A service-level agreement is what turns "we'll handle your ad ops" into something you can actually hold a partner to. Without one, every disagreement becomes a matter of opinion: was that creative swap late, was that discrepancy caught quickly enough, should someone have answered on a Saturday? With a clear SLA, those questions have answers written down before anyone is frustrated.
In short: a useful ad ops SLA defines turnaround times per request type, the QA standard every launch must meet, coverage hours and how urgent issues outside them are handled, an escalation path with named roles, reporting commitments, how errors are measured and remedied, and the terms for exiting the agreement, including how your documentation and access are handed back.
1Turnaround Times by Request Type
A single turnaround promise for "all requests" doesn't work, because a creative swap and a new multi-line-item campaign build are very different amounts of work. Split requests into categories and agree a turnaround for each. A typical breakdown separates these request types:
| Request category | Examples |
|---|---|
| Urgent fixes | A broken creative or a campaign that has stopped delivering |
| Routine changes | Swaps, flight changes, targeting edits |
| New campaign builds | Multi-line-item campaign setup |
| Reporting requests | Reports on delivery and performance |
Be precise about when the clock starts. The most useful definition is "from receipt of a complete request", which means the SLA should also define what a complete request contains: the brief, all creative assets, tracking, flight dates and targeting. That protects both sides, since the partner isn't penalized for waiting on missing assets and you know exactly what to send.
2QA Standards Every Launch Must Meet
Turnaround without quality is meaningless, so the SLA should state what gets checked before anything goes live. At a minimum: creative specs and rendering, click-through and tracking, flight dates, targeting against the insertion order, frequency caps, and a preview in a real environment. Ideally the agreement references a written checklist rather than describing it loosely; ours is published as the ad trafficking QA checklist.
Also agree whether a second person reviews each build. A two-person check costs a little more time but catches the kind of mistakes the person who built the campaign naturally overlooks.
3Coverage Hours and Escalation
State the hours the team is actively working, in your time zone, and what happens outside them. If campaigns run around the clock, decide which issues justify an out-of-hours response, for example a campaign that has stopped delivering or a creative that is visibly broken, and how that response is triggered.
The escalation path should name roles rather than individuals, so it survives staff changes: first contact, team lead, and account owner, with the time allowed at each level before the issue moves up. Agree the channel too. Urgent issues raised in a busy shared chat channel are easy to miss, so many teams use a separate route for anything marked urgent.
4Reporting Commitments
Spell out which reports are delivered, to whom, how often and in what format, and who is responsible for reconciling numbers between platforms when they disagree. Discrepancies between the ad server and third-party measurement are normal; the SLA should say who investigates them and how quickly. If reporting is a large part of the work, it may be worth scoping it as its own service, like our reporting and analytics support.
It also helps to agree a short monthly service report from the partner: requests handled, turnaround against the SLA, errors found and what was done about them. That report is the basis for every future review conversation.
5How Errors Are Measured and Remedied
Agree how a mistake is defined and recorded before one happens. A practical approach separates errors caught in QA, which are part of the process working, from errors that reach a live campaign. For live errors, the SLA should require a short written explanation of what happened and what will change to prevent it recurring.
Remedies vary between agreements. Some use service credits, some rely on a review and improvement plan, and many combine the two. Whatever the approach, make sure it is proportionate and that the process for agreeing an error happened is clear. Our article on the real cost of ad trafficking errors explains why prevention is worth paying for.
6Security, Confidentiality and Exit Terms
The SLA, or the contract around it, should cover confidentiality, how platform access is granted and revoked, and how client data is handled. If the partner works white-label under your brand, state that explicitly, including which email domains and names they use with your clients.
Finally, plan the exit at the start. Agree the notice period, a handover period during which the partner continues working while documentation is transferred, and that the runbook, naming conventions and any reports built for you belong to you. A partner who is confident in their service rarely objects to clear exit terms.
7A Quick SLA Checklist
| SLA area | What to specify |
|---|---|
| Turnaround | Turnaround times for each request category, with a definition of a complete request |
| QA | A written QA checklist and whether a second reviewer is required |
| Coverage | Coverage hours in your time zone and the out-of-hours policy |
| Escalation | An escalation path by role, with response times at each level |
| Reporting | Reporting deliverables and responsibility for reconciling discrepancies |
| Errors | Definitions of errors, the incident write-up process and the remedy |
| Access and confidentiality | Access, confidentiality and white-label terms |
| Exit | Notice period, handover period and ownership of documentation |
If you are comparing partners, put this checklist in front of each of them and compare the answers side by side. To see how we structure engagements, read about our managed ad services or ask us for a sample SLA.
