How long should you give a new marketing agency? Long enough for the channel you hired them for, which runs from weeks for paid search to several months for SEO and content. But you don't need to wait that long to find out whether the engagement is set up to work. That part shows inside the first 30 days. By then your company should own the logins, have a written baseline of where quote requests come from today, have a dated plan, and have tracking that counts quote requests rather than clicks. If those four things aren't in place by day 30 or so, more months won't fix it. If they are, judge the results on the channel's clock, not on your patience.
What can you actually see at day 30, 60 and 90?
It depends on the channel, and most frustration with a new agency comes from judging one channel on another's clock.
Paid search produces data within days: impressions, clicks, and the searches that triggered your ads. What takes longer is the bidding. Google's help page on the learning period says an automated bid strategy can take "a few conversion cycles (1-2 typically)" to calibrate, and defines a conversion cycle as the time it takes a click to become a conversion. For a manufacturer whose buyers take a couple of weeks between the first click and a quote request, that's a month before the bidding settles, and only if the account counts quote requests as its conversion and enough of them arrive for the bidding to learn from. At a handful a month it won't settle, and a sensible agency says so and bids differently. On a budget that buys a few hundred clicks a month, there should be a cost per quote request by day 60 to 90, and a direction it's moving in. On a small budget, two dozen clicks a month can't tell a bad campaign from bad luck, and a campaign can run two or three months with no quote at all. There, judge the search terms and the landing page instead, because the sample can't tell you yet.
SEO and content run on months. Google's own starter guide says some changes take effect within hours and "others could take several months." Google's page on hiring an SEO doesn't put a number on it at all. It tells you to ask the SEO what results they expect, and in what timeframe. At day 90, what you can fairly ask for is shipped work you can point to and early movement in Search Console impressions on the terms in the plan. Impressions move well before clicks and quote requests do, so a program can look flat on leads and still be working. Search Console data normally arrives within two to three days, so "we don't have data yet" isn't an answer at day 90.
A website rebuild has its own schedule. There's nothing to measure until launch, so you judge the project on its milestones, and the results clock starts on launch day. That's why the baseline has to be captured before launch, not after.
What should be in place by day 30, whatever the channel?
Four inputs. None of them needs marketing knowledge to check, and none of them depends on results, so none of them should be late.
Access your company owns. Your company should hold the top role on its own accounts: Owner in Search Console, Administrator on the analytics account and Admin on your own Google Ads account, with the agency added as a user in the first two. In Google Ads the agency usually links your account to its manager account rather than taking a seat in it. Google's page on manager accounts says anyone with admin access on your account can end that link at any time. Google's advice when an SEO offers an audit is to grant read access to Search Console, not write. If the agency set anything up under its own login, fix it now while the relationship is new. The steps for analytics are in what to do when you can't see your own analytics.
A written baseline. How many quote requests arrive and how each one reaches you: the website form, a phone call, an email straight to an estimator, a rep or distributor, a trade show. Few companies can rebuild a year of that from memory, so start from whatever the quote log, CRM or ERP already records, and count by source from now on. Any request that goes straight to an estimator's inbox never shows up in analytics, so the baseline has to come partly from the people who receive the requests, and a good agency will ask for it. One that builds it only from Google Analytics will report on whatever reached the form.
A dated plan. What ships in months one, two and three, by name. These pages, these campaigns, this tracking fix. "Ongoing optimization" isn't a line on a plan.
Tracking that ties to quote requests. Form submissions counted as conversions, calls tracked if your buyers call, and a way to match what the report says against what your estimators actually received.
How do you check the work without becoming a marketer?
Three checks, about half an hour once a month.
Open the change history in Google Ads. It lists changes to the account, campaigns and ad groups over the past two years, with the email of whoever made each one in the interface. If the report says the account was tuned every week and the history is quiet for a month, believe the history.
Hold the plan next to what shipped. Each line from the month-one plan should be done, in progress with a date, or dropped with a reason.
Ask your estimators first. Count what reached them this month, against the baseline, before you open the agency's report. You'll already know the number the report should arrive at, and whether it does.
When is it time to fire a marketing agency?
Patience is right for slow channels and wrong for missing inputs. These are the signals I'd treat as a decision point rather than a phase:
- Day 45 has passed without access or a baseline, and the agency hasn't asked you for them in writing or told you what it's blocked on. The delay is often on your side, but chasing it is the agency's job.
- The change history doesn't match what you're being billed for.
- You've asked in writing for quote requests by source, and the next report still stops at traffic and clicks.
- The plan at day 90 is the plan from day one, and the agency can't tell you what two months of data changed.
- The explanation for slow results can't be checked. "SEO takes time" is true. "These twelve pages went live on these dates, and impressions on these terms are up since" can be checked, and it's the kind of answer worth waiting on.
The pattern behind most of these isn't a slow channel. It's a promise made at signing that the channel could never keep. When I was asked about this recently, I put it like this: "The people that are overpromising and underdelivering, too, don't have the expertise to know when that's happening. Maybe when 6 or 7 months have passed, that relationship is ruined." So test the promise against the channel's real clock in the first month, while it's cheap to say so, rather than in the sixth. If you're on an annual term, find the cancellation notice date now too, because the notice window decides more than the renewal date.
The opposite mistake costs something as well. Leaving an SEO engagement at day 60, when the inputs are in place and the work is shipping on schedule, sets you back. The pages and fixes already shipped stay on the site and keep maturing on Google's clock, but the onboarding restarts, and the next agency often re-plans work that was already under way. If the four inputs are there, give the slow channel its months. If they're not, you already have your answer.
