline" until the cohort data catches up. Three warning signs surface 12-24 months before the financials do: win rate stays flat while cycle length grows 30%+; net revenue retention diverges sharply between the original cohort and the new one; and CAC payback diverges the same way, with the original profile paying back in 12-18 months while the expansion cohort takes 30-45. RevOps has to surface the tier-level breakdown well before the blended number moves, because by the time the blend deteriorates you are four to eight quarters into the mistake. The correct move when you want to go upmarket is to fork: separate AEs, separate process, separate packaging, separate forecast line — while leaving the mid-market motion intact.

Formalizing too early. Below roughly $5M ARR this entire apparatus is premature and can harm growth. With thirty customers you do not have enough closed-won data for the correlations to mean anything — you need on the order of 100 closed-won deals across diverse account shapes before the layers are statistically meaningful. Founder selling still dominates, and imposing a five-tier system on a two-person sales team is process theater. Worse, narrowing early kills the discovery that tells you what the real profile is. Write a one-page hypothesis, refresh it quarterly, and wait for the triggers: AE number five, RevOps hire number one, marketing spend past roughly $1.5M annually, or Series B/C diligence pressure on CAC payback.

Decision framework: choosing the model that fits your motion

Not every company should run the same segmentation shape. Three variables determine which model applies: how much of your revenue lands self-serve, what your median contract value is, and how mature your closed-won dataset is.

If more than roughly 70% of revenue is product-led, invert the weights. Behavioral signal should carry 55-60 of the 100 points, firmographics 25, technographics 15. In a product-led business you identify the profile from in-product activation patterns rather than from enrichment data — usage tells you the truth and firmographics only tell you the shape. The classic bottom-up developer and engineer-buyer motions work this way: let usage surface qualified accounts first, then apply firmographic discipline once the funnel has produced enough volume to segment meaningfully.

If median contract value exceeds roughly $300K with nine-to-fifteen-month cycles, the five-tier system collapses into two: a hand-curated strategic top 50 and a named-account long tail. Continuous scoring is replaced by written account plans owned by the AE and reviewed quarterly by the CRO. Behavioral signal still helps, but the firmographic and technographic layers degrade in usefulness because every account at that size is idiosyncratic.

If you are squarely mid-market with a mixed inbound and outbound motion, run the full 40/25/35 model with five tiers as described above. This is the default and it fits the majority of companies at $10M ARR.

On the question of AI scoring versus human-tuned rules, run both in parallel before you trust either. Vendor AI scoring is genuinely better at three things: weighting hundreds of weak signals at once, detecting non-linear interactions a human would never write a rule for, and updating weights as the market moves. It is genuinely worse at three others: explaining to your CRO why an account scored 87, handling categorical sparsity when your first twenty deals in a new vertical are too few to learn from, and avoiding feedback loops — the model learns from accounts your AEs worked and never sees the accounts they skipped, which entrenches whatever bias already exists. The workable blend at $10M ARR is roughly 70% human-tuned rules with 30% vendor AI used for tiebreaks and signal discovery, shifting toward AI-dominant by $50M ARR when the data volume justifies the trust. Run the human model in shadow for two to three quarters after adopting vendor scoring before you deprecate it.

The refresh cadence that keeps the model honest

Segmentation is an operating discipline on four clocks, not a document.

Weekly, recompute account scores against fresh signal and let accounts promote and demote across tiers automatically. Owner: RevOps and marketing ops. This is the cadence that was impossible a few years ago and is now table stakes — static quarterly segments waste the entire behavioral layer, since intent decays inside 30-60 days.

Monthly, refresh the enrichment inputs: headcount, funding, stack. Review intent-surged accounts that no one actioned, because an unworked surge is a process failure, not a data failure.

Quarterly, run the full review inside the forecast cadence, anchored on two exercises. The first is the best-50 reverse-engineer: rank your top 50 customers not by contract value alone but by contract value multiplied by second-year net retention, satisfaction tier, and expansion velocity. Enrich each with *current* firmographic data rather than signed-date data so you are looking at what they are now. For each, capture the persona who actually signed, the trigger that opened the deal (funding, executive change, an incumbent contract expiring, an internal mandate), and time-to-close. Cluster all 50 across the three signal layers and find the two or three clusters that contain thirty-plus of them. Compare those clusters against your written profile. Drift shows up here twelve to twenty-four months before it shows up in revenue.

The second exercise is the one most leaders skip because it is uncomfortable: the worst-50 reverse-engineer. Pull the fifty customers with the worst first-year net retention, the heaviest support burden, or the earliest churn. Cluster their signatures the same way. Those clusters become your Hard-No rules and your Soft-No deprioritization rules — and this is where the largest acquisition-cost savings actually live, because every category you remove from the top of the funnel reduces cost across everything that remains.

Annually, recommit the whole model. Between $10M and $50M ARR, expect two to three material revisions. The triggers that force one early: win-rate drift above 30% on a sub-segment, a new product line that changes the buying committee, a pricing change, or a genuine expansion of the addressable universe through a new geography or vertical.

One structural note on existing customers: include them, but as a separate expansion segment rather than mixed into new-business tiers. Your installed base often has a different firmographic and technographic profile than your ideal new logos, and scoring them on acquisition criteria produces nonsense. Segment expansion accounts on product adoption depth, satisfaction, and headroom instead. The same logic applies to multi-product companies — tag accounts with product-level tiers rather than averaging a single blended tier across products, because averaging destroys exactly the signal you built the system to capture.

Finally, a data-quality floor. Before any of this governs real routing decisions, you need roughly 80% completeness on core firmographic fields and roughly 60% on technographic fields across your target accounts. Below that, tier assignments are noise, and noise routed at scale is worse than no routing at all — it gives the organization false confidence in a system that is guessing.

Related questions

How many accounts should be in the Ideal tier?

Roughly 800-1,500 for a mid-market SaaS company at $10M ARR — enough to sustain eight to fifteen AEs at 80-150 accounts each. If Ideal-Plus exceeds about 15% of your serviceable market you are too broad; if Ideal holds fewer than 200 accounts you have capped your pipeline.

Should lead scoring or account scoring drive routing?

Account scoring, at this stage. Mid-market buying committees run three to eight people, so account-level signal aggregates far more reliably than any individual's email opens or form fills. Score the company to set the tier; use lead scoring only to prioritize which contact inside that account to approach first.

How long before a new segmentation model produces measurable results?

Plan on one full quarter running the score in shadow mode, then two to three quarters before win rate and cycle length by tier become statistically readable. CAC payback shifts show up last, typically four to six quarters out, because payback is a trailing measure by construction.

What triggers an off-cycle profile revision?

Four things: win-rate drift above 30% on any sub-segment, a new product line that changes the buying committee, a material pricing change, or entry into a new geography or vertical. Any of these invalidates the weights in your scoring model and warrants an immediate recalibration rather than waiting for the quarterly review.

Can two segmentation archetypes run at once?

Only if you can fund two separate go-to-market organizations — different AEs, different marketing budgets, different forecast lines. The archetypes use different signal weights, sales motions, pricing, and success metrics. At $10M ARR most companies cannot afford both and should pick one and execute it completely.

FAQ

How do I know if my segments are too broad or too narrow?

Check two ratios. If your Ideal-Plus tier holds more than about 15% of your serviceable market, it is too broad and your team will burn hours on low-fit accounts that carry a premium treatment they did not earn. If your Ideal tier holds fewer than about 200 accounts, it is too narrow and you have structurally capped pipeline before the quarter starts. A workable target is 5-10% of serviceable market in Ideal-Plus and 15-25% in Ideal.

What is the most common mistake at this revenue stage?

Relying on firmographics alone without layering technographic and behavioral signals. Headcount and revenue tell you the shape of an account, not its readiness or its fit with your product. A 500-person company on legacy on-premise infrastructure can be a hard no while a 200-person company on a modern cloud stack with a recent functional hire is your best opportunity this quarter. Skipping the intent layer means chasing accounts that look correct and are not buying.

How often should tier assignments change?

Composite scores should recompute weekly, because behavioral signal decays within 30-60 days and stale intent is worse than no intent — it produces confident action on dead leads. Enrichment inputs refresh monthly. The full model review is quarterly, inside the forecast cadence, with the CRO and RevOps present. Annual recommitment covers the strategic question of whether the profile itself is still right.

Should existing customers be scored in the same model?

Include them, but as a separate expansion segment rather than mixed into new-business tiers. Your installed base frequently has a different firmographic and technographic profile than the logos you want next, and scoring them on acquisition criteria produces misleading results. Segment expansion accounts on product adoption depth, satisfaction, and remaining headroom instead.

What data quality do I need before this is trustworthy?

Roughly 80% completeness on core firmographic fields — headcount, revenue, industry — and roughly 60% on technographic fields across your target accounts. Below those thresholds, tiers are too noisy to govern territory assignment or spend allocation, and you will be routing on guesswork while believing you are routing on evidence. Clean the CRM and enrich before you build the tiers, not after.

How do I handle accounts that score differently for different products?

Build product-specific profiles when your products have genuinely distinct buyers. An account can be Ideal for your core product and Stretch for a premium add-on, and that is useful information. Tag each account with product-level tier assignments and route based on which product triggered the engagement. Never average tiers across products — the average is always wrong for both and it destroys the very distinction you built the model to capture.

Sources

1. Pavilion — Ideal Customer Profile resources — Community-sourced ICP definitions and scorecard methodology for go-to-market operators.
2. Bessemer Venture Partners — 10 Laws of Cloud Computing — Foundational SaaS unit-economics and market-fit framework, updated annually.
3. OpenView Partners — SaaS Benchmarks — Annual benchmark data on CAC payback, net revenue retention, and growth by ARR band.
4. For Entrepreneurs — David Skok — Canonical unit-economics, CAC payback, and sales-funnel frameworks for SaaS.
5. Tomasz Tunguz — Data-driven analysis of SaaS growth stages and go-to-market efficiency.
6. Point Nine — Christoph Janz — The "five elephants versus 1,000 rabbits" framing for matching customer size to business model.
7. SaaStr — Practitioner writing on segmentation discipline, drift, and growth stalls between $5M and $100M ARR.
8. Predictable Revenue — Aaron Ross on outbound prospecting, specialization, and target-account selection.
9. MEDDICC — The MEDDIC and MEDDPICC opportunity-qualification frameworks used inside top-tier accounts.
10. U.S. Census Bureau — NAICS — Official industry classification codes used for firmographic filtering.

Related on PULSE

• How should you segment your customers into SMB / Mid-Market / Enterprise tiers?
• How do win-loss interviews refine ICP targeting and segment strategy?
• What is a healthy win rate by segment (SMB / Mid-Market / Enterprise) in 2027?
• Fractional CRO for B2B SaaS startups under $10M ARR
• How do you decide if a part-time revenue leader is right for a bootstrapped $10M ARR company when forecast missed twice in a row?