Ecommerce ads with a hand on margin, not just spend

For most ecommerce accounts we work on a flat retainer or a hybrid base-plus-percentage above a spend threshold, because a pure percentage of spend pays us more as your spend grows rather than as your profit grows. A proper build usually carries a one-time setup fee for the feed, tracking, and account architecture. We scope it against your spend level and margin, and there are no hidden markups, because your ad spend is billed directly by the platform, never routed through us.

Smart Bidding needs roughly 30 to 50 conversions per campaign each month to calibrate, and campaigns exit the learning phase in about two to three weeks. Honestly, first meaningful and stable results land around 45 to 60 days, and campaigns reach a mature state between 30 and 90 days. Early CPAs that look high, then settle over the first couple of months, are the normal trajectory, not a failure. Impatiently resetting learning is the single biggest cause of accounts that never work.

Yes. You have admin access from day one, spend is billed directly by the platform to your card, and your account structure, data, and history are fully portable. Accounts built under an agency MCC with the agency as billing middleman are a red flag we deliberately avoid, so there is nothing to claw back if you leave.

A senior media buyer owns your account and makes every decision that ships. This is the opposite of the common bait-and-switch where seniors sell the account and juniors execute it. AI handles the grunt work, audits, search-term mining, and round-the-clock monitoring, but a human sets the strategy and approves every change. We keep manager-to-account ratios deliberately low so seniors have real time on each account.

This is the sharp question, and it deserves a real answer. Branded search converts with or without an agency, so non-brand is the part actually attributable to us. We set a numeric non-brand ROAS floor with you during the audit, often around 1.5x as a starting line for many ecommerce accounts, and we agree upfront what we do if it's breached. We will not name a guaranteed number before auditing your account, because anyone who does is guessing.

Brand exclusions on PMax from day one, plus a dedicated brand Search campaign to capture that demand cleanly. Letting PMax bid on your own brand terms lets it claim credit for sales you would have won anyway, which inflates the dashboard. Once brand is excluded, the true performance of your prospecting shows up immediately, and it's usually a very different number.

AI accelerates the grunt work: auditing feeds and accounts, mining search terms and placements at full-catalog scale, monitoring pacing and MER around the clock, and drafting feed rules and creative variants. Senior humans own everything that ships: the non-brand ROAS floor, the account architecture, the scale-defend-kill decisions, and the incrementality tests. AI never pushes a change live unsupervised.

We anchor budget decisions to MER, your total revenue divided by total marketing spend, and to contribution margin, using channel ROAS only for tactical ad-set kill or scale calls. That matters because a campaign can report a strong ROAS and still lose money: platforms over-attribute by 30 to 40 percent, and some accounts at 4.0x reported ROAS have real incremental contribution below 1.0x. Healthy DTC MER runs roughly 3.0 to 5.0x but it's margin-dependent, so we set your real floor against your gross margin. These are directional benchmarks, not guarantees.

Feed work is foundational, not an add-on. We rebuild titles in the brand plus product type plus attribute order so the highest-value information shows in the roughly 70 characters that display, add missing GTINs through supplemental feeds, and build custom labels to segment by margin and bestseller. This is the unglamorous craftsmanship most providers skip, and it's often where the biggest gains come from before we touch a single bid.

A short initial commitment, typically three to six months so the account has time to exit learning and reach a mature state, then a rolling 30 to 60 day notice. We avoid the 12-month-plus lock-ins some shops push. The first few months are where the foundation gets built, and resetting that work early is exactly what causes accounts to fail, so the initial term protects your results, not our revenue.