Choosing a B2B marketing agency is usually treated as a shortlist exercise: collect four proposals, compare scope and price, pick the one that presents best. That process reliably produces the same outcome, which is buying an excellent fix for a problem you did not have.
The better sequence is diagnosis first, archetype second, mechanics third. This guide walks all three. We run an integrated B2B marketing agency, so we have an obvious bias toward one of the four archetypes below. We have tried to describe the other three the way their best clients would describe them, and to name the cases where they beat us.
Step one: name your constraint
A constraint is the single thing that, if it stays fixed, prevents everything else from working. B2B marketing programs stall for four distinct reasons, and each one points at a different kind of partner.
- Content quality. You publish on cadence, but the work is interchangeable with your competitors' and nothing earns links, rankings, or citations. The gap is editorial depth and subject matter expertise on the page.
- Site capability. Your content is decent and your rankings are capped anyway: slow pages, client-side rendering, no internal link architecture, a CMS that requires a developer for a headline change, no conversion path below the fold.
- Channel fragmentation. SEO, content, paid, and web live with four vendors reporting four metrics. Every one of them is hitting their number and pipeline is flat.
- Capacity. The strategy is right and there is one of you. You do not need a new plan, you need execution volume.
Write your constraint down in one sentence before you take a single sales call. Hiring an editorial shop to fix a technical problem is the most common and most expensive mismatch in this category, and it happens because the diagnosis step gets skipped and the vendor with the best deck defines the problem for you.
Step two: match the archetype to the constraint
Almost every option in the market is a version of four archetypes. They are not ranked. They are built for different constraints, and the one that is wrong for you is not a bad agency.
| Archetype | Best when your constraint is | Typical monthly range | Main risk |
|---|---|---|---|
| Generalist full-service | Capacity, plus no internal marketing leadership | $5K to $20K | Broad but shallow; B2B depth varies by account team |
| Channel specialist (SEO, paid, or content only) | One channel underperforming while the rest works | $4K to $25K | Optimizes its own metric; someone internal must integrate |
| Integrated growth partner | Fragmentation across site, search, content, and paid | $7.5K to $50K+ | Higher commitment; wasteful if you only need one channel |
| In-house build | Deep product complexity and a multi-year horizon | $35K+ fully loaded for three roles | Six to nine months to assemble; hiring risk is yours |
Two practical notes. First, a channel specialist is the right answer more often than integrated agencies like to admit: if your site converts, your content is strong, and only paid is broken, buy paid. Second, in-house and agency are not exclusive. The most stable arrangement we see is an internal owner of positioning and product marketing with an agency owning the execution surfaces they cannot staff.
The cost of fragmentation, calculated
If your constraint is fragmentation, the comparison is not four retainers against one retainer. It is total cost including the coordination tax and the compounding you are not getting. Here is the arithmetic for a typical mid-market setup.
- SEO vendor: $4,000/mo
- Content shop: $5,000/mo
- Paid agency: $3,500/mo plus 12% of a $30,000 spend, so $7,100/mo
- Web dev partner on retainer: $2,500/mo
That is $18,600/mo in fees. Now add the coordination tax. A head of marketing at a $180,000 fully loaded salary costs roughly $86/hour. Four vendor relationships realistically consume eight hours a week in calls, reviews, briefs, and reconciliation, which is $2,750/mo. True cost: about $21,350/mo.
The larger number is the one that does not appear on any invoice. When the SEO vendor cannot edit templates, the content shop cannot see paid search query data, and the paid agency cannot change a landing page without a dev ticket, the surfaces stop compounding. Each channel performs at standalone efficiency instead of feeding the others. That is the gap an integrated partner has to justify, and the honest way to test the claim is to ask a candidate to describe a specific change they made last quarter that touched three surfaces in the same week.
Step three: verify the mechanics
Once the archetype is right, the remaining risk is execution. Four mechanics predict it better than anything in the proposal.
- Who does the work, by name. Ask for the actual strategist and the actual practitioners, their other account load, and whether the people on the pitch call remain on the account. A pitch team that disappears at signature is the single most reliable predictor of a disappointing engagement.
- What ships in the first 90 days. You want a list of artifacts, not phases. "Technical audit with prioritized fix list, query map covering the buying committee, measurement and CRM handoff configured, eight pages live" is an answer. "Discovery, strategy, and implementation" is not.
- One metric, agreed before signature. Pipeline influenced, qualified conversations, or opportunity count. If the engagement is judged on sessions or pieces published, you will get sessions and pieces published.
- A written out-of-scope list. Every honest scope has one. Ours says we do not run cold email, SDR teams, or ABM outreach. An agency that claims to do everything either subcontracts silently or will discover its limits on your budget.
Questions that actually separate candidates
Most sales-call questions are answered from a script. These are not, which is why the quality of the answer tells you something.
- Describe your last engagement that failed, and what you would do differently.
- What would have to be true about our business for you to decline the work?
- Which of our current channels would you turn off, and what would you lose?
- If the leading indicators are flat at day 90, what changes in month four?
- What do you need from us monthly, in hours, for this to work?
- How do you handle AI search visibility, and how is that different from SEO?
The last one matters more each quarter. Buying committees increasingly shortlist through generated answers, which is a different discipline from ranking. An agency that treats answer engine optimization and generative engine optimization as synonyms for "we add FAQ schema" has not done the work. Ask specifically how they measure citation share and how they measure whether models describe your product accurately, because those are two different problems with two different levers.
Reading a proposal without getting sold
Proposals are optimized for signature, not comprehension. Three passes make them readable.
- Pass one, strip the adjectives. Rewrite each deliverable as a noun with a quantity and an owner. Anything that resists that treatment is not a deliverable.
- Pass two, find the margin. Estimate the hours implied and the seniority required. If a $6,000 retainer implies 60 senior hours, either the hours are junior or the hours are not there.
- Pass three, check the case studies. A percentage without a baseline is decoration. "Traffic up 300%" from 400 sessions is a rounding error. Ask for the starting number and the time window.
On pricing specifically: an agency that will not publish or state ranges before a qualification call is pricing you, not the work. We wrote a full breakdown of what each retainer band actually buys in our guide to SaaS marketing agency pricing, and our own tiers are on the pricing page with no gate.
Common mistakes
- Shortlisting before diagnosing. The vendor who defines your problem will define it as the thing they sell.
- Optimizing the retainer instead of the outcome. Saving $3,000/mo on a program that produces nothing is a $36,000 loss, not a saving.
- Buying a three-month trial of a nine-month discipline. You pay for the build and leave before the compounding.
- Withholding subject matter expert time. Content without real product specificity is the reason most B2B content is invisible. Thirty to sixty minutes a month is the difference.
- Judging month two on pipeline. Agree leading indicators up front so you can be rigorous early without being wrong early.
- Ignoring the exit. Ask who owns the site, the analytics configuration, the content, and the ad accounts if you leave. The answer should be you, in writing.
A workable decision process
Write the constraint in one sentence. Pick the archetype that fixes it. Talk to three candidates inside that archetype rather than one from each, so you are comparing like with like. Score them only on the four mechanics. Then choose the one whose out-of-scope list you can live with, because that list, not the capability list, is what you will actually be managing around for the next six months.
If your constraint is fragmentation and you sell software, that is precisely the case we are built for: website, SEO, AEO, GEO, content, and paid run by one team under one strategist, with published pricing and a written scope. Bring your current vendor mix and last quarter's reporting to a 30-minute growth call and we will tell you which of the four archetypes you actually need, including when it is not us. If you would rather read first, the B2B marketing agency page lays out exactly what we own and what we refuse, and the comparison of B2B marketing companies sorts every firm you are likely to shortlist into the five types. Once your shortlist exists, the five axes that decide which one is best for you turn it into a decision.