Consultant vs Agency vs In-House

The choice is not about cost, whatever you have been told

The folk model says agencies are expensive, consultants are the middle option and freelancers are cheap. The only decent published dataset — a survey of 439 search service providers, published in December 2023 and updated in August 2024 — does not support it. Independent consultants reported charging slightly more per month on average than agencies did, not less. Freelancers sat in a different band entirely, far below both. The distribution across all groups is extremely wide, and the modal retainer sits well below the average, meaning the average is dragged up by a minority of large engagements.

Two things follow. If you are choosing a consultant to save money, the market data says you probably are not; what differs is cost structure — no account management layer, no sales layer, no junior execution layer — so you buy fewer people rather than cheaper hours. And in-house is not automatically cheaper either, once benefits, recruiting, onboarding, training, tools and management overhead sit on top of the salary and you count the months before a new hire is productive.

Which means this decision has to be made on something else. Three things, in fact: capacity, accountability and continuity. The rest of this page is about reading which of them is actually binding for you.

What an agency is structurally good at

Agencies exist because some problems are a headcount problem, and no individual can solve a headcount problem however good they are.

Parallel capacity. If you need technical work, content production, digital PR and paid media running at once, an agency can staff all four simultaneously. A single practitioner cannot, and any who says otherwise is planning to subcontract without telling you.

Breadth across specialisms. International annotation work, feed management, developer resource, design and analytics engineering are different skills, and a firm carrying all of them moves between them without a new procurement cycle.

Continuity through absence. Somebody covers illness, leave and departure. Administrative, until the quarter it matters.

Procurement compatibility. Larger organizations frequently have vendor requirements — insurance thresholds, security review, invoicing structures, master service agreements — that an established firm satisfies routinely and an individual may not satisfy at all. This is a real and legitimate constraint, not corporate friction.

Buying power and tooling. Enterprise crawlers, rank tracking at volume and data pipelines are expensive, and a firm amortizes them across clients.

Choose an agency when the work is broad, needs several things at once, must run regardless of any individual, or has to pass procurement. Those are good reasons and none is a compromise.

What an independent consultant is structurally good at

The consultant's advantages are all versions of the same fact: there is only one person, and you know who it is.

Who does the work is knowable in advance. In an agency the pitch team and the delivery team are commonly different people — which is why some firms advertise a minimum experience standard for their consultants rather than naming who you get, and others promise you will speak directly to a senior practitioner. Those promises exist because buyers learned to worry. With one practitioner the question does not arise.

No translation layer. The person diagnosing is the person you talk to, so nothing is lost between the analysis and the account manager's summary of it.

Seniority on every hour. You are not paying a blended rate that averages a principal and three juniors. That is also why the fee is not lower than an agency's.

Fewer incentives to expand scope. An individual has no bench to keep occupied, which makes it easier to say a piece of work is not worth doing.

Choose a consultant when the problem is one of judgment rather than volume — what to do, in what order, whether the last provider's plan was sound, whether a platform decision is about to cost you — and when you have people who can implement once somebody has decided what to implement.

What an in-house hire is structurally good at

In-house is frequently the right answer, and it is the one most under-recommended by everyone who writes about this, for obvious reasons.

Context nobody external can accumulate. An employee knows the margin on each product line, which supplier is unreliable, why that category page exists, and which senior stakeholder will block which change. That knowledge takes an external provider quarters to approach and years to match.

Proximity to implementation. Being in the standup, in the sprint planning, in the corridor. A great many search recommendations die in a queue, and the single most effective intervention is often a person present when priorities are set.

Compounding institutional memory. The knowledge stays when a contract would have ended.

Cost efficiency at sustained volume. If search work is continuous, substantial and central to how the business acquires customers, an internal team is usually more economical than paying external rates indefinitely, and the argument strengthens with the horizon.

The honest costs: hiring takes months and is high variance, since you are evaluating a field with no licensing and considerable noise. New hires ramp slowly — in-house teams generally take longer to get going, while external providers start faster because their people, tools and workflows already exist. A sole internal specialist has no peer review, and skills drift without exposure to other environments.

Choose in-house when search is a permanent, central function, when implementation dependency is your real bottleneck, and when you can support the person properly rather than isolating them.

How each of the three fails

Every model has a characteristic failure, and knowing them is more useful than knowing the benefits.

An agency fails through dilution. The senior person who won the account moves to winning the next one; delivery falls to someone learning on your budget. Reporting becomes activity-shaped, because activity is what a team can reliably produce. Nobody is individually accountable when work is spread across five people and a process. The countermeasures are real: name the delivery lead in the contract, meet them before signing, and ask what proportion of their time you are buying.

A consultant fails through capacity and singularity. One person cannot staff four simultaneous programs, cannot cover their own absence, and becomes a bottleneck the moment your organization wants more throughput than one brain can supply. There is also a real single point of failure. The countermeasures are stating capacity honestly, documenting everything so the work survives the relationship, and being clear early when a piece of work needs a team rather than a person.

In-house fails through isolation and drift. A single specialist with no peer becomes either unchallengeable or unheard, and neither is good. Skills narrow to the systems in front of them. The role gradually becomes ticket-writing for other teams. The countermeasures are external review at intervals, a training budget that survives cost cutting, and a reporting line senior enough that recommendations can reach a roadmap.

Reading which constraint is actually binding

Answer these honestly and the choice usually makes itself.

• Do you know what to do, or not? If not, you have a diagnosis problem, and that is a small purchase from an individual or a specialist team rather than a year's retainer with anyone.
• Can anyone implement? If nothing can ship, buying more advice makes the pile of unshipped advice taller. Buy execution capacity — an agency, a contract developer, or an internal hire with authority.
• How many things need to happen at once? One or two, sequentially: an individual is fine. Four in parallel, permanently: that is a team.
• Is this permanent? A function that will exist in five years justifies building it internally. A twelve-month push does not.
• Does procurement require a firm? Then it requires a firm, and the rest of the analysis is academic.
• What is the cost of a wrong decision? High-stakes, low-frequency decisions — a replatform, a domain change, a market entry — favor concentrated senior judgment. High-volume execution favors capacity.

Hybrids, which is what most mature setups actually are

The three-way framing is a simplification. Companies that have been doing this for a while almost always run a combination, and the combinations are more interesting than the pure cases.

In-house owner plus external specialists is the most common mature arrangement. Someone internal owns the function, holds the context and gets things into the roadmap; specialists are brought in for depth — a migration, an international build, a penalty determination — and leave when it is done.

Agency for execution plus an independent for review addresses a legitimate worry: you are paying a substantial retainer with no way to judge whether the work is good. An occasional independent read of the strategy and the deliverables is a small cost against it. This works only if framed as verification rather than a hunt for failure, and if the agency is told from the start.

Consultant now, in-house later is a deliberate and underused sequence. Bring someone in to establish what the work is, write the specifications and the measurement framework, then hire against a role you now understand — and have them help screen the candidates, which is where most companies get it wrong.

What does not work is two providers with overlapping scopes and no named owner. Ambiguity about accountability does more damage than picking the slightly wrong model.

The uncomfortable summary

This page is on the site of an independent consultant, so it should end by being explicit about where that is the wrong choice.

If you need four workstreams running simultaneously and permanently, hire an agency. If your procurement process requires a firm with a master service agreement and a security review, hire an agency; there is no point admiring an alternative you cannot buy. If search is central to how your business acquires customers and will still be central in five years, build the function internally — it will cost less over that horizon and the knowledge will stay. If your bottleneck is that nothing ever ships, hire someone internal with the authority to get things into a sprint, because no external provider will ever have that authority.

An independent is the right answer to a narrower question: when the binding constraint is knowing what to do rather than having the hands to do it, when the decision is expensive to get wrong, when you want the person who diagnosed the problem to be the person you can call about it, and when you have or can obtain implementation capacity elsewhere.

Those are genuinely different situations. A provider who tells you their model is right for all of them is describing their business rather than your problem.

Frequently Asked Questions

Published 2026-08-11