How to Choose a Marketing Agency: A Practical Guide to Finding the Right Growth Partner

Most buyers start by comparing agencies. That is often one step too early.

A company struggling to generate pipeline needs different help than one struggling to convert pipeline. A company that doesn't understand why performance is poor needs something different than one that knows exactly what to do but lacks execution bandwidth.

This guide covers how to identify what you actually need, test whether an agency can deliver it, and define what it should be accountable for. It is written primarily for B2B and other high-consideration businesses where growth involves longer buying cycles, multiple variables and meaningful acquisition costs.

Start With What You're Actually Buying

Most agency engagements fall into one of four categories:

1. Execution capacity You know what needs to happen. You need more hands.

This is the simplest engagement. You have a working playbook - the strategy, targeting and channels are understood - and you need bandwidth to run it.

2. Specialist capability You know the problem and the channel, but you lack the depth to solve it well internally.

For example: you know Google Ads should work for your business, but your team doesn't have the expertise to make it profitable. You need someone who has solved this specific problem before.

3. Diagnosis Performance is weak or inconsistent, but you don't yet know what's causing it.

This is different from execution. You're not looking for someone to run campaigns. You're looking for someone to figure out why the current approach isn't working - and whether the fix is even a marketing fix.

4. Strategic growth leadership You're not just missing execution. You need someone to determine priorities, challenge assumptions and connect marketing decisions to the wider business.

This is the hardest engagement to evaluate, because you're hiring for judgment rather than a specific deliverable.

These require different partners and different evidence

For execution capacity, you need proof the agency can deliver volume reliably. For specialist work, you need evidence they've solved this specific problem before - not just worked in the same industry. For diagnosis, you need someone who asks hard questions before recommending tactics. For strategic leadership, you need evidence of business judgment.

Not every weak result is a marketing problem

Poor Google Ads performance may actually be an offer problem, a positioning problem, a tracking problem or a sales problem. Plenty of leads but no revenue may be a qualification, sales or economics problem. When companies say "marketing isn't working," they often mean "we're not growing" — and the channel is rarely the actual problem.

If your ICP, positioning, channel economics and acquisition mechanism are already reasonably understood, an external partner can multiply your execution.

If none of those are clear, you're asking the agency to discover the strategy while simultaneously executing it. That's a much harder engagement to get right - and a much easier one to get wrong.

The clearer you are about what's missing, the easier it becomes to determine what kind of partner you need.

Should You Build This Internally or Buy It Externally?

Build internally when the work depends heavily on proprietary knowledge, needs to compound inside the business, is strategically core, or requires constant cross-functional involvement.

Use an external partner when you need specialist expertise you don't have, need it faster than you can hire, the work is temporary or variable, or it doesn't need to become a permanent internal function.

Use a hybrid model when strategy should remain internal but specialist execution can be external, or when you eventually want to absorb the capability.

One agency operator put the distinction this way:

Click to expand

Examples:

• Strong internal marketing leader hitting a paid-search performance ceiling → specialist external capability may make sense, but first confirm the channel is commercially viable before hiring for it.
• Founder-led company with unclear positioning and acquisition economics → hiring a channel executor is probably premature. You need diagnosis first.
• Strong strategy internally but insufficient execution bandwidth → external execution capacity may be the right fit.

Don't choose based on labels

Terms like "full-service agency" or "growth agency" tell you very little. Instead ask: What problem will this partner own? What decisions will they make? What work will they execute? What remains our responsibility?

Once you know what you're buying, the harder question begins: how do you determine whether a particular agency can actually deliver it?

Why Marketing Agencies Are Hard to Evaluate

Before hiring, most of what a buyer can see is case studies, testimonials, client logos, certifications, pitch decks and confident sales conversations. These signals are easy to display - and easy to optimise for regardless of actual delivery quality.

What the buyer actually needs to know is harder to observe: Can this team diagnose our problem correctly? Will the pitch team actually be involved? How much attention will our account receive? What happens when the first hypothesis fails? Are their past results genuinely transferable?

The agency naturally knows more about its own team, workload and operating quality than the buyer does before signing — a dynamic researchers describe as characteristic of outsourced marketing services, where providers have information advantages the client cannot easily verify. The job of due diligence is not to eliminate uncertainty — it is to replace easy-to-display signals with evidence that is harder to fake and more relevant to your situation.

Don't Just Ask for Case Studies. Test Whether the Evidence Transfers.

A case study proves, at most, that an agency participated in a result under particular conditions. The real question: does it demonstrate capability that transfers to your situation?

1. What was the starting point? Ask for absolute context. A 200% increase from 2 leads to 6 tells a different story than 200 to 600. Get the baseline spend, volume, conversion rate and economics.

2. What did the agency actually change? What specific intervention did they make? What part was strategy versus execution? What did the client change at the same time? Understand what produced the improvement - not just that improvement happened.

3. Did the economics improve? A company generating twice as many leads after tripling spend is not necessarily evidence of better performance. Ask whether acquisition efficiency, lead quality and downstream economics improved.

4. Why is this example comparable to us? Industry similarity alone is not enough. Look at business model, ACV, sales cycle, market maturity and starting position. A company in a different industry with similar acquisition economics may be more relevant than one that shares your industry label.

5. Was this result typical? Most showcase case studies cannot answer this. Ask: was this exceptional or representative? Can we speak to a comparable client? What unsuccessful experiments preceded it?

The strongest case study is not the one with the biggest ROAS. It is the one where you understand what produced the result - and whether those conditions resemble yours.

Understand Who Will Actually Run the Engagement

Execution can be delegated. Accountability for judgment cannot be. WFA/VoxComm's agency selection guidance emphasises a people-first, partnership-oriented approach to selection, rather than treating it simply as a comparison of agency credentials.

A lean agency with direct founder oversight may be structurally stronger than a large agency with diffuse accountability. Equally, a larger agency may be superior where the work requires specialist depth or execution capacity. Understand the operating model rather than assume one structure is inherently better.

Four questions matter:

1. Who owns judgment? Who diagnoses the problem, chooses priorities and decides when the approach should change?

2. Who owns execution? Who performs the day-to-day work? Execution does not need to be senior - it needs to be competent and appropriately supervised.

3. How is quality controlled? How are campaigns and recommendations reviewed before mistakes become expensive?

4. Who is accountable when something goes wrong? Is there one clear person responsible for outcomes and course correction?

Red flags: The pitch team cannot explain who owns strategic decisions after signing. Execution is delegated without a review layer. Nobody has clear accountability. The actual delivery model differs materially from what was implied during sales.

Poor communication can mean several different things in practice. One Reddit discussion broke it down like this:

Click to expand

You do not need senior people doing every task. You need the right judgment applied to the right decisions, capable execution, and clear accountability.

Watch What the Agency Does Before It Makes a Recommendation

A strong agency should not prescribe before understanding enough of the problem. But this does not mean it should provide a complete diagnosis before being hired. The depth of discovery should match the mandate.

If you already know paid search works and need specialist execution, discovery may primarily validate economics and account condition. If growth has stalled and you don't know why, recommending any channel immediately should invite more skepticism.

During conversations, observe whether the agency:

• Asks questions before giving answers - Does it investigate economics, funnel and constraints?
• Separates facts from assumptions - Does it say what it knows versus what needs verification?
• Considers more than one explanation - If performance is weak, does it consider offer, positioning, tracking and sales - not just the channel?
• Is willing to say "not yet" or "not us" - If prerequisites are missing, can it say so?

If deeper diagnosis is genuinely required, a paid diagnostic may itself be the right first engagement.

Weak: "You need Google Ads. Here is our package."

Stronger: "Google Ads may be appropriate, but before recommending investment we need to establish your tracking accuracy, unit economics and what's happening after leads enter the CRM."

You are not testing whether the agency can solve your business during a sales call. You are testing whether its recommendations are driven by evidence rather than by the service it sells. The best audits start with the business problem, not the channel — and so should the agency's approach.

Define What the Agency Should Be Accountable For

The right success metric depends on what the agency actually controls.

Activity metrics (impressions, clicks, CTR, CPC) are useful for diagnosis but rarely sufficient as business measures. Commercial outcomes (qualified leads, opportunities, CAC, pipeline) are often closer to business value and appropriate for acquisition mandates. Ultimate outcomes (revenue, margin, retention) matter most but the agency may not fully control them.

A metric becomes fairer as the agency has more control over the variables that produce it.

A paid-search agency with no control over pricing, sales or onboarding should not be judged solely on closed revenue - cost per qualified opportunity may be more appropriate. A growth partner involved in ICP, messaging and full-funnel experimentation should carry more responsibility for pipeline outcomes.

But don't let agencies hide behind channel metrics either. If CPC falls and CTR rises while lead quality and pipeline deteriorate, the campaign is not succeeding because platform metrics look better.

Before signing, agree on:

Also agree on: primary success metric, supporting diagnostics, and how disagreements about attribution or lead quality will be resolved.

Good accountability is not about choosing the biggest metric. It is about choosing the most meaningful outcome the agency can reasonably influence - and making the remaining dependencies explicit.

Ask What Happens When the First Plan Doesn't Work

Growth work involves uncertainty. Not every hypothesis will be correct.

Ask agencies directly: "Tell me about a campaign or strategy that did not work as expected. How did you know, and what changed afterward?"

A strong answer includes: what they expected and why, what signals showed it wasn't working, what they decided to do, what they learned, and whether they changed their process based on that learning.

Experiment volume is not the goal. The goal is useful learning per unit of time and money. An agency running twenty unfocused experiments is not necessarily better than one running five disciplined ones.

Red flags: Every failure is blamed on the client or market. The agency cannot describe a meaningful failed hypothesis. Underperforming activity continues because "it needs more time" without a clear decision rule. Reporting describes what happened but not what was learned.

Avoid the opposite extreme. A mature agency should not abandon a sound strategy because of noisy data. Look for disciplined adaptation, not constant reaction.

You are not hiring an agency because it will always be right. You are hiring it partly for what happens when it is wrong: how quickly it notices, how accurately it learns and how intelligently it reallocates.

Evaluate Your Own Readiness Too

Due diligence should run both ways. In a discussion among agency operators, several describe firing clients who couldn't make decisions or wouldn't share the information needed to do good work. Before hiring, identify whether internal constraints could prevent the agency from succeeding.

Is there one clear decision owner? Someone needs authority to approve priorities, resolve disagreements and unblock execution. Multiple stakeholders are normal. Multiple final decision-makers are not.

Can the agency get the information it needs? Analytics, ad accounts, CRM, historical data, sales feedback, customer insights. An agency working from incomplete information cannot make high-confidence decisions.

Can your company act quickly enough? A growth system cannot learn if every change takes three weeks to approve. Decision speed must match the mandate.

Can the downstream business handle the demand? More leads do not help if sales cannot follow up, qualification is undefined or onboarding is broken.

Are the economics realistic? Both sides should understand budget, time horizon, baseline, customer economics and what success looks like before making promises.

Choosing a growth partner is only half the decision. You also need to know whether your organisation can give that partner the access and operating conditions required to do good work.

Make Sure the Commercial Model Rewards the Right Behaviour

Both sides need economics that work. If the agency cannot deliver proper attention at the fee, or the client cannot justify the cost, the relationship breaks.

Percentage-of-spend can be reasonable where workload rises with spend, but rewards increasing spend unless efficiency guardrails exist. Performance fees align incentives only if the outcome is measurable and within the agency's control. Fixed retainers provide predictability, but understand what they fund. Project pricing suits bounded work, not continuous optimisation.

No model is inherently superior. Ask: What behaviour does this pricing reward? Could the agency increase its fee while making our economics worse?

Before signing, clarify:

• Scope: What is included, excluded and considered a change in scope?
• Commitment and exit: Initial term, notice required, exit conditions.
• Ownership: Client retains access to ad accounts, analytics, domains, tracking and reporting history.
• IP: Who owns landing pages, creative, research, code and strategic artefacts?
• Handover: What gets transferred if the relationship ends, in what format, how quickly?

The best commercial model is the one where the agency can do the work properly, the client can justify the economics, incentives do not reward the wrong behaviour, and either side understands what happens if things need to change.

Sometimes You Should Not Hire an Agency at All

An agency is useful when the missing capability is something an external partner can provide. Sometimes the constraint sits elsewhere. Question hiring an agency when:

1. You don't know if acquisition is the problem. The constraint may be conversion, sales, pricing, retention or product-market fit.
2. The offer or positioning is fundamentally uncertain. Scaling channels before you know who buys and why is premature.
3. The downstream system cannot convert additional demand. More leads do not help if sales or onboarding are broken.
4. You need a permanent internal capability. Strategically core work that depends on proprietary knowledge may be better built in-house.
5. The economics cannot support experimentation. An agency cannot manufacture certainty if budget or time horizon make it impossible to learn.

This does not mean "do nothing." The right next step might be a paid diagnostic, customer research, fixing measurement, or a bounded experiment.

A growth partner is valuable when the constraint matches what you are buying. If it does not, even an excellent agency becomes an expensive way to work on the wrong problem.

A Practical Process for Choosing a Growth Partner

Step 1: Define the problem before defining the vendor

Write down what is not working, what is known, and whether the gap is execution, specialist expertise, diagnosis or strategic leadership.

Step 2: Decide what should remain internal and what should be external

Clarify what decisions stay inside the company, what the partner will provide, and what your team still owns.

Step 3: Shortlist using broad signals - but do not mistake them for proof

Referrals, case studies and certifications are reasonable for creating a shortlist, not for making the final decision.

Step 4: Interrogate the evidence

Test whether case studies transfer to your problem — baseline, what changed, economics, whether the result was typical.

Step 5: Evaluate how they think and how the work will run

Observe quality of questions, who owns judgment, who executes, and how work is reviewed.

Step 6: Agree on accountability before agreeing on targets

Define primary metric, supporting diagnostics, what each side controls, and how disagreements will be resolved.

Step 7: Test the commercial and operating model

Clarify pricing, scope, commitment, exit, ownership and handover — and whether the structure could reward behaviour that makes your economics worse.

Step 8: Reduce uncertainty before scaling

Begin with something bounded — a diagnostic, audit or pilot — and scale when the evidence supports it.

The objective is not to remove all uncertainty before hiring. You cannot. The objective is to progressively replace assumptions with better evidence until the remaining uncertainty is reasonable for the size of the decision.

Choose for Evidence, Not Certainty

There is no reliable pre-hire test that proves an agency will grow your business. Certainty before hiring is unrealistic. But you can still make a substantially better decision — by moving from what the agency says about itself to what you can actually observe about how it thinks and works.

A good selection process eliminates obvious mismatches, makes bluffing harder, exposes dependencies and makes course correction possible. The best growth partner is the one whose capabilities match the problem, whose evidence survives scrutiny, whose operating model you understand, and whose incentives make sense for both sides.

You are not trying to remove uncertainty. You are trying to make the remaining uncertainty worth taking.