What should a marketing agency be accountable for?
An agency cannot be accountable for revenue, and any agency that agrees to be is either about to fail or about to redefine revenue. They do not control your pricing, your sales team, your product or your competitors' behaviour. What they can be held to is narrower, more boring and considerably more useful: the quality of the work, the honesty of the reporting, and a specific set of outputs that they alone determine.
What can an agency genuinely control?
Three things: what gets made, where it runs, and what gets reported. That is the honest boundary. Everything past the click passes through your website, your pricing, your sales follow-up and your product, none of which the agency touches, and holding them to outcomes downstream of those things produces either padded numbers or an early exit.
The reason this matters is that the further an agreed metric sits from the agency's actual control, the more incentive there is to influence the metric rather than the business. An agency held to lead volume will find you cheap leads. An agency held to cost per acquisition will quietly shift spend to brand keywords that would have converted anyway. Neither is dishonest exactly; both are the predictable response to being measured on something they only partly control.
So write the accountability where the control is. The agency is accountable for the work being done, done competently, and reported truthfully. You remain accountable for what happens to the traffic once it arrives.
Should the contract contain revenue or lead targets?
Lead targets are usually a mistake and revenue targets almost always are. A lead target converts the relationship into a volume game within about two months, because the fastest route to more leads is looser qualification, and looser qualification is invisible in the report the agency produces.
There is a version that works, and it is worth knowing because it is rare: a target on qualified leads, where qualification is defined by your sales team, recorded in your CRM, and reviewed jointly. That puts the number under shared control and gives the agency a reason to care about what happens after the form. It requires you to actually maintain the CRM, which is why most attempts collapse.
The safer structure for most relationships is to agree targets as shared expectations rather than contractual obligations, and to make the contractual commitments about process: what runs, how often it is reviewed, how quickly things are fixed when they break.
Which commitments are worth writing down?
Four, and they are all things an agency can control unilaterally. Everything else belongs in a conversation rather than a contract.
| Commitment | Why it matters | How to verify it | Common evasion |
|---|---|---|---|
| Named people and their time | You are usually sold seniors and delivered juniors | Ask who worked on last month's output, by name | "A team of specialists" with no names |
| Account and asset ownership | Ad accounts, analytics and domains must be yours | Check you are the admin, not a user, today | Accounts held under the agency's business manager |
| Reporting on your data, not theirs | A report from the agency's own dashboard cannot be audited | Reproduce one number yourself from source | A branded PDF with no link to the platform |
| Notice and handover terms | Exit is where value is destroyed or preserved | Ask what you receive on the last day, in writing | Thirty days notice, no handover clause |
| Change approval threshold | Prevents silent strategy shifts mid-quarter | Agree what needs sign-off before it happens | "Ongoing optimisation" as a blanket permission |
Who should own the ad accounts and the data?
You should, without exception, and this is the single most common failure in agency relationships. If the ad accounts, analytics property, tag manager container and domain registrations sit inside the agency's structure, then leaving them costs you your entire measurement history and your campaign learning, which is a real switching cost that has nothing to do with the quality of their work.
The mechanism is not usually malicious. It starts as convenience during setup and never gets revisited. That is precisely why it should be checked at the start rather than negotiated at the end, when you have no bargaining position left.
Do this today, in ten minutes: open your ad platform and analytics, and confirm your own company account is listed as the owner or administrator rather than a granted user. If it is not, ask for that to change this week, and treat reluctance as information about the rest of the relationship.
What is a fair way to judge the first ninety days?
By whether the work happened, whether it was competent, and whether the reporting told you anything you did not already know. Not by results, because in most markets ninety days is not enough time for a channel change to show up in revenue, and judging on results that early rewards whoever is best at generating early-looking numbers.
A useful specific check at the end of a first quarter is to ask what they now know about your customers that they did not know at the start. A good answer is uncomfortable and specific: which messages failed, which segment turns out to be unprofitable, which page loses people. A bad answer is a summary of activity.
The second check is whether anything got stopped. An agency that has added five things and stopped nothing in three months has not been evaluating, and the account will get more expensive and less legible every quarter.
When is the agency not the problem?
When the leads arrive and nothing happens to them. This is common and rarely acknowledged, because the diagnosis sits inside your own organisation. If enquiries wait two days for a reply, or the sales conversation contradicts the campaign promise, the marketing can be working exactly as intended and the money is still being wasted.
It is also not the agency's problem when the offer itself is the constraint. No amount of channel work fixes a price that is uncompetitive, a product that is behind, or a category that has moved. Agencies rarely say this out loud, partly out of self-interest and partly because it is not their place to.
The test is to look at what happens to the traffic you already have. If your existing enquiry-to-customer rate is poor, more traffic multiplies an existing loss, and the sensible next spend is on the conversion path rather than on another channel.
Common questions
- Should a marketing agency be accountable for revenue?
- No. Revenue passes through pricing, the website, sales follow-up and the product, none of which an agency controls. An agency held to revenue will redefine the metric or exit. Hold them to what they determine unilaterally: what gets made, where it runs, and whether the reporting is honest and reproducible from your own accounts.
- Should a marketing contract include lead targets?
- Usually not as a contractual obligation. A lead target becomes a volume game quickly, because the fastest route to more leads is looser qualification, and that is invisible in an agency-produced report. A qualified-lead target defined by your sales team and recorded in your own CRM can work, but only if the CRM is genuinely maintained.
- Who should own the ad accounts when working with an agency?
- The client, always. Ad accounts, the analytics property, the tag container and domain registrations should sit under the client's own ownership with the agency granted access. When they sit inside the agency's structure, leaving costs the client its measurement history and campaign learning, creating a switching cost unrelated to work quality. This is worth checking on day one rather than at exit.
- How do you evaluate a marketing agency in the first ninety days?
- On work delivered, competence, and whether the reporting revealed anything new, rather than on results, since ninety days is rarely long enough for a channel change to reach revenue. Two useful checks: ask what they now know about your customers that they did not at the start, and ask what they have stopped doing. An agency that has only added things has not been evaluating.
- How do you tell if poor results are the agency's fault?
- Look at what happens to the traffic that already arrives. If the existing enquiry-to-customer rate is poor, if replies take days, or if the sales conversation contradicts the campaign promise, then more traffic multiplies an existing loss and the constraint is internal. The same applies when the offer is uncompetitive, which no amount of channel work fixes.
