Ask a founder how they screened their last channel partner and you will usually get a real answer: they checked the company's history, spoke to two existing partners, argued over the revenue split, and agreed how deals would be attributed. Ask the same founder how they picked the agency that now runs their website, their search presence and their ad budget, and the answer is often "they came recommended and the deck looked good."
That gap is odd, because the agency relationship has more of a partnership's characteristics than most channel deals do. It shares your outcome, it holds your data, it speaks in your name in public, and unwinding it is genuinely expensive. The short version of this piece: treat the agency search as partner sourcing, run the same four checks you would run on any alliance - capability, scope, measurement and exit - and settle all four before money moves.
Why an agency is a partnership, not a purchase
A purchase ends when the thing arrives. A partnership keeps producing - or keeps costing - long after the first invoice, and that is exactly the shape of an outside marketing relationship.
- The outcome is shared. If the campaigns work and the site cannot convert, you both lose; if the site converts and nobody sends traffic, you both lose. Neither side can succeed alone, which is the textbook condition for a partnership rather than a supply arrangement.
- The data is shared. They will hold your analytics, your ad accounts, your keyword history and often your domain records. Shared systems are a partnership problem, and they need the same clarity about ownership you would demand of any joint asset.
- The reputation is shared. Whatever they publish, bid on or post carries your name, not theirs. You are lending an external team your brand's voice for the duration.
- Exit is expensive. Losing an agency is not like losing a supplier. Campaign history, tracking setup, content and account access all sit on the other side of the relationship unless you arranged otherwise up front.
That last point is why this belongs on a partnerships blog rather than a procurement one. Everything that makes partner due diligence useful - checking capability honestly, writing down who owns what, agreeing how results are counted - applies here almost unchanged. The full guide to finding and vetting partners sets out that discipline in general terms; what follows is how it lands when the partner in question sells marketing.
Check one: capability, described in verbs
The weakest signal in the market is a service list. Every agency lists the same nouns - SEO, social, web design, paid media, branding - and the list tells you nothing about how any of it is done.
What tells you something is a description written in verbs, published where you can hold them to it. "Technical SEO, on-page optimisation and content direction" is a scope you can compare between two proposals. "We are passionate about growth" is not. When an agency publishes its own service lines in specific language, you can at least check whether the proposal in front of you covers what the website claims the firm does.
A worked example of the format: Spifex Digital, a full-service digital marketing and website design agency, publishes six service lines - digital marketing and growth, website design and development, branding and creative, SEO and content marketing, social media management, and automation and lead systems - each with its own description rather than a single word. Whether or not that firm is the right fit for you, the published-scope format is what you want to see, because it gives you something concrete to hold every candidate to.
Then ask the question the list cannot answer: which of these do you do in-house, and which do you subcontract? There is nothing wrong with a subcontracted line, but you need to know where the handoffs are, because handoffs are where accountability goes missing.
Check two: scope, and the multi-vendor trap
The most common failure in outsourced marketing is not incompetence. It is a boundary problem: the website sits with one firm, the ads with another, the search work with a third, and each one's reporting proves the problem lives somewhere else.
It is a familiar pattern to anyone who has run a partner program - three partners each optimising their own number, with nobody accountable for the number the business actually cares about. The fixes are the same too. Either consolidate the scope so one team owns the chain from click to enquiry, or keep the vendors separate and appoint someone internally who owns the whole funnel and can overrule all of them.
What does not work is the middle ground, where three suppliers each hold a third of the outcome and none of them holds the definition of success. Before you sign, draw the chain on one page - traffic source, landing page, form or call, follow-up, reporting - and write a name against every link. Any link with two names or no name is where the next argument will start.
Check three: measurement, agreed before the budget
Partner-sourced revenue is a solved problem in channel programs: you agree what counts, how it is attributed, and where it is recorded, before anyone starts selling. Marketing engagements routinely skip all three, and then relitigate them in month four when the numbers disappoint.
Settle this first:
- What counts as a result? A form, a phone call, a booking, a completed checkout. Pick one primary action and name the runners-up.
- Where is it recorded? In your analytics property, your CRM, or a dashboard the agency owns? The third option is not a record; it is a report.
- When do we review? Set the date at signing, not when the first invoice lands. Work judged on feelings gets cancelled on feelings.
- What is a fair leading indicator? Search and content work compounds slowly, and the honest answer to "when will it show" is a range, not a date. An agency that gives you a specific position by a specific week is telling you something about their sales process, not their method.
If you have run a co-marketing campaign with a partner, this is the same negotiation as agreeing how a joint campaign's leads get counted - and it is just as unpleasant to have after the fact.
Check four: exit, arranged while everyone is friendly
Every asset the engagement touches should be in your name, with the agency added as a user. That means the domain registration, the hosting, the analytics property, the ad accounts, the business profile and any tracking container. It costs nothing to arrange at the start and is painful to unpick when the relationship has already soured.
Ask two more exit questions while you are there: what happens to the content and the creative if we stop, and what is the notice period? Neither answer needs to be generous. It needs to exist in writing.
Red flags worth walking away from
- A promised ranking position, a promised revenue figure, or any outcome guarantee attached to a channel neither of you controls.
- A proposal that will not name what is in scope this month versus later.
- Reporting that lives only in the agency's own dashboard.
- Reluctance to put accounts in your name.
- A price quoted before anyone asked what a result looks like for you.
FAQ
Is a full-service agency always better than specialists? No. A specialist can be the better answer when you already have someone internally who owns the whole funnel and can direct them. Consolidation earns its keep when nobody in-house is holding the chain together, because then the handoffs are the risk you are actually buying out.
How long should the first engagement be? Long enough to be judged and short enough to leave. Prefer a first piece of work that is specific, measurable against the result you defined, and reviewable on a date you set before signing.
Should I ask for references? Yes, and ask them partnership questions rather than performance ones: what happened when a month went badly, who they spoke to, and how long it took to get an answer. Anyone can supply a happy reference; how a supplier behaves during a bad quarter is the useful signal.
Is it safe to give an agency access to my ad and analytics accounts? Yes, provided the accounts are yours and they are added as users rather than owners. The risk is not access; it is access to accounts created in someone else's name, which turns a routine offboarding into a negotiation.
What to do next
Run the marketing decision through the same process you would use for any alliance: profile the partner you actually need, shortlist on published scope rather than on decks, agree the single number the engagement is judged on, and put the accounts in your own name before work starts. If you are also building referral or channel relationships alongside this, browse the partnership programs and PRM comparisons in the Alianzy directory to see how the same vetting discipline gets applied to partner software and programs.
And if the conclusion you reach is that the handoffs are the problem - that the site, the search work and the campaigns need to answer to one plan rather than three invoices - start by looking at how a candidate publishes its own scope. Spifex Digital sets out its service lines and what each one covers, which is exactly the level of detail you should be able to compare, question and hold any marketing partner to before you sign.
