Vendor evaluation in B2B marketing usually goes wrong in the first week, before anyone has read a proposal. A team collects five referrals, books five calls, and ends up holding five documents that describe five different jobs at five different prices. There is no way to score that set, so the decision falls to whoever presented best.
The fix is structural, not diligence effort. Sort by type, shortlist within a single type, and score everyone on the same five dimensions. This guide walks that process, including the artifacts to demand and the specific answers that predict a bad engagement. We run one of these types ourselves, an integrated firm, so treat the positioning here as informed and biased. The full type comparison lives on our B2B marketing companies page.
Step one: name the constraint, not the goal
Every team's goal is more pipeline. That is not useful for evaluation because every firm sells it. The constraint is the one thing that, left unchanged, prevents everything else from working. There are four common ones in software:
- Nobody owns the whole picture. SEO, content, paid, and web sit with four vendors. Each hits their number and pipeline is flat.
- The site cannot support the program. Content is fine and rankings are capped anyway: slow pages, no internal link architecture, a CMS that needs a developer for a headline.
- The positioning is unclear. Traffic converts poorly everywhere, sales calls start with confusion, and no channel fixes that.
- Capacity. The plan is right and there is one of you.
Each constraint maps to a different type of company. Buying an excellent fix for the wrong constraint is the single most expensive mistake in this category, and it is nearly always a sorting error rather than a judgment error.
Step two: shortlist inside one type
Three candidates from one type produce a real comparison. One candidate from each of three types produces a preference. If your constraint is fragmentation, talk to three integrated firms. If your constraint is positioning, talk to three product marketing consultancies and revisit channels afterward.
This also fixes the price confusion. Within a type, ranges cluster tightly, and an outlier means something specific: unusual seniority, unusual scope, or a scope gap you have not found yet. Across types, price tells you almost nothing. Our own breakdown of agency pricing covers what the ranges buy at each level.
Step three: the five-dimension scorecard
Score each candidate one to five on the same five dimensions. Total the scores, but pay more attention to which candidate you cannot score, because an unscoreable candidate is selling a deck rather than a delivery model.
- Constraint fit. Does this firm's core build address the sentence you wrote in step one, or does it address it as an adjacent capability.
- Named ownership. Who does the work after signature, by name and role. A strategist who appears only in the pitch is a sales asset.
- Dated 90-day list. Specific deliverables with dates. A phase diagram labeled discovery, strategy, execution is not a plan.
- Measurement depth. Does reporting terminate at pipeline influenced or at impressions, rankings, and activity counts.
- Scope honesty. Is there a written list of what they will not do. Firms that refuse to write one negotiate it later, at your expense.
How to read a proposal
Proposals in this category are optimized to look thorough. Read them in a fixed order and most of the volume becomes skippable.
- Find the deliverable list. If there are no dates and no counts, stop reading and ask for them before the next call.
- Find the team page and check whether roles are named or generic. "Senior SEO strategist" without a name means unassigned.
- Find the metric the engagement is judged on. There should be exactly one primary, with leading indicators underneath it.
- Find the exclusions. Media spend, licenses, and third-party fees should be explicitly outside the retainer at any honest firm.
- Find the term and the exit. A six-month minimum is reasonable for compounding work. An auto-renewing twelve-month term with a ninety-day notice window is not.
The questions with a cost to answering
Capability questions are rehearsed and every firm passes them. Use questions where an honest answer costs the firm something.
- What would you cut from our current program first, and what evidence triggers it.
- Describe an engagement that failed and what you changed structurally afterward.
- Which prospective clients do you turn down, and why.
- If every leading indicator is flat at day 90, what specifically happens.
- Who on your team will I speak to in month four, and how often.
The failure question is the most informative one in the set. A firm with no failure story either has not run enough engagements or is unwilling to be accurate with you, and both answers matter.
Red flags that reliably predict a bad engagement
- Pricing withheld until a qualification call. The number is being set from your funding announcement rather than your scope.
- Percentages without baselines. A 340 percent traffic increase from a base of 900 sessions is a rounding event.
- Guarantees on organic outcomes. Nobody controls a ranking or a citation. Guaranteed deliverables are fine; guaranteed results are not.
- Pitch team disappears at signature. Ask directly, then confirm the answer in the contract.
- No out-of-scope list. Every scope has edges. A firm unwilling to write them down is preserving room to renegotiate.
- Channel answers with no tradeoffs. A firm that recommends everything has not diagnosed anything.
A worked example of the scoring
A Series A software company with four vendors and flat pipeline writes its constraint as fragmentation. It shortlists three integrated firms at $9K, $14K, and $22K per month.
The $9K firm scores well on constraint fit and price, has one named strategist across eleven accounts, and its 90-day list contains four content pieces and an audit. The $22K firm names five people, delivers a dated list with a site rebuild inside the first 60 days, and reports to pipeline influenced. The $14K firm cannot produce a dated list at all and describes measurement as "full-funnel visibility".
The middle option drops out on unscoreability, not price. The real decision is between capacity risk at $9K and budget risk at $22K, and it turns on whether the site itself is part of the constraint. If it is, the cheaper option cannot fix the thing that is broken and the price difference is irrelevant.
Reference checks that produce signal
Ask for two references: one current client past the two-year mark and one client who left. The second request is the test. A firm that cannot produce a former client who will speak either has never retained anyone long enough to churn well or does not want you talking to the ones who did.
On the call, skip outcomes and ask about mechanics: what the reporting cadence was in month four, what got missed and how it was handled, whether the seller stayed involved, and what they would scope differently. Those answers describe delivery. Outcome anecdotes describe the sale.
What a good first 90 days looks like
Regardless of type, a competent engagement front-loads diagnosis and instrumentation so spend follows measurement rather than preceding it. Month one should produce working analytics and CRM handoff, a mapped query and channel set, and a technical and AI-readiness baseline. Months two and three should produce shipped pages, published content on a stated cadence, and live campaigns with weekly iteration.
By day 90 you should be able to check leading indicators agreed before signature: pages shipped against plan, index and citation coverage, ranking movement on the mapped set, conversion rate on money pages, and cost per qualified conversation. If all of them are flat, the lag argument stops protecting the program. Our month-by-month methodology shows the version of this we commit to in writing.
Making the call
Pick the candidate whose out-of-scope list you can live with. Capability lists converge across serious firms; exclusion lists are where the real differences sit, and they are what you manage around every month for the length of the term.
If your constraint is fragmentation and you sell software, that is the case we are built for: website, SEO, AEO, GEO, content, and paid 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 type of company you actually need, including the cases where it is not us. If you would rather read first, the agency archetype guide covers the diagnosis step in more depth, and what best actually means for your company explains why ranked agency lists cannot answer it.