A Scorecard for Choosing a Marketing Agency (and Firing the Wrong One)
Most agency selection processes reward the best pitch deck, not the best fit. By the time you notice the agency you hired can't actually do the work described in their proposal, you're three months and a retainer deep with a contract that's hard to unwind.
This is a scorecard-based approach to picking an agency, built around what predicts a good working relationship rather than who presented the most polished slides.
What are you actually buying before you send an RFP?
Write down the specific deliverable and the specific problem before you send a single RFP. "Help with marketing" produces vague proposals from agencies telling you what they think you want to hear. Say your real goal is a rebuilt paid search account structure with a lower cost per lead within two quarters: naming that specific outcome produces proposals you can actually compare against each other and against your own numbers later, instead of a pile of generic capability decks.
Decide up front whether you need a generalist retainer or a specialist for one function, such as paid media or lifecycle email. Generalist agencies are rarely excellent at everything on their service list, and the function that matters most to you is usually not their strongest one. If you can't name the deliverable in one sentence, you're not ready to send the RFP yet, no matter how urgent the need feels.
The scorecard: five criteria that predict a good relationship
Score each finalist against the same five criteria instead of your gut reaction to the pitch:
- Relevant case studies with numbers you can verify, not just logos
- The actual team who will work on your account, not the partner who ran the pitch
- Reporting cadence and what's in the report, checked against a sample
- Contract terms: notice period, ownership of creative and data, and exit cost
- References from clients in a similar size range, contacted directly by you
Weight these based on what matters most for your situation. A startup evaluating agencies for the first time should weight references and team access highest, since those are the hardest things to fix after signing. Write the weights down before you see any proposals, or you'll unconsciously adjust them to favor whichever pitch impressed you most in the room.
Which questions separate a strong pitch from a strong agency?
Ask each finalist to walk through a real campaign that underperformed, not just their wins. How they diagnose and communicate a miss tells you more about the relationship than any case study of a success. Ask who specifically will be on your account day to day, and ask to speak with that person directly rather than only the salesperson who ran the pitch.
Ask what they'd do differently if your budget were cut by a third next quarter. The answer reveals whether they think in terms of your goals or their retainer. An agency that immediately names which tactics they'd cut first, and why, is thinking like an operator. One that gets defensive or vague is telling you something about how a real budget conversation will go later.
Red flags in the proposal itself
Watch for proposals that promise specific results (a percentage lift, a guaranteed ranking) without qualifying them against your starting point, since no agency can honestly guarantee outcomes they don't fully control. Watch for pricing based on ad spend percentage with no cap, which can quietly reward the agency for spending more of your budget rather than spending it well.
A proposal that reuses language you recognize from the agency's own website, with your company name inserted, suggests the same is happening to every other prospect they're pitching. Also watch the org chart in the proposal: if the named strategist has six other named accounts listed in their bio or on the agency's site, ask directly how much time per week you're actually getting, since "dedicated strategist" and "available strategist" aren't the same commitment.
Setting a 90-day review before you sign a full year
Push back on any contract that locks you in for 12 months with no formal review point. A 90-day check-in with specific, agreed metrics gives both sides a clean off-ramp if the fit isn't right, and a good agency won't resist this since they're confident they'll pass it.
Put the review criteria in writing before work starts, not after. Retroactively deciding what "good" looks like after three months of underwhelming results turns an honest conversation into a negotiation over whose fault it was. Tie the review to the same specific deliverable you wrote down before the RFP went out, not to a new set of goals the agency proposes once the original ones look hard to hit.
What Good Looks Like
Good agency selection means a written brief with a specific deliverable, a scorecard applied consistently across finalists, direct references you contacted yourself, and a 90-day review point built into the contract before work starts.
Building The Capability (5-Stage Skill Ladder)
How to Get Started
Frequently Asked Questions
How many agencies should we get proposals from?
Three to five is usually enough to see real variation in approach and price without turning the process into a part-time job. Fewer than three and you can't tell if a proposal is genuinely strong or just the only option you saw.
Should we ask agencies to do free work as part of the pitch?
Asking for a paid, scoped audit or a short paid pilot works better than free spec work. Free work attracts agencies willing to work for nothing, which isn't the trait you want, and it doesn't test how they perform once money is actually on the line.
What's a reasonable notice period to negotiate into the contract?
30 days is standard and fair to both sides. Anything longer than 60 days should raise questions about why the agency needs that much runway, since it usually means unwinding is harder than they're letting on.
About the numbers
This guide doesn't quote a sourced benchmark. Figures in it are estimates or general guidance, so check them against your own numbers.
Related Guides
Lifecycle Email Conversion Benchmarks: What 'Good' Actually Means
Generic conversion benchmarks don't fit lifecycle email. Here's how to set a realistic target for onboarding, post-purchase, and win-back flows.
Writing an Onboarding Email Sequence New Users Actually Read
How to structure a post-signup nurture sequence around what a new user needs to do next, instead of a fixed schedule of feature announcements.
Four Safeguards That Catch Churn Risk Before It's Too Late
Churn prevention often fails quietly, not dramatically. Four production safeguards that catch at-risk customers before a cancellation request arrives.
The SEO Checklist for a Company That Isn't a Content Team
A practical checklist for improving organic search visibility when marketing is one or two people, with the mistakes that waste the most time.
Content Syndication: Which Channels Are Worth the Setup Time
A comparison of the main ways to syndicate content beyond your own site, including partner republishing, industry newsletters, and paid distribution.
Building an Executive Dashboard Leadership Will Actually Open
Most marketing dashboards get built once and ignored. A practical approach to designing one that leadership actually opens before a meeting.