A paid search agency is paid to make decisions inside advertising auctions on your behalf: which searches to bid on, which to refuse, what the ads say, where the clicks land, how much a conversion is worth and when to stop. The platforms have automated a growing share of the mechanics, which has not made the agency redundant so much as moved the job. The work that matters now is deciding what the automation is optimising toward, feeding it accurate information about which conversions were actually valuable, and refusing the searches it would otherwise happily buy. Understanding that shift is the fastest way to tell a serious candidate from one still selling manual bid management as though it were 2015.
What the month actually contains
A well run account has a rhythm. Search terms are reviewed and negatives added, because automated matching will keep finding adjacent searches and some of them are expensive nonsense. Budgets are moved between campaigns according to what is converting rather than what was planned. Ad copy and assets are tested in a structured way rather than swapped when someone gets bored. Landing pages are checked against what the ads promise. Conversion data is audited for double counting and for actions that were never worth counting. And someone writes down what changed and why, which is the deliverable you can actually inspect. Ask a candidate for a redacted example of a monthly report before you sign. A report that lists what was decided and what will be tested next tells you how the account will be run. A dashboard screenshot with a summary paragraph tells you it will not be run at all.
Fee models bias the advice, so name the bias
Percentage of ad spend is the oldest model and aligns the agency with growth, which is comfortable when growth is right and awkward when the correct recommendation is to cut budget. Flat retainers are easy to budget and reward attention early in the relationship, then quietly reward less of it once the account is stable. Hybrid models with a floor plus a percentage above a threshold split the difference. Performance based fees tied to conversions or revenue sound like the obvious answer and fail on trust: they require both sides to agree that the conversion data is accurate, which is precisely the thing that becomes contested when money depends on it. None of these are wrong. What is wrong is choosing one without saying the incentive out loud and agreeing how you will handle the moment when the fee model and the right decision point in different directions.
Conversion tracking is the actual deliverable
Most underperforming accounts are not badly managed, they are well managed toward the wrong target. If the platform counts every form submission as a conversion, it will find you more form submissions, including the ones from people who will never buy. The fix is unglamorous plumbing: define the conversion that reflects real value, and get that back into the ad platform even when it happens days later in a sales system rather than on the website. Google Ads supports importing conversions that occurred offline, which is how a booked demo, a qualified opportunity or a closed deal becomes something the bidding can optimise toward. Ask every candidate how they will do this, who builds it, how long it takes and what happens to it at the end of the engagement. Candidates who treat it as a phase two item are telling you the first phase will be judged on numbers that do not mean anything.
Who needs an agency, and who does not
A single platform account with a handful of campaigns, a stable offer and someone in house who enjoys the work does not need an agency and will not get much from one. The case for hiring strengthens with complexity: several platforms, several locations, seasonal swings, a large catalogue, or a budget large enough that a few points of waste exceeds the fee. It also strengthens when the in house owner is a marketing generalist whose attention is genuinely needed elsewhere. If you do hire, be clear about which of the two purchases you are making: continuous management of an existing programme, or a strategic rebuild followed by management. Buyers comparing paid search marketing services usually get better quotes by asking for those two phases to be priced separately, because it forces candidates to say what the first ninety days contains.
Questions people ask about paid search agency
How much of the fee should go to strategy versus execution?
Ask the question in hours rather than proportions. A fee that buys ten hours a month is a different product from one that buys thirty, and the split between senior thinking and account maintenance matters more than a percentage. Any agency that cannot describe how your fee converts into named people and time is not ready to be compared.
Does account structure still matter with automated bidding?
Yes, though differently. Structure now mostly controls budget separation, reporting clarity and how conversion signals are grouped, rather than manual bid control. Fragmented accounts split data thinly and starve the automation of signal. Over consolidated accounts make it impossible to protect a priority segment's budget. A good candidate explains that tradeoff for your specific case.
Should the agency also build the landing pages?
It is cleaner when they do, because landing page quality is inseparable from campaign performance and split responsibility produces finger pointing. If your web team owns pages, agree turnaround times and a change process in advance. The worst arrangement is an agency accountable for conversion rate with no ability to change the page.
What is a reasonable review point?
A full quarter, with criteria agreed on day one. The first month is usually tracking repair and restructuring, so judging on it measures the previous state of the account rather than the new agency. Write down before you start what result would justify continuing and what result would mean stopping.