Why paid search suits online sales
A shopper who types "waterproof walking boots women's wide fit" has done most of the deciding. The engine shows product listings drawn from the shop's feed, with an image, a price and the shop's name, beside text ads that lead to category pages. Few channels meet demand that close to the order, and none lets the advertiser choose so precisely which products enter which auctions. Engines also let bidding optimise towards the value of each purchase rather than a count of orders, which is the feature an online sales objective needs most.
The limit is the same as the strength. Search captures demand; it does not make it. The number of people shopping for a category in a given week sets how many orders the channel can win, and once the shop appears on most of those searches, extra spend buys vaguer queries and shoppers further from buying. The channel also attracts credit for orders it merely sat in front of: the returning customer who searches the shop's name, the shopper who saw a social ad yesterday and searched today. A plan that reads search on its own reported return will fund it past the point of profit and starve the channels that create the searches.
Paid search tactics for online sales
| Tactic | What it does for online sales | Playbook |
|---|---|---|
| Run feed-led shopping beside text search | Uses product listing ads for product queries and text ads for category and problem queries | Product feeds |
| Bid to margin, not revenue | Sets value targets from break-even ROAS by margin band so bidding never buys unprofitable orders | Smart Bidding strategies |
| Segment brand, category and product queries | Gives each query type its own budget and target because their incrementality differs | Keyword match types |
| Fix the feed before raising bids | Improves titles, attributes, images and price competitiveness, which decide shopping visibility | Shopping and catalogue ads |
| Land each query on the matching page | Sends product queries to product pages and category queries to filtered category pages | Product page optimisation |
| Exclude lines that cannot pay back | Removes out-of-stock, low-margin and high-return products from campaigns | Performance Max structure |
Let the feed carry product queries
Product listings are matched from the product data rather than from a keyword list, so the feed decides which searches each item can enter. Text ads still matter for searches the feed cannot answer well: "gifts for keen cooks", "how to stop condensation", a category with hundreds of variants. Run both and give each the queries it serves best. The product feeds guide covers titles, identifiers and custom labels; at channel level the point is that a shop without a maintained feed is competing for product searches with one hand tied.
Turn margin into the value target
Value-based bidding asks the engine for revenue, and revenue is not what the business keeps. Work out Break-even ROASMinimum ROAS to avoid a loss. for each margin band from gross margin after fulfilment and returns, add the profit the plan requires, and set a separate target per band. A single account-wide target lets the system fill the basket with whatever sells most easily, which in many shops is the lowest-margin stock. The Smart Bidding strategies guide explains how targets are applied and moved in steps; the break-even calculator does the arithmetic.
Give brand, category and product queries their own money
The three query types differ in how many of their orders would happen without the ad. Brand searchers have mostly chosen the shop already. Product searchers have chosen the item and are choosing where to buy it. Category searchers are still choosing what to buy, and that is where the shop meets new customers. One budget shared across all three lets the cheapest orders, usually brand, absorb the money and flatter the return. Separate campaigns, with match types set per segment as the keyword match types guide describes, keep each segment's cost and contribution visible.
Improve the listing before the bid
When a product is not showing, the first question is whether the listing deserves to show. Vague titles, missing attributes, poor images and a price above what rivals charge for the same item all reduce visibility, and no bid fixes them. The shopping and catalogue ads guide sets out what feed quality means across platforms. Treat disapproved and limited products as a weekly check, because an item that cannot enter the auction looks like an item nobody wants.
Match the landing page to the query
A product query should land on that product with the variant the shopper searched for selected. A category query should land on a category page already filtered to what the query named. Sending both to the home page throws away the precision that makes search valuable. The product page optimisation guide covers what the page needs to convert once the shopper arrives.
Take out what cannot earn
Some products should not be advertised at all: lines that are out of stock, lines whose margin cannot cover a click, and lines returned so often that the order is a loss. Exclude them from campaigns or move them to a group with a target that reflects their purpose, such as clearance. The Performance Max structure guide shows how listing groups and exclusions put this into practice.
Budget posture and KPI targets
Search for online sales is funded to a ceiling set by margin and by demand. Fund each segment until marginal return on ad spend (ROASRevenue generated per unit of ad spend.) falls to break-even on contribution: the last pound spent in a segment should still return at least what that segment's margin requires. Because the average return usually sits above the marginal one as spend rises, a segment can look healthy on its blended figure while its last slice of budget loses money, so read the change in contribution when budget changes, not the average.
Keep brand spend separate and small. Its job is to hold the top position against competitors and resellers, not to generate reported revenue. Let seasonal demand set the overall ceiling: search scales with how many people are shopping, so extra budget outside peak weeks buys weaker queries rather than more buyers, and the same budget in a peak week may be too little. Hold a small, fixed test allowance for new query themes, a second engine or a new product range so experiments do not borrow from proven segments. When a segment's impression share (Impression ShareShare of eligible impressions you received.) is already high and its marginal return is at break-even, the next pound belongs in a channel that creates demand, usually paid social.
The tactics in a worked plan
Steepwhistle Leaf Co, an invented online loose-leaf tea retailer
Steepwhistle sells around 240 teas and brewing accessories. Gross margin after fulfilment is about 60% on own-blend teas and 30% on branded teaware, so break-even ROAS is roughly 167% and 333%. It spends £9,000 a month in one automated shopping campaign with a single target ROAS of 350%, and the account reports 420% overall.
| Tactic | Before | After the rebuild |
|---|---|---|
| Feed and text search | Shopping only; titles copied from the site | Titles rewritten by type, origin and format; text ads for "loose leaf tea gifts" and "tea for sleep" |
| Bid to margin | One 350% target | Own blends at 220%, teaware at 400% |
| Query segments | Brand mixed in | Brand campaign at £600 a month; brand excluded from shopping |
| Feed before bids | 31 products disapproved | Disapprovals cleared; price checked against rivals for teaware |
| Landing pages | Category queries to the home page | Category queries to filtered collection pages |
| Exclusions | Out-of-stock and sample packs served | Both excluded; clearance teaware in its own group |
Before the rebuild, brand searches supply £14,000 of the £37,800 reported revenue and new customers place 38% of non-brand orders. Two months later, non-brand spend of £8,400 returns £30,100 at a lower headline ROAS of 358%, but contribution after ad cost on non-brand rises from £4,900 to £8,200 and new customers place 49% of those orders. Own-blend impression share reaches 78%, so the next £1,500 goes to a paid social test.
The headline return fell and the business kept more, which is the pattern a margin-led rebuild usually produces. Separating brand made the non-brand figure honest, and the impression share reading told the shop when search had done what it could.
KPI set for paid search online sales
| KPI | What it tells you | Read how often |
|---|---|---|
| Contribution after ad cost | What search orders leave once product, fulfilment, returns and media are paid | Monthly, per segment |
| ROAS against break-even ROAS by segment | Whether each margin band and query type clears its own threshold | Weekly |
| New-customer share of orders | Whether search is finding buyers or re-selling to existing ones | Monthly |
| Impression share on core category queries | Headroom: whether more buying searches exist at this price | Weekly |
| Average order value from paid search | Basket size, compared with the store average to spot cheap-item bias | Monthly |
Read the set together. A rising return with a falling new-customer share means the account has drifted towards brand and returning buyers. A falling return with rising contribution, as in the example, usually means the mix has moved towards products that earn more. Cost per click (CPCAverage cost of each click.) and AOVAverage revenue per order. explain movements in the return without being targets in themselves. The paid search operating playbook sets the weekly and monthly routines for reading these figures, and the contribution margin guide defines the deductions behind the first row.
When paid search is the wrong route for online sales
Search cannot sell a product people do not yet search for: a new invention, a brand-led category where shoppers look for inspiration rather than items, or an impulse purchase nobody plans. In those cases search captures the small amount of demand other channels create, and the budget belongs in paid social or creator content until searches appear. It is also a poor fit where the shop cannot compete on the results page itself: a reseller of branded goods priced above the manufacturer and larger retailers, or a catalogue of low-priced items where one click costs more than the margin on the order. The margin arithmetic above shows both problems before money is spent.
For the engine-specific build, the Google Ads setup for online sales covers feeds, margin labels, brand exclusions and value bidding.
