Human SDR vs AI SDR for UK SMEs: Cost Per Meeting

A 20-person UK B2B SaaS company was paying £52,000 a year for a senior SDR who booked an average of nine qualified meetings per month. At twelve months, the fully loaded cost — salary, employer NI at 15% above the £5,000 secondary threshold (the rate and threshold that applied from April 2025 after the Autumn 2024 Budget), 25 days holiday, laptop amortisation, software licences, and a dead month of ramp at hire — came to £62,500. Nine meetings a month for twelve months: one hundred and eight meetings at £579 each. The month after the SDR resigned, we rebuilt their outbound on Clay, Apollo, and HubSpot Sequences for £1,800 per month. Same ICP, same contact volume, different economics entirely.

A second data point from a different ACV tier: a 35-person UK legal-tech company was running two junior SDRs at £38,000 each — fully loaded to £48,500 per head after employer NI at 15% on earnings above £5,000 and tools — booking fourteen meetings per month combined at £291 per meeting. After migrating to Clay and Apollo cold sequences at £2,400 per month combined and retaining one SDR to handle warm and referral pipeline, meeting volume rose to twenty-four per month at a blended cost of £165 per meeting. Show rate on the human-SDR-handled bookings held at 76%, against 53% on the fully automated ones — a gap that compounds at a £28,000 average deal size when each AE hour has a real cost attached.

That is the calculation this post works through: not "is AI better than humans" — it is not, categorically — but where the numbers diverge, what drives those gaps, and what configuration makes sense for a UK SME building predictable outbound in 2026.

The real cost of a UK SDR: salary, NI, tools, ramp time, and the turnover maths

The £52,000 salary line is the easy part. The rest gets rounded down or forgotten entirely in most business cases.

Employer NI: 15% on earnings above the secondary threshold (£5,000 from April 2025, after the Autumn 2024 Budget raised the rate from 13.8% and cut the threshold from £9,100). On a £52,000 salary that is £7,050 per year — over £1,100 more than the pre-2025 calculation most spreadsheets still use.

Tools: A competent SDR needs a CRM seat (£1,200–£2,400/year on an SME plan), LinkedIn Sales Navigator (£960/year), an email sequencing tool if not using CRM-native sequences (£600–£1,200/year), and dialler access. Budget £3,000–£5,000/year in tools even when they share infrastructure with the broader sales team.

Ramp time: Most UK B2B SDRs hit consistent output around month two. Month one produces two or three meetings while they learn the product and talk track — roughly £4,300 in salary absorbed before targets become meaningful.

Holiday and absence: 25 days statutory plus eight bank holidays equals 33 days out per year — 13% of working time unproductive on meeting volume.

Turnover: LinkedIn Talent Insights puts average UK SDR tenure at 14–18 months. Restart the ramp clock, absorb £3,000–£5,000 in recruitment costs, and the true annualised SDR cycle cost over 18 months climbs to £68,000–£74,000.

Year-one cost per booked meeting: £579 at nine meetings per month.

What AI SDR tools actually do in 2026: Clay, Apollo, Outreach, and HubSpot Sequences compared

These four platforms are not interchangeable — they sit at different points in the outbound workflow and combining them without understanding those boundaries wastes budget fast.

In practice, a cost-effective UK SME stack runs Clay on enrichment and ICP filtering, Apollo on cold email delivery and contact sourcing, and HubSpot Sequences on warm contacts who engage with the cold outreach — the configuration used for the SaaS company above at £1,800/month combined.

Before going live, verify the EU data residency settings on each platform against ICO PECR requirements — standard Apollo and Clay plans default to US-based data processing.

Cost-per-booked-meeting analysis: the honest numbers at 50, 100, and 200 contacts per day

The table below uses a 1.5% meeting rate for AI sequences (industry range 0.8%–2.5% for cold B2B) and 4.5% for a human SDR running 10–12 targeted, researched touches per day — the volume achievable with meaningful per-contact personalisation, not the 80–100 template ceiling. At 10–12 touches per day over 21 working days, 4.5% conversion produces nine meetings per month.

The human SDR cost stays flat because one person has a hard ceiling on daily contact capacity — around 80–100 personalised touches before quality degrades. The AI stack scales cost sub-linearly: tool costs increase modestly while output tracks volume linearly.

At 200 contacts per day, the AI stack books seven times the meetings of a single human SDR at one-sixteenth of the cost per meeting.

The catch belongs here, not a footnote: those 65 AI-booked meetings are colder. No-show rates, qualification dropout, and AE time-burn are all higher — factor a 20–25% uplift in downstream sales cost before treating the headline numbers as the full picture. See our LinkedIn AI SDR case study for what that pipeline quality difference looks like across six months.

Where AI SDRs fail: the scenarios a £52k human beats a £1,200/month tool

Four specific situations where a human SDR consistently out-performs the AI stack on hard meeting and conversion metrics — not relationship warmth, which is subjective.

Inbound intent signals: A prospect who visits your pricing page three times and downloads a whitepaper needs a call within the hour. Most AI stacks require a custom webhook from site analytics into the sequence trigger to act on that signal in real time — until that's built, a human beats the automation on timing every time. The B2B intent data prospecting guide covers how to wire that layer up.

Multi-threaded buying committees: When two contacts on a six-person committee know your CEO from a previous business, a human SDR threads the outreach accordingly. An AI sequence fires the same four-touch cadence to every contact on the list.

Complex objections mid-sequence: "We tried something similar and it failed" requires a human to acknowledge, diagnose the root cause, and reframe. The AI sends email three of four regardless.

Deals above £75k ACV: Buyers at that contract value typically expect a human conversation before agreeing to a meeting. An automated booking link after email four drops show rates by 30–40% compared with the same meeting booked by a human after a brief qualifying call.

Where human SDRs fail: the volume ceiling and consistency problems AI doesn't have

A human SDR's hard ceiling sits at around 80–100 genuinely personalised touches per day before quality degrades into template-pasting. An AI stack running Clay and Apollo does not have that ceiling. At 200 contacts per day, output is as consistent on day 180 as day one.

The consistency problem: Human SDR output varies 30–50% week to week — the Monday after losing a deal, the Friday before a bank holiday. The AI stack sends email three at 09:14 on Tuesday regardless.

The documentation problem: Salesforce research on CRM data quality shows only 64% of sales activities get logged by human reps. The AI stack records every touch, open, click, and reply automatically — giving you the data to iterate on sequence performance.

The handoff problem: When an SDR leaves, their relationship context walks out with them. The SDR-to-AE handoff must be structured so institutional knowledge stays in the CRM regardless of turnover.

Hybrid architecture: how UK SMEs use AI for volume and humans for conversion

The configuration that works for most 10–50-person UK B2B companies is neither full AI nor full human:

ICP filter → Clay enrichment → Apollo cold sequence (100–200/day)
                                            │
                                   [Opens / replies / clicks]
                                            │
                                Human SDR reviews warm signal
                                            │
                            Personalised follow-up or direct call
                                            │
                                Meeting booked → discovery call
                                            │
                              HubSpot deal created → AE handoff

The AI handles cold volume at scale. The human SDR works the 10–20 warm leads per week surfaced by engagement signals — a prospect who opened twice and visited the pricing page is a different conversation from a cold dial. The SDR's time on the phone doubles while administrative overhead drops to near zero.

Building the business case: the break-even calculation for replacing or augmenting your SDR team

The question is not "is AI cheaper?" — it is. The real question is at what deal size and close rate AI-sourced meeting revenue justifies the configuration change. This is the model we run before recommending any change:

{
  "scenario": "augment_existing_sdr_with_ai_stack",
  "human_sdr_baseline": {
    "annual_loaded_cost_gbp": 62500,
    "meetings_per_month": 9,
    "show_rate": 0.85,
    "qualified_after_meeting": 0.65,
    "cost_per_qualified_meeting_gbp": 1250
  },
  "ai_stack_addition": {
    "monthly_cost_gbp": 1800,
    "annual_cost_gbp": 21600,
    "incremental_booked_meetings_per_month": 25,
    "show_rate": 0.62,
    "qualified_after_meeting": 0.42,
    "ai_qualified_meetings_per_year": 78,
    "derivation": "25 booked × 0.62 show rate × 0.42 qualification rate × 12 months = 78",
    "cost_per_qualified_meeting_gbp": 277
  },
  "break_even_model": {
    "close_rate": 0.18,
    "projected_closed_deals_per_year": 14,
    "required_deals_to_cover_ai_cost_at_acv_15000": 2,
    "payback_period_months": 1.7
  }
}

At £15,000 ACV and an 18% close rate, the AI stack pays for itself after fewer than two closed deals. With fourteen projected closed deals per year from seventy-eight qualified meetings, the second deal closes around week seven. For UK B2B SaaS above £10k ACV, payback is typically two to four months from go-live — even accounting for the lower qualification rate of AI-sourced meetings.

What changed in 2025–2026: AI SDR platforms adding real-time personalisation and intent scoring

The biggest shift in the twelve months to July 2026 is intent data moving from bolt-on to core sequence trigger. Apollo's intent score integration with G2, Bombora, and LinkedIn signals (launched late 2025) lets sequences fire off buying intent rather than elapsed time — open and reply rates improve measurably for accounts actively researching.

Clay's AI research agent mode pulls live company news, leadership changes, and funding announcements into per-row enrichment at build time. A personalisation line referencing last week's CEO appointment outperforms a static company description, at scale and automatically.

HubSpot's Breeze AI layer, launched mid-2025, added AI-generated follow-up emails pulling context from prior CRM interactions — closing the warm-sequence gap that previously required a human SDR.

The counterpoint: Salesforce's 2025 State of Sales Report finds most buyers now filter out outreach where the only personalisation is a job title or funding round from a public database — a pattern the report ties to the volume surge in AI-generated prospecting since 2023. For high-ACV deals, a personalisation token referencing a LinkedIn headline often signals automation rather than research. That cost compounds the further up-market you go.

Good / Bad / Ugly: three team configurations and their actual cost-per-meeting

Good — Hybrid (AI volume, human warmth)

One human SDR on warm and inbound leads, Clay and Apollo running cold outreach at 100 contacts per day. Cost per booked meeting: £120–£180. Show rate: 70–75%. Downstream close rates are 5–10 percentage points higher than full-AI-sourced meetings. The human SDR's conversation quality improves and they typically stay 6–8 months longer than in a pure cold-call role. This is the configuration we deploy for most UK SME clients after the first month of AI-only testing.

Bad — Full AI, no human qualification layer

Clay and Apollo running cold sequences with an automated booking link sent inside the sequence. Cost per booked meeting: £35–£75. Show rate: 50–55%. No qualification happens between sequence completion and calendar invitation — AEs spend 35–40% of call time on misaligned prospects. Net cost per closed deal often exceeds the human SDR model once downstream time burn is costed in.

Ugly — Full human team at scale, no AI tooling

Four SDRs, no AI tooling beyond a basic CRM and LinkedIn. Cost per booked meeting: £480–£600. Consistency varies 30–50% week to week. When one person leaves — and one always does, usually around month fifteen — output drops 25% for six weeks while the replacement ramps. The most expensive, least predictable configuration of the three, with the highest single-point-of-failure risk per pound spent.

The clearest decision rule from this analysis: if your ACV is below £25k and you are booking fewer than fifteen meetings a month from outbound, the AI-first stack recovers its configuration cost inside sixty days; above £50k ACV, lead with the hybrid model and keep a human SDR on warm pipeline from day one — the show-rate gap alone (76% human-handled versus 53% fully automated in the legal-tech example above) justifies the additional salary line before you factor in downstream close-rate differences. The multi-channel outbound sequence guide covers how to structure the hybrid model across email, LinkedIn, and phone once the AI tooling is in place.