Most disappointment with agencies comes from unmanaged expectations rather than bad work. Here is a realistic shape, and what should worry you at each stage.
Week one: access, diagnosis, rebuild
Access granted, tracking verified against store orders, account history read, structure rebuilt, campaigns live. Three days is enough for this; four is fine. You should receive a written account of what was wrong before anything was changed.
Warning sign: week one ends and nobody has mentioned your conversion tracking.
Weeks two to four: learning, not judging
Google relearns the account. Numbers will be volatile and are not representative. This is the hardest stretch, because it looks like nothing is working and the temptation to change everything is enormous.
Warning sign: big structural changes every few days. That resets learning and guarantees another month of noise. Also: an agency promising results during this window.
Month two: the first honest read
Bidding has settled enough to read. You should now be having a real conversation about which products and campaigns earn and which do not, with budget moving accordingly.
Warning sign: the report is full of impressions and clicks rather than orders against your break-even. Or nobody has mentioned your margin yet.
Month three: scale or stop
By now there should be a clear answer to whether Google can carry this store profitably. If it can, spend grows deliberately against performance. If it cannot, a good agency says so.
Warning sign: month three arrives and the answer is still "give it more time" with no specific reason. Time is not a strategy.
Send access, answer when they ping you, sign off on the plan, and read the weekly update. If the relationship needs more from you than that, something is wrong with how it is set up.
Ours is written out on the process page, including what happens daily and weekly — so you can hold us to it.