How Sandbox Media can scale revenue without scaling headcount — and protect what's been built through acquisition.
Prepared for Anthony Colreavy, Chairman — Sandbox Media Pty Ltd
A lean, high-performing agency at capacity
Sandbox Media generates $25M in revenue with ~25 people, placing $150–200M in annual media spend. Revenue per head ($1.25M) is strong. But growth requires more account executives — costing $80–110k each and taking 6–12 months to ramp.
Headcount is the ceiling on revenue
~70% of account team time is consumed by manual, repetitive work: pulling data, formatting reports, reconciling invoices, drafting emails. Each executive can only carry 5–7 accounts before quality degrades. Headcount is the ceiling on growth.
AI reclaims capacity without a hire
AI and automation can reclaim 4–5 hours per person per week — freeing capacity equivalent to 1.5 senior executives. Applied correctly, this absorbs 2–3 additional accounts per year with zero new headcount. On $250k average account revenue: $500–750k per cohort, compounding annually.
The recommendation: begin the first three initiatives within 30 days. Automated reporting, reconciliation, and client comms drafting require no new infrastructure and deliver measurable capacity within weeks. The question is not whether to invest. It is how quickly to begin.
Three lenses on how Sandbox Media operates — what it is, how it makes money, and how it delivers its value every day.
Sandbox sits between businesses that want to advertise and media companies that sell the space. Like a mortgage broker — it doesn't own the inventory, it facilitates the transaction and earns the margin. The advantage it sells is expertise, relationships, and buying scale that clients can't replicate internally.
Two streams: a 10–15% commission on every dollar of media placed (on $150–200M in billings, that's $15–22M); and programmatic trading margin — buying digital inventory at wholesale through DSPs and deploying it at a spread. Revenue scales with billing volume and client retention, not headcount.
Example: Showpo's $3M budget → $375k commission + $60k programmatic margin = $435k from one account. One executive manages this relationship.
Three layers: client service (account directors and executives managing relationships and campaigns), buying and trading (media schedules, programmatic desk, publisher relationships), and a thin operations layer (one financial controller, no dedicated data or tech function).
The defining characteristic: delivery capacity scales linearly with headcount. Adding a client means adding hours. Adding hours means adding people.
Every additional account requires a new hire. That hire costs $80–110k fully loaded and takes 6–12 months before producing value. Revenue can grow without proportional cost — but only until servicing capacity is exhausted. Then it jumps.
3–4 hours per account per week pulling data from ad platforms into formatted client decks. Happens every week, for every client. Across 15 account-facing people, that's 45–60 hours of manual work weekly.
A media schedule takes 1–2 days to build from a client brief from scratch. Bottlenecks response time, slows campaign starts, and limits how many new briefs the team can turn around in a month.
Post-buy invoice matching is Excel-based and error-prone. On $150–200M in managed spend, even small reconciliation errors carry financial exposure. This work is currently distributed across the account team and finance.
One person (Guy Taffs) owns all revenue development. The capacity ceiling on growth is a single individual's available hours. No structural relief, no leverage, no scale without another hire.
The arithmetic: to absorb one additional account worth $300–500k in revenue, Sandbox must hire at $80–110k fully loaded and wait 6–12 months for ramp. The cost of capacity is high. The time to capacity is long. AI changes both.
This is not a cost reduction. It is a capability expansion. Every hour recovered from manual work is an hour available for the work that actually grows the business — deeper relationships, proactive strategy, account expansion, new business support.
Not all work is equally automatable. The framework maps five levels of the value hierarchy against three ownership modes. Reading the levels bottom-up shows where AI can act immediately — and where human judgment remains irreplaceable.
| Level | AI only | Human × AI | Human only |
|---|---|---|---|
| Step | Data pulls, cell formatting, template sends, standard email triggers | — | Final client approval before send |
| Task | Report formatting, email drafting, data extraction, invoice matching | Report narrative, brief analysis, first-cut schedule | Client advisory, negotiation, relationship judgment |
| Process | Post-buy reconciliation, invoice-to-plan matching | Brief-to-plan drafting, programmatic optimisation, campaign pacing | Campaign strategy, channel planning, client onboarding |
| Outcome | Performance dashboards, anomaly detection alerts | Campaign optimisation, QBR preparation, retention analysis | Client relationships, accountability, escalation decisions |
| Strategy | Market research synthesis, competitive data aggregation | Pitch preparation, new business intelligence, competitive briefs | Client strategy, pricing decisions, business development |
| Item | Estimate | Note |
|---|---|---|
| Year-one investment (all-in) | $65–140k | Tooling, integration, change management |
| Revenue uplift (conservative) | $1.5–2.25M | 60–100% of unlocked capacity realised |
| Additional gross margin | $900k–1.35M | 60% margin; no new fixed cost on absorbed accounts |
| Payback period | < 3 months | On conservative revenue assumptions |
| 3-year cumulative impact | > $4M | Freed capacity compounds — reinvested into growth each year |
Prioritised by impact-to-effort ratio. The first three can begin immediately with existing infrastructure — no new platforms, no long lead times.
| Initiative | Framework layer | Time saved | Revenue impact | Effort |
|---|---|---|---|---|
| Automated campaign reporting | Step / Task | 3 hrs × 15/wk | ~$900k uplift | Low |
| Client comms drafting | Step | 1 hr × 15/wk | ~$300k uplift | Low |
| Initiative | Framework layer | Time saved | Revenue impact | Effort |
|---|---|---|---|---|
| Post-buy reconciliation automation | Step / Task | 2 hrs × 15/wk | ~$600k uplift | Low |
| Brief-to-plan acceleration | Task / Process | 2 hrs × 5 planners | ~$300k uplift | Med |
| Initiative | Framework layer | Time saved | Revenue impact | Effort |
|---|---|---|---|---|
| Programmatic optimisation (AI-assisted) | Process / Outcome | Performance uplift | ~$500k+ performance | Med |
| New business intelligence (Guy Taffs) | Task / Process | Guy Taffs × AI | ~$500k new revenue | Med |
The AI initiatives are not just operational improvements. They change the strategic position of the business — what it can win, how it retains clients, and how it competes.
When account executives are freed from 5 hours of manual work per week, that time can be redirected to new business activity. The distinction between "farmer" (managing existing) and "hunter" (developing new) roles becomes viable without additional headcount. Guy Taffs gains a multiplier — AI-powered prospect research, pitch prep, and competitive intelligence effectively makes one revenue person operate like two.
The best new revenue is from clients already in the building. When account executives have more time for proactive strategy — reviewing campaign performance, identifying untapped channels, bringing new ideas — existing clients spend more. More campaigns. More ad spend. More commission. The account relationship shifts from reactive management to proactive partnership, which is both stickier and more lucrative.
Sandbox's advantage is personal service over holding-group scale. AI closes the capability gap without losing that advantage. Automated reporting that rivals what WPP delivers. Programmatic optimisation that matches GroupM's trading desk. Brief turnaround that beats any holding group on speed. The boutique intimacy stays. The operational gap disappears. That's a winning pitch for any mid-market advertiser choosing between Sandbox and a large network.
The gap between agencies talking about AI and agencies deploying it systematically is enormous. That gap is the opportunity.
Industry surveys consistently show 70–80% of media agencies claim to be "using AI." The reality: most are using ChatGPT for occasional copywriting or experimenting with single-use tools. Systematic deployment — AI integrated into core workflows, replacing specific manual processes, measured against time and revenue — is rare.
The difference between "using AI" and "AI-enabled operations" is the difference between having a gym membership and being fit. One is a statement. The other is a structural capability.
Holding groups (WPP, Publicis, IPG) are investing in AI at scale — but their implementations are slow, bureaucratic, and often designed to protect existing revenue structures rather than create new ones. Their AI initiatives typically take 18–36 months to reach the account team.
Independent agencies that move first have a 12–18 month window to establish a genuine operational advantage — better reporting, faster turnaround, more proactive strategy — before the holding groups catch up. That window is open now.
The first mover position for Sandbox: an independent agency that can genuinely say "our account executives spend 70% of their time on strategy, not administration" is a fundamentally different product than what the market currently offers. That becomes a pitch differentiator in new business conversations, a retention argument with existing clients, and a talent attraction story for hiring the best account people. AI is the operational enabler. The market position is the prize.
A decade of senior operations roles in scaling businesses — building the systems, processes, and capability that let companies grow without proportional overhead.
Built for scale without overheadDesigned and deployed operational infrastructure at NEC that supported 3x growth without proportional headcount increase. Selected for a newly created COO-adjacent role; promoted 3 times in 4 years.
AI and automation deploymentBuilt working AI/automation tooling and demonstration environments — including an ad agency client reporting automation demo — for use in prospect conversations. Applied the outcome/task/process/step framework to identify and prioritise automation opportunities.
Systems thinking at scaleConsistent pattern across Microsoft, NEC, Zoomo, and Wesfarmers: selected for newly created roles, trusted with ambiguous mandates, and delivered structural improvements that outlasted the role itself.
Direct relevance to SandboxThe constraint Sandbox faces — revenue growth capped by account team capacity — is structurally identical to constraints solved in previous roles. The difference is the tools available to solve it are now an order of magnitude more powerful.
"The question is not whether to invest.
It is how quickly to begin."
Prepared for Anthony Colreavy, Chairman — Sandbox Media Pty Ltd · April 2026 · Confidential