How to Choose a Marketing Agency: 25 Questions to Ask Before You Sign

Most marketing agencies lose clients in the same four ways, and you can spot every one of them before you sign. The senior person who pitches you hands you to a junior. The plan is a template that worked for someone else. The agency solves half the problem and calls the other half “outside scope”. And the pitch is a lot of talking with nothing built.

We are a boutique growth partner, so we have sat on both sides of this table. Below are the 25 questions we would ask if we were hiring an agency, with what a good answer and a bad answer sound like, and a scoring sheet you can copy into a spreadsheet and use on your whole shortlist.

Quick answer

To choose a marketing agency, ask 25 questions across five areas (who does the work, how they will solve your whole problem, how results are measured, who owns what, and how you pay), score every answer from 0 to 2, and compare totals. Prefer the agency that shows you work before you pay, names the people who will do it, lets you see the numbers in tools you control, and lets you leave with everything you paid for. Under 30 out of 50 means keep looking.

The four ways agencies let clients down, and what to look for in their place.

The four patterns to watch for

1. The handover. You meet a charming senior person in the pitch. After you sign, the work goes to someone two years out of university who has never met your buyer. Businesses that have been burned often say the same thing: “I don’t want to be dumped on a junior.” The fix is a named team and a promise about who is on each call.

2. The cookie-cutter. The agency copies what worked for another company in a different market and accepts marginal results. Good work starts with your buyers’ words, your numbers and your constraints. Ask to see two plans for two different clients. If they look alike, you have your answer.

3. Half the problem. Traffic agencies bring visitors and leave you with a conversion problem. Conversion agencies say traffic is someone else’s job. If the site leaks enquiries, more traffic only wastes more of it. Pick a partner who owns the whole path from a stranger’s first search to a booked call.

4. All talk. Many pitches are long on slides and short on proof. The agencies worth hiring put their money where their mouth is: a free audit, a sample page, a working prototype of your homepage, before you pay anything. If an agency will not show you a sliver of the work first, ask why.

How to use this list

Send it to every agency on your shortlist, ideally three to five, before the call. Ask for written answers, then use the call to push on the weak ones. Score each answer using the sheet at the end. A good agency will enjoy the questions. A bad one will get defensive.

Area 1: Who does the work (questions 1 to 5)

Answer first: you are buying people, so ask who they are, how many clients each one handles, and who you will speak to.

Q1. Who, by name, will work on my account day to day?

• Good: “[Name] writes, [Name] builds, [Name] runs paid. You will meet all three in the first call.” Real names, roles and a short bio each.
• Bad: “Our team of specialists.” No names.

Q2. What does my main contact do each week, and how many other clients do they have?

• Good: A real number (“six clients, each gets a half day a week”) and a clear description of the work.
• Bad: “We scale resources to your needs.”

Q3. Is anything outsourced or sent offshore?

• Good: “Everything is in-house” or an honest list: “We use a freelance photographer for shoots.”
• Bad: Vague answers, or reluctance to say.

Q4. Will the person who pitched me stay on the account?

• Good: A specific commitment: “Our founders are on every monthly call.”
• Bad: “Of course!” with no detail on how often.

Q5. What happens if my main contact leaves?

• Good: A handover process, a second person already briefed, a notice period.
• Bad: “That doesn’t happen here.” (It happens everywhere.)

Common mistake: being impressed by a polished pitch team. Ask to speak to the person who will actually write your copy or build your pages for ten minutes. You learn more from that call than from the whole deck.

Area 2: The plan (questions 6 to 10)

Answer first: a good plan starts with your offer and your buyer, covers traffic and conversion together, and shows you what will be delivered in the first 90 days.

Q6. What would you do first, and why?

• Good: A clear, ordered view: “We would fix the enquiry path first, because traffic to a leaking site is wasted. Then content.” You can disagree with it, which is a good sign.
• Bad: “We would start with a full discovery phase.” Six weeks of meetings.

Q7. Show me two plans you made for different clients. What differed?

• Good: Two plans with different priorities, because the businesses differ.
• Bad: Two near-identical documents with the logo swapped.

Q8. Who owns conversion and follow-up, not just traffic?

• Good: “We look at the whole path: the site, the form, the first reply and the follow-up.”
• Bad: “That’s a web design question” or “Your sales team handles that.”

Q9. What would you need from me, and how many hours a month?

• Good: A precise ask, agreed up front: who they need to speak to, when and for how long.
• Bad: “Whatever you can give,” or a demand for ten hours a week.

Q10. What would you not do for me?

• Good: An honest “no”: “I would not run paid social for your high-ticket service yet. The channel does not suit the price point.”
• Bad: “We can do it all.” A partner who never says no has no strategy.

Spiky view: a channel mismatch kills more budgets than bad creative. Take a hypothetical advisory firm whose engagement is worth 15,000 pounds. Say paid social form fills cost 30 pounds each and one in twenty is a real buyer. That is 600 pounds per qualified lead before any sales time, and the buyer still needs trust before a first call. The same arithmetic can work for a cheaper product with a short sale. These figures are made up to show the sum, not benchmarks. An agency that will take your budget for a channel without asking about your price point is guessing, so ask them to run the sum with your numbers.

Area 3: Measurement and leads (questions 11 to 15)

Answer first: you should be able to see results in tools you control, in terms of enquiries and revenue, with a lead counted the same way every time.

Q11. How do you count a lead?

• Good: A written definition: “A form submission, call or booking from a person in your target market, excluding spam and existing customers.”
• Bad: “Any contact.” Or a number that includes clicks and likes.

Q12. Which metrics will you report, and which will you not?

• Good: “Qualified conversations, cost per qualified conversation, enquiry rate, revenue where we can see it. We do not report followers as a result.”
• Bad: Impressions, reach and followers as headline numbers. We say it plainly: you can have two million of the wrong people.

Q13. Whose tools measure the work?

• Good: “Google Analytics and Search Console in your account, plus a third-party tracker you can log into.” You get access on day one.
• Bad: “We have a proprietary dashboard.” If only they can produce the numbers, they mark their own homework.

Q14. How often will we talk, and what is in the report?

• Good: A weekly short call, a monthly one-page report, a quarterly plan and targets. A report that says what worked, what did not and what changes.
• Bad: “Monthly PDF.” No call. A report that is 30 pages of charts.

Q15. What results should I expect in the first 30 days, and what should I not expect?

• Good: Leading indicators (tracking set up, site fixes shipped, first conversations) and a refusal to promise revenue.
• Bad: A guaranteed number of leads, or “page one in 30 days”. Anyone promising that is making it up.

For a sense of why honest timelines matter: in an Ahrefs study of one million random new pages, 1.74% reached the top 10 of Google within a year, and 6.11% in a second sample of two million pages filtered for real content. Ranking takes time and most pages never get there. An agency that sells you speed is selling you something it cannot control.

Score each area out of 10. Compare agencies side by side.

Area 4: Ownership and exit (questions 16 to 20)

Answer first: you should own every account, domain, file and dataset from day one, and you should be able to leave without losing anything.

Q16. Who owns the website, domain and hosting?

• Good: “You do. Registered in your name, with us added as a user.”
• Bad: “We host it on our servers.” If you leave, the site stays with them.

Q17. Who owns the ad accounts, analytics, email list and CRM data?

• Good: All in your name, with the agency as an admin. Your customer data is yours.
• Bad: The agency’s ad account, or a list held in their system with no export.

Q18. What happens to my content, designs and code if we part ways?

• Good: “Everything we create is yours the day it is paid for. We hand over files and logins in a week.”
• Bad: Licence terms, “ownership transfers after 18 months”, or an exit fee. If the model is “no upfront cost, and you own it after a long term”, read the small print.

Q19. What is the minimum term and the notice period?

• Good: Short and plain, with a reason for the term. A minimum term can be fair if it gives both sides time to do the work and to be held to account.
• Bad: Twelve months locked in, auto-renewal, hidden notice windows.

Q20. Can I speak to two current clients and one who left?

• Good: “Yes, here are three names.” They explain why the third left.
• Bad: Only polished testimonials on the website.

Common mistake: negotiating price before ownership. A discount on a contract that lets the agency keep your domain is no bargain.

Area 5: Money and the pay model (questions 21 to 25)

Answer first: there are three common ways to pay an agency, and each rewards different behaviour. Choose the one that matches what you want.

Fixed, hourly or commission: each pays the agency to behave differently.

Model	Rewards	Risk for you	Suits you when
Fixed retainer	Delivering the agreed scope	Paying for activity, not results	The work is ongoing and the scope is clear
Hourly	Time spent	No reason to be efficient	A small, defined job or advice
Commission or performance share	Results the agency can influence	Disputes about attribution	You can track leads and sales cleanly

Most small firms do best with a fixed scope, a short term, and clear targets, with a performance element only if tracking is solid. If you are comparing a fixed price with an hourly quote, ask for both and divide: the hourly agency will usually spend its hours.

Q21. Which pay model do you offer, and why?

• Good: A reasoned answer and a choice of two options: “Here is a lighter and a fuller version.”
• Bad: One option, take it or leave it.

Q22. What is included in the fee, and what costs extra?

• Good: A written scope listing deliverables, plus a list of exclusions such as ad spend, stock images and software.
• Bad: “Everything you need.” Then surprise invoices.

Q23. Is ad spend separate, and who controls it?

• Good: Spend paid directly by you to the platform, with the agency’s fee separate.
• Bad: A bundled figure where you cannot see what goes to the platform.

Q24. Will you work to a smaller scope if my budget is tight?

• Good: “Yes, start small and focused, and grow as trust builds.” They shrink the scope and keep the quality.
• Bad: A discount on the same promise. Cutting the price without cutting the work means somebody is cutting corners. Ask: “Would you accept lesser quality work in return for a discount?”

Q25. What do you do if the results are not coming?

• Good: “We review at 30 days, change the plan, and if it is not working you can leave on notice.” Plus a view on what they will not do: “We do not guarantee outcomes, because we do not control them.”
• Bad: A guarantee. Guarantees invite disputes and usually come with a clause that voids them.

The scoring sheet

Copy this into a spreadsheet. Score each answer 0 (bad or evasive), 1 (acceptable), or 2 (specific, honest, in writing).

Area	Questions	Max	Agency A	Agency B	Agency C
Who does the work	Q1 to Q5	10
The plan	Q6 to Q10	10
Measurement and leads	Q11 to Q15	10
Ownership and exit	Q16 to Q20	10
Money and pay model	Q21 to Q25	10
Total		50

How to read it:

• 40 to 50: a strong shortlist candidate. Ask for a small paid or free first piece of work.
• 30 to 39: workable. Ask them to fix the weakest area in writing.
• Under 30: keep looking. Any area scored under 4 is a red flag by itself, and ownership under 6 is a deal-breaker.

A quick test that beats any pitch

Before you sign, ask each agency to show you something real. For a website, that might be a sample homepage built around your offer. For search and AI visibility, a short audit of where you appear when buyers ask ChatGPT or Google your questions. For LinkedIn, a profile rewrite. You will learn how they think, how fast they work and whether they listen. Agencies that put their money where their mouth is will usually do it. The ones that will not are telling you something.

Questions people ask

How do I choose a good marketing agency?

Shortlist three to five, send each the same 25 questions in writing, score the answers and ask for a small sample of real work. Prefer the agency that names the people doing the work, shows results in tools you control and lets you own everything you pay for.

What should I ask a marketing agency before hiring?

Ask who does the work and how many clients they carry, how a lead is counted, which tools measure results, who owns the accounts, domain and content, the minimum term and notice period, and what the agency will not do for you.

Should I pay a marketing agency a retainer, hourly or on commission?

A fixed retainer suits ongoing work with a clear scope, hourly suits small defined jobs, and commission suits cases where leads and sales are tracked cleanly. Each rewards a different behaviour, so match the model to the outcome you want.

What are the red flags when hiring a marketing agency?

Guaranteed rankings or leads, no named team, proprietary dashboards only they can see, agency ownership of your website or ad accounts, long lock-in terms, a pitch that cannot be told apart from a competitor’s, and reports built on followers and impressions.

How long should I give a marketing agency before judging results?

Judge the first 30 days on leading indicators such as tracking, fixes shipped and early conversations, and the first 90 days on qualified conversations. Revenue from search and AI visibility builds over months, so be wary of anyone promising results in weeks.

Sources

• Ahrefs, “How Long Does It Take to Rank in Google?”, based on random URL samples from 2023

Want us to do this for you?

We are happy to be scored on this sheet like anyone else. Get in touch and we will send our written answers to all 25. If you would like to see how we think first, read our guides on AI search optimisation and SEO, or our Uniflex case study.