Ecommerce needs conversion volume before any algorithm can optimise, and it needs enough creative throughput to keep finding winners. Below this floor you get neither, and a retainer takes a share of the budget that the margin cannot absorb.
Most ecommerce accounts we audit are being managed to a number that cannot be spent. Here is what replaces it.
Marketing efficiency ratio is total revenue divided by total ad spend across every channel. It cannot be inflated by double-counting, it does not care about attribution windows, and it is the only revenue figure that reconciles with your store. Platform ROAS stays on the dashboard as a directional signal for optimising within a channel — it is not the number that decides your budget.
Revenue is not profit. We build your unit economics in properly: cost of goods, payment fees, shipping, packaging, returns and refunds. What comes out is the amount a sale actually contributes, which is what sets the ceiling on what you can pay for a customer. Brands routinely discover their "profitable" 3x ROAS product loses money once returns are counted.
A blended ROAS that looks healthy is often just your existing customers buying again through retargeting. That is not growth, and it is not something you needed to pay Meta for. We separate first-order acquisition cost from repeat revenue so you can see what new customers actually cost.
If customers reorder, the acquisition cost does not have to be recovered on the first order. Knowing the real repeat rate and the interval lets you decide how long you are willing to wait, which is usually the difference between a brand that scales and one that stalls at a profitable but small number.
Before we touch bidding we verify what the account is actually being told, because optimising to broken signal is worse than not optimising at all.
Browser pixels lose events to tracking prevention, ad blockers and consent refusals. A server-side Conversions API recovers a large share of that, but only if event IDs are matched so browser and server events deduplicate. Done wrong you get inflated conversion counts, a ROAS that looks great, and scaling decisions built on a fiction. We check the event match quality rather than assuming a green tick means correct.
In Shopping and Performance Max, the feed is the campaign. Titles, product types, GTINs, brand fields, image quality, custom labels for margin tiers and stock levels — these decide what you show for and what you are willing to pay. Most feeds we inherit are whatever the store platform exported on day one, which leaves a lot of cheap performance unclaimed.
In the UK, EU and increasingly in US states, your consent banner directly determines how much conversion data reaches the platforms. Consent Mode has to be configured, not merely installed. When a client tells us conversions fell off a cliff and nothing changed, a banner update is the first place we look.
Audience selection matters far less than it did. What decides a Meta account now is how many genuinely distinct concepts you can put into the auction and how fast you can replace the ones that fatigue.
Twenty crops of the same product shot is one concept, and the algorithm treats it that way. A real testing roadmap moves through different angles — problem, demonstration, comparison, social proof, founder, unboxing, objection handling — because a losing angle cannot be rescued by a better thumbnail.
Rising frequency with falling click-through and a climbing cost per acquisition is creative fatigue, and it has a predictable arrival time per account. We track it per concept so replacement is scheduled rather than reactive, which is what stops the monthly cycle of panic when the account suddenly stops working.
We write copy and direct concepts, and we can work with your studio, your UGC creators or your in-house designer. Full production is scoped separately. If creative throughput is your bottleneck, we will say so in the audit, because no amount of account management fixes a starved creative pipeline.
A real audit of your live campaigns — structure, tracking, creative and where the budget is leaking. Free, no obligation, and yours to keep even if you never work with us.
Because ROAS is a platform's own claim about its own contribution, and every platform counts the same sale. Add up the claimed revenue across Meta, Google and email and it usually exceeds total store revenue. We use it inside a channel and judge the business on blended MER and contribution margin.
We need unit economics: cost of goods, shipping, payment fees and a realistic return rate. Not your P&L. Without those numbers nobody can tell you what a customer is worth, and any agency that does not ask for them is guessing.
Meta and Google — Shopping, Performance Max, search and YouTube — as the core. TikTok where the product and audience justify it. We will tell you when adding a channel would just thin out the budget.
Weeks one and two are audit, tracking and feed work, and the numbers move around. The first honest read is week four. Judge it fairly at 90 days, which is how long it takes to find winning creative and scale behind it.
Yes, and it is usually the highest-return work in the first month. Feed rules, title structure, custom labels for margin and stock, and excluding the products that quietly eat the budget.
You do. Ad accounts, pixel and datasets, feeds, audiences and creative files all stay in your name. We are a partner on them and removable in two clicks.
Feed, tracking, structure and creative — a written read on all four within two working days.