Listing A's ad ROAS dropped from 4 to 1.8. The operator's first reaction is usually to raise the budget, and the second is to change the main image. Both were done, money was spent, time was spent, and the listing still looks the same. The problem is that these two actions work on different funnel stages: raising the budget amplifies traffic efficiency, while changing the main image repairs conversion handling efficiency. Taking action without doing stage-by-stage diagnosis first is equivalent to changing two variables at once, and in the end no one can tell which one had an effect.
First distinguish two things: adding ads amplifies traffic efficiency; optimizing conversion rate amplifies conversion handling efficiency.
Ads do not generate conversions; they only amplify the impressions the platform has already allocated to you. Conversion rate determines whether those amplified impressions still make money. So the order is fixed: first confirm there are no obvious holes in the conversion handling stage, then decide whether to amplify.
Conversely, if conversion handling is already up to standard but you do not add ads, you are wasting traffic efficiency that has already been validated—the platform is willing to give you impressions, clicks and payments are keeping up, but you are not pushing the budget up to capture this share.
Four data watersheds: impressions, clicks, add-to-cart, payment—whichever stage declines beyond baseline is the one to act on first.
The comparison target must be your own 7-day or 14-day rolling baseline for this listing, not the industry average, and not public data from peers. Shopee sites and categories vary too much in scale; applying the same number across stores will only lead to misjudgment.
| Funnel segmentation | What to focus on | How to compare with the baseline | What to do first when below baseline |
|---|---|---|---|
| Impressions | Impression volume, keyword impression share, organic search position | Compare with the previous 7 days under the same keyword group and same campaign type | First check budget, bids, category competition, and major promotion traffic diversion; do not change the page |
| Clicks | Click-through rate (CTR) | Previous 7-day average for the same link and same placement | Change the main image, the first 30 characters of the title, the price anchor, and the sales volume label |
| Add to cart | Add-to-cart rate = Add-to-cart count ÷ Visitor count | Rolling 14-day average for the same link | Check the above-the-fold selling points on the detail page, whether the specifications are complete, and the frequent questions in reviews |
| Payment | Payment conversion rate = Payment count ÷ Add-to-cart count | Compare under the same link and the same shipping plan | Check shipping fees, dispatch time, inventory, and negative reviews |
If only one segment is below baseline, the action is clear; if two or three segments are below baseline at the same time, the bottleneck is upstream, so start fixing from the segment closest to impressions. When click-through rate is low, adding ads buys more expensive traffic, not more orders; when there is a break at the payment segment, adding ads amplifies refunds and cancellations.
The three most common combinations correspond to three different actions
- Impressions are sufficient, but click-through rate is below baseline.Money is being spent but no one clicks, which means the initial gate of conversion has not been opened. Start by improving the main image and title, and also check the price anchor and sales label; hold off on increasing the budget during this period.
- Clicks are normal, but there is a gap between add-to-cart and payment.The add-to-cart rate is fine, which means the product itself is attractive; the drop-off is at the payment stage, most likely due to shipping costs, delivery time, inventory, or missing sizes. First check these fields, then look through the reviews for a cluster of logistics complaints. If this part is not fixed, increasing ads will only amplify refunds.
- All four stages meet the standard; the only issue is insufficient impression share.This is the only situation in which you should increase ad spend. The way you increase it must also be restrained: first raise the budget by a small percentage and observe whether marginal ROAS is still rising, rather than doubling it all at once—doubling it all at once often buys repeated impressions from the same group of people.
On the advertising side, you only need to watch two numbers: spend share and marginal ROAS
Ad spend as a percentage of total sales determines how much gross margin remains after increasing the budget. To calculate this percentage, use the same basis: count only settled orders and deduct cancellations and refunds; otherwise, you will overestimate efficiency.
Marginal ROAS refers to the incremental sales generated by every additional 10% of budget, not overall ROAS. Two counterintuitive situations need to be treated separately: if overall ROAS is high but marginal ROAS is declining, it means the traffic pool is already saturated, and continuing to increase spend will only drive up costs; if overall ROAS looks low but marginal ROAS is rising, it means you are still in the ramp-up stage, and you can observe for another one to two settlement cycles. If the agency operation quote includes an advertising service fee, include this service fee in the spend share as well, so that it can be compared with self-operated cash expenses on the same basis. For the specific breakdown, seeHow to calculate the economics of agency operation vs. self-operation.
Multi-store and agency operation scenarios: if the basis is not unified, this judgment framework will fail
The most common mistake in multi-store operations is using Store A's conversion rate baseline to decide whether Store B should increase ad spend. The two stores differ in site, category, average order value, and logistics plan, so the baselines are inherently not comparable, and any conclusion drawn from the comparison will naturally be wrong.
The minimum approach to unifying definitions has only four rules: same site, same category, same reporting period (a 14-day rolling window is recommended), and same ad type. Turn the baseline deviation values of the four segment metrics into one cross-store dashboard, with one store per row and one segment per column, and mark recommended actions in cells that exceed the baseline. When dividing work across the team, this judgment should be written into the handover checklist rather than left in the operator's personal experience—once the person is replaced, the standard disappears. Unifying definitions is an action that should be completed by the 2-to-3-store stage; for specific stage tasks, refer toStage tasks from single-store validation to multi-store replication.
The same set of metric definitions can be reused on Shopee, Lazada, AliExpress, Mercado Libre, and even Wayfair, provided that business flows are centrally consolidated and login environments and proxy egress are strictly isolated. These two matters belong to different layers, and mixing them will cause problems in both at the same time. For the layered approach, seeCentralized Processing of Orders, Inventory, and Messages, and the key configuration points on the account environment side are inMulti-Store Account Management.
In these three situations, do not add ads or change pages
First resolve the constraints, then return to this judgment.
- Insufficient inventory or key variants out of stock: conversion data is artificially depressed; fixing the page won't help, and increasing ad spend will help even less.
- Store rating or logistics timeliness has already triggered platform restrictions: the impression structure itself has been interfered with, so segmented data is not a valid reference.
- Traffic structure changes drastically before and after major promotions: the platform skews traffic to the major promotion venue, the usual baseline is distorted, so compare again after it returns to a regular cadence.
FAQ
Conversion rate and ad budget: which should you look at first?
First check whether conversion capacity meets the standard. Of the four data segments, if any one of clicks, add-to-cart, or payment is clearly below its own baseline, fix that segment first before adding budget; only when all four segments are above baseline and only impression share is insufficient is it time to add ad spend.
Should the baseline use 7 days or 14 days?
For listings with stable daily order volume and low traffic fluctuation, a 7-day rolling baseline is sufficient; for listings with high average order value, sparse orders, or those currently switching main images for product testing, use 14 days. Do not make cross-period comparisons around major promotions; that data cannot be used as a baseline.
Can the same conversion rate standard be used to judge multiple stores?
No. They must be on the same site, in the same category, over the same statistical period, and for the same ad type to be comparable. The value of a cross-store dashboard is in seeing the direction and magnitude of deviation from the baseline, not in comparing whose absolute conversion rate is higher.
Which fields should an ad monitoring tool focus on?
Look at three things: whether the attribution methodology is consistent with the backend, whether segmented data can be aligned to impressions and clicks, and whether data permissions are tiered by role.What to look for when choosing store operations toolsThe comparison dimensions in it can be directly transferred to the Shopee context.
After deciding to increase the budget, how much should you increase it by at a time?
Increase it in steps of 10% to 20%, and after increasing, observe at least one full settlement cycle before deciding the next step. The advantage of incremental steps is that you can clearly see the trend of marginal ROAS; if you double the budget and still reach the same group of people, costs will immediately rise while order volume stays flat.
