The agency RFP: how to run one that doesn't waste everyone's time.

A good RFP gets you three or four comparable proposals from agencies who understand your business, and it takes weeks — not months — to run. Most RFPs fail because they ask for the wrong things, invite the wrong shops, and reward the best writers instead of the best operators. The fix is to run a tighter process with fewer, sharper questions.

I have sat on both sides of this table. I have written proposals that took forty hours and lost to a two-page email. I have also watched owners pick the flashiest deck and regret it inside a quarter. What follows is how to run an RFP that respects everyone's time and still gives you enough to decide well.

Do you even need an RFP?

You need an RFP when the work is large enough, unclear enough, or important enough that a phone call cannot settle it. For a $2,000-a-month retainer, an RFP is overkill — two calls and a reference check will do. For a $150,000 annual engagement that touches your website, your search visibility, and your reporting, a written process protects you.

The test is money and reversibility. If the contract is cheap and easy to leave, skip the formal process. If it is expensive and hard to unwind, write it down. A formal RFP forces you to say what you actually want, and that clarity is worth more than the proposals it generates.

Before you write anything, decide whether you want an agency at all. The alternative — a senior person who works inside your business a few days a month — solves a different problem. We wrote about the tradeoff in Fractional CMO vs agency: the difference that matters. If you are not sure which you need, settle that first. Running an RFP for the wrong model wastes everyone's time by definition.

What a good RFP actually contains

A good RFP tells the agency what you sell, what problem you are trying to solve, what you will spend, and how you will decide. That is four things, and most RFPs leave out at least two of them.

Start with context. One page on your business — what you do, who buys, where the money comes from, and what your last twelve months looked like. If you run a downtown San Diego dental practice, say so. Say whether you fill chairs from Google, from referrals, or from paid ads. The agency cannot propose good work without this, and if you withhold it, you get generic proposals.

Next, state the problem in your words. Not "we need SEO." Say "new-patient inquiries from search dropped 30% after our site redesign, and we do not know why." A named problem gets you a diagnosis. A named service gets you a menu.

Then give a budget range. This is the part owners resist most. You worry that naming a number invites the agency to spend all of it. The opposite is true — without a range, every proposal comes back at a different scale and you cannot compare them. Say "we expect to spend between $6,000 and $10,000 a month" and let them show you what that buys.

Finally, state how you will decide and when. Name the criteria. Name the dates. "Proposals due the 20th. Two shortlisted agencies present on the 27th. Decision by the 30th." Agencies plan their week around your timeline. Give them one.

The questions worth asking

Ask questions that reveal how an agency thinks, not how well it writes. The best RFP questions cannot be answered with a case study or a stock paragraph. They force the agency to reason in front of you.

Here are the ones that earn their place:

1. What would you look at first, and why? This shows whether they diagnose or prescribe. A good answer names your data before it names their services.
2. What have you stopped doing for clients in the last two years? This reveals whether they update their methods or sell the same thing they sold in 2019.
3. Who does the actual work, and will I meet them? Many agencies sell with senior people and deliver with juniors. Ask for names and roles on your account.
4. How do you report, and what do you refuse to report? A shop that leads with impressions and rankings is selling you comfort. A shop that ties work to inquiries and revenue is selling you outcomes.
5. What would make you turn down this engagement? An agency that will take any client is not choosing you. The good ones have a shape of client they serve well and say so.

Five questions is enough. Ten is a homework assignment, and homework assignments get delegated to the intern who writes the fluffiest answers. Keep it short so your best prospects say yes to answering.

How to keep it from wasting everyone's time

Invite fewer agencies, ask for less up front, and pay for real work before you commit. Those three moves account for most of the waste in a typical RFP, and fixing them costs you nothing.

Invite three or four shops, not ten. Every agency you invite spends real hours on your proposal, and every proposal you receive costs you an hour to read. Ten invitations produce ten thin proposals and a week of your life gone. Four good invitations produce four proposals you can actually compare. Screen down to your shortlist before you send anything.

Do not ask for free strategy. "Show us what you'd do with our account" sounds reasonable and is not. You are asking for eight hours of senior work for free, which means either the agency declines — the good ones do — or it gives you a shallow answer dressed up as a plan. If you want to see how someone works, pay for a paid audit or a paid discovery sprint. A few thousand dollars of real work tells you more than any free pitch. This is close to how we run our own front door — Discover, then Audit — because a paid first step filters out the shops that only want to win pitches.

Avoid the beauty contest. A ninety-minute final presentation rewards the agency with the best presenter, not the best operator. If you must meet finalists, make it a working session. Bring a real problem. Watch how they think when the slides are gone. The agency that reasons well in the room is the one you want in the room every month.

And set a hard timeline. RFPs rot. A process that drags for two months loses the good agencies — they take other work — and leaves you choosing among the ones with nothing better to do. Publish your dates and hold them.

How to read the proposals you get back

Read for diagnosis first, price second, and design last. The order matters, because the proposal that looks best is rarely the one that thinks best, and the one that thinks best is the one you will still respect in six months.

Start by asking whether the agency understood your problem. Did they repeat your words back, or did they paste in a section about their five-pillar framework? A proposal that opens with your business and your data is a proposal from a shop that listened. A proposal that opens with the agency's mission statement is a shop that will manage you the same way — with slides, not answers.

Then look at how they handle uncertainty. The honest proposals say "we would need to see your analytics before we commit to a number." The overconfident ones promise a first-page ranking in ninety days. Certainty this early is a red flag. Nobody can promise search outcomes without seeing your foundation first, and anyone who does is either guessing or lying.

Compare price against scope, not against each other. A $4,000 proposal and a $9,000 proposal are not competing on price if one is doing twice the work. Line up what each includes. Reporting, content, technical fixes, meetings. Then decide what you actually need. Owners often pay for volume they will never use because a proposal listed a lot of activity.

Watch for who signs off. If the proposal comes from a partner but the delivery team is unnamed, expect the partner to vanish after the contract. The agencies worth hiring stay in the room. We saw this play out with McShanes Solicitors — the work landed because the same people who scoped it did it, and the reporting tied straight back to inquiries the firm could count.

Where this breaks down

An RFP cannot fix an owner who does not know what they want. If you have not decided whether your problem is traffic, conversion, or positioning, no proposal will decide it for you — it will just give you four different guesses. Sort out your own diagnosis first, even roughly. We wrote about that clarity step in When you actually need a Fractional CMO — and when you don't, and it applies here too.

An RFP also cannot tell you whether you will like working with someone every week for a year. Proposals are written. Relationships are lived. Use a paid trial to close that gap before you sign a long contract.

Run the process to serve the decision, not the process itself. Fewer invitations, sharper questions, a paid first step, and a hard timeline. That is enough to pick well without burning a month of everyone's time. Foundations first — including the foundation of knowing what you are actually buying.