How to hire a SaaS marketing agency

The worst agency relationships do not fail dramatically. They drift. Month one is energetic, month three is a status call, month seven is a slide deck of impressions, and month eleven is a founder asking what exactly they have been paying for. The failure was in the buying process, not the agency.

This is a buying process, not a list of agencies. It covers when an agency beats a hire, the six questions that expose the pitch, how the pricing models shape behaviour, a pilot structure that de-risks the commitment, and the contract terms you will regret not having.

When does an agency actually beat hiring someone?

Three situations, and it is worth being strict about them because the default assumption in most companies is wrong in both directions.

The need is temporary. A site migration, a rebrand rollout, a category launch, a one-off research study. You need six months of intense capability and then you need none. Hiring for that creates a person you will have to redeploy or let go.

You need speed you cannot hire for. A good demand generation manager takes 60 to 90 days to find and another 60 to ramp. If the board wants pipeline movement inside a quarter, an agency starting in two weeks with an existing playbook is the only honest path, even if you hire in parallel.

The specialism is too narrow for a full-time role. Technical SEO for a JavaScript-heavy app. Paid search management when you spend $25,000 a month, which is not enough to justify a full-time specialist but too much to run badly. Conversion rate optimisation as a discipline. These are real skills you need eight hours a month of.

And the two situations where an agency never wins. First, when the work requires deep product knowledge that takes months to acquire, such as product marketing or technical content for a complex developer tool. An agency writer will produce competent generic material and your engineers will hate it. Second, when you have no internal owner. An agency without a counterpart inside the company who makes decisions, provides access and chases approvals will fail, and it will look like their fault.

Which six questions expose the pitch?

Six, and you should ask all of them in the second conversation, not the first. The first call is sales. The second is where you find out what you are buying.

Who does the work day to day, by name, and what else are they on? The people in the pitch are frequently not the people in the delivery. Get names, then look them up. An agency that answers this cleanly is confident in its bench. One that says “we assign based on fit after onboarding” is telling you they do not know yet, which means neither do you.

What is the pod ratio? How many client accounts does the strategist carry, and how many does the specialist carry. One strategist across twelve accounts produces templates, and you will recognise your deliverables as a lightly-edited version of somebody else’s. Four to six accounts per strategist is a workable ratio. Above eight, expect volume over judgement.

• What ships in week one? Not “discovery”. A specific artefact. Good answers include an audit with prioritised findings, a tracking implementation review, or the first three content briefs. An agency whose first four weeks are entirely discovery is either doing genuinely deep strategy work, which they should be able to describe, or padding the ramp.

What data access do you need, and why? This surfaces both competence and risk. An agency that wants CRM access to tie work to pipeline is thinking about outcomes. One that only wants Google Analytics is going to report on sessions. It also surfaces whether they want to own your ad accounts, which is a contract issue covered below.

What is your client retention rate over 24 months? Ask for the number and for the average engagement length. Agencies with high churn are either bad or working in a segment with structurally short engagements, and either answer is useful. Expect some evasion, and note that evasion is itself data.

Describe an engagement that failed and what you would do differently. The highest-signal question on the list. An agency that cannot name one has either never reflected or is not telling the truth. The quality of the answer, specifically whether they own their part of the failure, predicts how they will behave when something goes wrong with you.

How do the pricing models change agency behaviour?

Four models, each with an incentive it creates whether anyone intends it or not. Pick the one whose incentive matches the work.

Model	Typical shape	What it incentivises	Best for	Watch for
Monthly retainer	$6K to $40K per month, 6 to 12 month minimum	Retention and steady output	Ongoing programmes with unclear scope	Coasting after month four
Fixed project	$20K to $120K for a defined deliverable	Finishing and moving on	Migrations, rebrands, audits, one-off builds	Scope disputes, thin post-delivery support
Performance	Base plus fee per lead, opportunity or percentage of spend	Whatever is easiest to attribute	High-volume, short-cycle self-serve motions	Lead quality collapse, attribution gaming
Hybrid	Reduced retainer plus outcome bonus	Balanced, if the outcome metric is right	Mature relationships with agreed measurement	Arguing about the measurement instead of the work

The performance model deserves a warning. Paying per lead in B2B SaaS reliably produces more leads and worse ones, because the agency optimises the metric you pay on and lead quality is not that metric. If you use performance pricing, tie it to opportunities created or pipeline value, accept the longer feedback loop, and add a quality guardrail. The detail on typical rate structures is in what B2B SaaS marketing agencies charge.

Percentage-of-spend pricing for paid media has the obvious flaw: the agency earns more by spending more. It is still the industry norm and it works fine as long as you review spend decisions independently. SaaS PPC agencies covers the specific version of this problem.

What does a good paid pilot look like?

Sixty days, paid at roughly the normal rate, with three to five specific deliverables and a written definition of what success looks like. Do not ask for free work, because free work gets the bench.

The pilot also tells you something no reference call will: how they behave when something goes wrong in the first eight weeks. Something always does. Access gets delayed, an approval sits for a week, a deliverable slips. How they handle that is the relationship.

Which contract clauses will you regret not having?

Five, and every one of them is about the day the relationship ends rather than the day it starts.

Asset ownership on payment. All content, creative, design files and code created under the engagement become yours when the invoice is paid. Specify source files, not just outputs. A brand refresh delivered as flattened PNGs is a hostage situation.

Your accounts, your admin. Google Ads, LinkedIn Campaign Manager, Search Console, analytics and your CMS should be your accounts with the agency granted access, never agency-owned accounts you are a guest in. Agencies that insist on owning the ad account usually cite billing convenience, and the effect is that your historical performance data walks out with them. Agency tech stack and tool ownership covers the fuller list, including who holds the Ahrefs or Semrush seat and what happens to it.

Named key person. The strategist you evaluated cannot be swapped without your written consent, and a swap you refuse is grounds for termination without penalty. This is the clause that makes the “who does the work” question enforceable.

Notice you can live with. Thirty to sixty days after the pilot. Anything longer and you are paying for months of a relationship you have already decided to end.

A transition obligation. Thirty days of handover at normal rates on termination, including documentation, access transfer and a written state-of-play. Without it, the last month is silence.

What are the honest downsides of using an agency?

Three, and any agency that denies them is not being straight with you.

Agencies never learn your product as well as an employee does. They will be competent about your category and shallow about your product, which is fine for paid media and painful for product marketing. Budget internal time for subject matter input and accept that you cannot outsource that part.

Attention follows revenue and risk. You are one of a portfolio, and when a bigger account has a crisis your strategist is on that call, not yours. This is not malice, it is capacity, and it argues for being a meaningful account to a smaller agency rather than a small account at a large one.

Institutional knowledge leaves with them. Two years of context about what was tried and why lives in an agency’s heads and their drive. The transition clause helps and does not solve it. If a channel is going to be permanently core to your business, the long-run answer is in-house, and the question is when you make the switch. The in house vs agency cost calculator gives you the breakeven arithmetic, and in house marketing team vs agency covers the structural tradeoffs.

What to do this week

Write the one outcome you want in a single sentence with a number and a date. If you cannot, the problem is not agency selection, it is that nobody internally has decided what marketing is supposed to do this year.

Then shortlist three agencies, run the six questions in second conversations, and scope a 60 day paid pilot with the strongest one. Look at how the SaaS growth agency model works internally so you know what you are buying, check the shortlists in best SaaS marketing agencies, and make sure you have the internal owner in place first, which the SaaS marketing hiring process covers. For the wider context on building a team around outside help, see SaaS marketing careers and the SaaS marketing hub.