Choosing among the best marketing agencies is less about finding the “perfect” partner and more about reducing downside risk before you sign a contract.
The common failure pattern looks like this: a strong pitch deck, a charismatic strategist, a retainer agreement, then three months later you realize you bought activity, not outcomes.
This guide is a pre-signature question bank designed for decision-stage buyers (founders, marketing leaders, and operators). Use it to pressure-test an agency’s strategy, measurement, execution capacity, and commercial terms before you commit.
Start with a one-page “success definition” (or every answer will sound good)
Great agencies can only be evaluated against a clear definition of success. Before you interview anyone, write a one-page brief you can share consistently.
Here’s a simple template you can copy into a doc.
| Item | What to write | Why it matters |
|---|---|---|
| Primary business goal | Revenue target, pipeline target, or retention target | Agencies optimize what you actually reward |
| Primary conversion event | Demo booked, trial started, purchase, qualified lead | Prevents reporting theater |
| Target segment | Industry, company size, use case, geography | Drives channel and message decisions |
| Time horizon | 90 days, 6 months, 12 months | Sets expectations for SEO vs paid |
| Constraints | Budget, internal resources, compliance, dev bandwidth | Avoids “we need engineering for that” surprises |
| Non-goals | What you will not do (TikTok, affiliates, rebrand, etc.) | Helps avoid scope creep |
When an agency answers your questions, you should be able to map every claim back to this brief.
Questions to ask about strategy (fit beats hype)
What do you believe is the highest-leverage growth constraint for us right now, and why?
A serious agency will ask follow-ups and propose a hypothesis tied to your funnel (awareness, conversion, retention). If they jump straight to “you need more content” or “you need paid,” they are selling their default playbook.
Good signs:
- They reference your market, your buying cycle, your price point, and your existing demand.
- They separate symptoms (low traffic) from constraints (weak offer, unclear positioning, poor conversion paths).
Red flags:
- They claim certainty without access to data.
- They over-index on vanity channels because they are trendy.
Which channel(s) would you not prioritize first, and what evidence would change your mind?
This question forces tradeoffs. The best marketing agencies can articulate a “not now” list and the trigger that would alter priorities.
How do you translate our business model into a funnel plan with stage-specific assets?
Listen for clarity on:
- TOFU demand capture (education, search intent, social distribution)
- MOFU proof and evaluation (case studies, comparisons, webinars, email sequences)
- BOFU conversion (landing pages, pricing/packaging, sales enablement)
If you want a reference point for funnel vocabulary, HubSpot’s overview of the marketing funnel is a useful baseline, even if you don’t use their software: marketing funnel explanation.
Questions to ask about measurement (because attribution is where retainers go to die)
What will you measure weekly vs monthly vs quarterly, and who is responsible for each metric?
Many agencies report monthly because it hides slow feedback loops. You want a cadence that matches the channel.
A practical answer usually includes:
- Weekly: leading indicators (impressions, clicks, spend efficiency, content shipped, technical fixes completed)
- Monthly: conversion rates, qualified leads, pipeline influenced
- Quarterly: CAC trends, retention, LTV, brand lift proxies
How do you define a conversion in GA4, and how do you validate tracking is correct?
You are not testing their GA4 memorization, you are testing whether they treat tracking as a deliverable. Google’s own documentation can help you sanity-check terms in the discussion: GA4 conversions overview.
Follow-up: What’s your plan if we discover we can’t trust the data today?
A credible agency has a “measurement recovery” approach (audit events, fix tagging, align CRM stages, dedupe leads) before promising performance.
How do you handle attribution vs incrementality?
Attribution shows where conversions are credited. Incrementality asks whether marketing caused additional conversions.
You do not need a PhD-level answer, but you do want maturity. Good signs:
- They acknowledge attribution limits (cross-device, privacy, walled gardens).
- They propose experiments (geo tests, holdouts, creative rotation) when stakes are high.
Questions to ask about SEO (to separate real operators from “content factories”)
How do you decide what keywords to target, and how do you map them to content types?
You want to hear “intent-first” thinking, not volume-first thinking.
Strong answers include:
- SERP format checks (are the top results guides, product pages, tools, lists?)
- Clustering (topics, not one-off posts)
- Internal linking and information architecture
You can also ask: Show me one example where you chose not to pursue a high-volume keyword, and why.
What is your link-building philosophy, and what links will you refuse to build?
This is non-negotiable. You are looking for a clear boundary against risky tactics.
Good answers sound like:
- Digital PR, partnerships, original research, and “earned” linkable assets
- Clear rejection of PBNs, paid link networks, and manipulative placements
Google’s documentation is explicit that link spam is a policy violation, which is useful context for this conversation: Google Search Central on link spam.
What does your technical SEO checklist look like, and what can you fix without engineering?
Most SEO roadblocks are operational. An agency should distinguish:
- Items they can execute (titles/meta, internal links, content updates, redirects in CMS)
- Items that require developers (rendering, CWV improvements, structured data, template changes)
How will you prove the content is helping the business, not just rankings?
A solid approach ties SEO pages to actions:
- Assisted conversions and CRM influence (when available)
- Lead quality signals (segment fit, sales acceptance)
- Conversion rate improvements on high-intent pages
If we are a local or service business, how do you approach local SEO content without fluff?
This matters if you serve a geography. A useful test is whether they can point to content that actually helps a local buyer solve a problem.
For example, a local service blog that publishes practical maintenance guidance and pricing expectations can support trust and long-tail search demand. Here’s a relevant reference point: Phoenix appliance repair blog.
(You are not buying appliance repair marketing specifically, you are evaluating whether the agency understands what “useful local content” looks like.)
Questions to ask about paid media (if they manage spend, they must explain the math)
How do you structure accounts and budgets to learn fast without wasting spend?
Look for:
- A testing plan (audiences, creative angles, landing pages)
- A clear definition of “learning budget” vs “scaling budget”
- Negative keyword and search query management (for paid search)
What are your rules for creative testing, and how do you avoid false winners?
Good agencies discuss:
- Minimum sample size logic (even if approximate)
- Controlling variables (one change at a time)
- Creative fatigue and refresh cadence
What happens if performance drops for two weeks, what do you do first?
This reveals their troubleshooting order. You want to hear a sequence like:
- Check tracking and spend delivery
- Check auction and competition shifts
- Check landing page and offer changes
- Check creative fatigue and message match
(Keep this as a discussion, not a rigid procedure.)
Questions to ask about team, process, and communication (the “who actually does the work?” section)
Who will be on the account day-to-day, and what percentage of their time is committed?
Ask for names and roles, not just titles. Many agencies pitch with senior people, then staff you with juniors.
Follow-up: What work is subcontracted?
Subcontracting is not automatically bad, but you should know what is outsourced, and how quality is managed.
What does your first 30 days look like, and what will be delivered by day 30?
If they cannot articulate tangible deliverables (audit, plan, measurement fixes, prioritized roadmap, first tests shipped), you are likely buying “planning” disguised as progress.
How do you handle approvals and feedback loops so nothing gets stuck?
Ask about:
- A single owner on your side
- A single owner on their side
- Where work lives (project management tool, shared doc)
- Review windows and escalation paths
What do you need from us to be successful, and what happens if we cannot provide it?
This surfaces hidden dependencies. A mature agency will say things like, “We need fast creative approvals,” or “We need one engineering slot per sprint,” and explain the consequences if that does not happen.
Questions to ask about proof (case studies are easy to fake, decision logic is harder)
Can you walk me through a recent win and a recent loss?
Most agencies can sell a win. Very few will discuss a loss transparently.
Listen for:
- Clear starting condition
- Actions taken
- What changed in the metrics
- What they learned
Which results were driven by the agency vs internal factors?
A trustworthy answer acknowledges confounders like:
- Product changes
- Seasonality
- Pricing shifts
- Sales team performance
Can we speak to two clients: one long-term and one who left?
You may not always get the “who left” reference, but the reaction matters. Defensiveness is a signal.
Questions to ask about ownership, governance, and risk
Who owns the ad accounts, analytics properties, tag manager, and creative assets?
You should own your accounts. The agency should be granted access.
Ask directly:
- Will we retain admin access at all times?
- Can we export everything if we end the relationship?
How do you handle privacy, consent, and compliance?
Even if you are not in a heavily regulated space, the agency should treat compliance as real.
If you do influencer or affiliate-like promotions, you can reference the FTC’s endorsement guidance to set expectations about disclosures: FTC endorsement guides.
The “contract and commercials” questions that prevent expensive lock-in
This is where great partnerships are often ruined. You are not trying to squeeze the agency, you are trying to align incentives and reduce ambiguity.
| Question | What a fair answer often includes | Watch-outs |
|---|---|---|
| What is included vs out of scope? | A written scope with examples | Vague “support” language that invites surprise invoices |
| What is the contract term and termination clause? | Clear notice period, clean exit steps | Long lock-ins with no performance checkpoints |
| How do you handle strategy changes mid-quarter? | Change control process, re-prioritization rules | “That’s a new project” for every adjustment |
| What are your SLAs for communication? | Response time expectations, escalation path | No defined owner, no operating cadence |
| How do you bill, and what triggers extra fees? | Transparent rate card or project pricing | Undefined “miscellaneous” or vague hourly overages |
| How do you report results and what tools do we see? | Shared dashboards, raw access, annotated insights | PDF reports only, no underlying data access |
A simple pre-sign scorecard (use it to compare agencies consistently)
If you are talking to multiple agencies, don’t rely on gut feel alone. Use a scorecard, even a simple one.
Rate each area from 1 to 5, then compare totals and the “why.”
| Area | What you are scoring | 1 (weak) | 3 (okay) | 5 (strong) |
|---|---|---|---|---|
| Strategy fit | Hypotheses match your funnel and constraints | Generic | Some tailoring | Clear choices and rationale |
| Measurement maturity | Tracking, definitions, validation | Hand-wavy | Basic GA4 | Strong instrumentation plan |
| Execution capacity | Ability to ship work consistently | Unclear | Some proof | Clear team, clear throughput |
| SEO approach | Intent, content quality, technical, links | Volume-driven | Mixed | Intent-led, risk-aware |
| Paid approach | Testing discipline and economics | “Boost posts” vibes | Competent | Clear learning plan and scaling rules |
| Communication | Cadence, ownership, speed | Ad hoc | Monthly calls | Operating system, fast loops |
| Commercial fairness | Terms, exit, scope clarity | Lock-in heavy | Acceptable | Clear, flexible, aligned |
One final question that often reveals the truth
If we sign and six months pass with mediocre results, what would you assume went wrong first?
This flips the frame from selling to diagnosing. Great agencies will talk about:
- Unclear ICP or offer
- Tracking blind spots
- Slow execution due to dependencies
- Creative that does not match intent
- Insufficient budget to buy learning
If they only blame “the market” or “the algorithm,” be careful.
