Price Testing on Google Shopping: Raise, Cut or Hold?

Raise prices first on products that sell steadily at or below the market price. Cut only products priced more than 20% above the market, and only after checking that Google compared the same item. Change prices by about 5%, one group of products at a time. Update the feed and the site on the same day, and leave budgets and bid targets alone while Smart Bidding adjusts. Judge each step after 14 days, on gross profit as well as ROAS: a raise can lower ROAS while profit grows.

Price is one part of pricing on Google Shopping, next to discounts, shipping and order value. This guide covers one decision: which prices to move, in which direction, and how to move them so your ads keep working.

Why does a price change hit your ads twice?

Return on ad spend (ROAS) breaks down into three parts: ROAS = conversion rate × average order value ÷ cost per click. A price change moves two of them at once, in opposite directions. A higher price raises your average order value and usually lowers your conversion rate, the share of clicks that end in a purchase. A lower price does the reverse.

That matters because ROAS follows these two parts most closely. In GetProfit data (1,360 store-months, June 2025 – June 2026), ROAS within a store moved together with conversion rate (rho +0.527) and average order value (+0.540). It hardly moved with cost per click (−0.100). Rho shows how closely two numbers move together: the further from zero, the closer the link.

Price elasticity is how strongly demand reacts to a price change. It differs by product, by season and by how many other stores show the same item, so you can’t look it up in advance. That is why you test prices instead of calculating them.

Which products can take a higher price?

In our study of 1.4 million products, we matched Merchant Center price competitiveness data against results for 213,913 products across 51 accounts, over 13 months:

Among products 5–20% above the market, 14.7% still got conversions, with an average ROAS of 801%, close to 855% at market. The drop starts above 20%: 11.7% with conversions and an average ROAS of 674%. Across all 213,913 products, the correlation between price gap and product status was r = −0.075, close to none.

This is an observation, not an experiment: each product had its price for its own reasons. It doesn’t prove that a raise keeps ROAS where it was.

It does show that a moderate premium over the market goes together with an average ROAS close to that of market-priced products. A smaller share of those products sold: 14.7% against 20.4%. Even so, a raise is worth testing.

Where to start:

• Raise: products at or below the market price that sell steadily. They have room to rise before they reach the 5–20% band.
• Hold: products 5–20% above the market that still sell. They already sit in the band where average ROAS stayed close to that of market-priced products.

Which products should get cheaper?

Start with products priced more than 20% above the market that get clicks but few sales. In our study, that band had the lowest share of products with conversions: 11.7%.

Before you cut, check two things:

1. Google has a benchmark for the product. According to Merchant Center Help (About Pricing in Merchant Center Analytics), Google builds the benchmark price from all retailers that sell a product with the same GTIN in Shopping ads and organic listings. To show one, it needs a valid GTIN from you. A product without a benchmark may have no competitor selling that exact item, or it may be missing its GTIN. Our guide to reading benchmark prices shows where the report can mislead you.
2. Google compared the same item. Open the products with the biggest gap and look through them yourself before you change a single price. The portal’s feed check lays your catalogue out along Google’s benchmark and warns that Google may have compared your product with a different size or configuration.

Hold prices on products that are already far below the market. In our study, products more than 20% below it had the lowest average ROAS of all five bands, 552%. Being much cheaper did not go together with better results. A deeper analysis of whether being cheaper wins looks at the same 213,913 products.

Google also suggests sale prices. The Merchant API reference (REST Resource: accounts.reports) lists a predicted change in impressions, clicks and conversions for each suggested price. Google’s help page on pricing calls these suggestions directional guidance and says the predictions don’t guarantee results. Treat a suggestion as a candidate for a test, not as an answer.

If a product gets no impressions at all, price may not be the first thing to fix. In our study of 1.4 million products, price explained only a small part of which products sold.

Why does ROAS misjudge a price change?

ROAS counts revenue. Profit depends on margin: what is left of a sale once you take out the cost of goods. A price change moves margin far more than it moves revenue, and ROAS leaves margin out. The portal works from your ad account, which usually holds no cost-of-goods data, so you add margin yourself.

Example store, not client data.

A tableware store gets 8,000 clicks a month for 40,000 of ad spend. The clicks bring 300 orders at an average of 600, so revenue is 180,000 and ROAS is 4.5. The gross margin is 35%: each order costs 390 in goods and leaves 210 of gross profit, 63,000 a month.

The store then raises prices by 10%. An order is now 660, and the cost of goods stays at 390, so each order leaves 270. Here is the same month after the raise, with the same clicks and the same spend:

At 240 orders, ROAS falls from 4.5 to 3.96, and the report reads like a failed test. Yet gross profit is 64,800, up from 63,000. Revenue drops below its old level once orders fall by about 9%. Gross profit only drops below its old level at 233 orders or fewer, a fall of more than 22%.

A cut works the other way round: if the same store lowers prices by 5%, an order is 570 and leaves 180. To keep 63,000 of gross profit, it needs 350 orders, 16.7% more. At 316 orders, revenue is 180,120 and ROAS still reads 4.5. Gross profit is 56,880, 9.7% less than before.

So a raise looks worse in ROAS than it is in profit, and a cut looks better. Read every price test on two numbers: ROAS for the ads, gross profit for the store.

How do you test prices without breaking Shopping ads?

Merchant Center Help (Price [price]) requires the price in your product data to match the price on the landing page and at checkout. Any user in the target country must be able to buy the product at the price you submitted. Google disapproves products that don’t meet these requirements. That rules out the classic split test, where some visitors see one price and others another.

The strongest set-up combines a product split with a before-and-after comparison. Compare how the test group changed with how the control group changed over the same days. The season hits both halves, so the difference between them shows the effect of the price more clearly.

Test on groups, not single products. In our study of 1.4 million products, only 17.6% of advertised products got even one conversion in 13 months. One product rarely brings enough sales to judge its price.

Change the feed and the site together. In the same Merchant Center Help article, Google recommends turning on automatic item updates for price. These updates only happen when Google crawls your landing page, and you still need to resubmit your product data regularly.

How do Smart Bidding and Shopping ads react to a price change?

The bid strategy status won’t flag a new price. Google Ads Help (Duration of the learning period for campaigns and what affects it) lists what puts a bid strategy into the “Learning” status: a new strategy, a setting change, or a change in which campaigns, ad groups or keywords it covers. In some cases, an ad group target change in Shopping campaigns does too. A new price in your feed isn’t on that list.

Google adds that its algorithms keep learning after the status no longer shows “Learning”. So expect bidding to adjust to your new conversion rate and average order value as conversions arrive, without any warning.

Give each step time. For a change to the bid strategy itself, the same help page says calibration can take up to 3 weeks or 1–2 conversion cycles. Google gives no such figure for a price change.

The portal advises cutting prices by about 5% once every two or three weeks, which gives each step about that long. A raise moves the same two numbers, conversion rate and average order value, so give it the same rhythm.

Keep your bid target where it is while a price test runs. If you also change your Target ROAS in the same window, the two effects mix and you can’t tell which one moved the result. When you adjust it later, keep each step within 15% and change it no more than once every one to two weeks. The portal counts a larger Target ROAS step as sharp.

Two more Google mechanics affect a cut:

• Sales lasting a few days. For a sale of one to seven days where you expect the conversion rate to jump, a seasonality adjustment lets you warn Smart Bidding in advance. The Google Ads API guide (Create seasonality adjustments) says adjustments may not work as well over periods longer than 14 days, so don’t use one for a permanent price change. For a temporary discount, Google’s price requirements point you to the sale price attribute.
• The “Price drop” badge. According to Merchant Center Help (About price drop annotations), if you lower a price below the average you listed over the past 60 days, Google can show a “Price drop” badge. Google designed it for a significant drop from a fairly stable price. Its help page names no threshold, so small steps may never qualify. Decide which you need: a test in small steps, or one clear cut that shoppers can see.

A price test, step by step

1. Pull two lists. Cut candidates are products priced more than 20% above the market that get clicks but few sales. Raise candidates are products at or below the market price that sell steadily.
2. Check the matches. Open the products with the biggest gap. Remove any that Google compared with a different size, set or configuration.
3. Split each list into test and control. Pick products that are alike in category, price level and sales. Change the price on one half and leave the other as it is.
4. Write down the baseline. For each half: orders, revenue, ad spend, ROAS and gross profit for the period before the change.
5. Change prices by about 5%. Update the feed and the site on the same day. Leave budgets and bid targets on these campaigns as they are while the test runs.
6. Read the first result after 7 days. Treat it as preliminary: the first days after a change are noisy.
7. Judge after 14 days. Compare test with control on ROAS, orders and gross profit.
8. Take the next step or roll back. If the step worked, repeat it two or three weeks after the last change. If it hurt, restore the old price.

How do you read the result after 7, 14 and 30 days?

The portal judges changes in your ad account over fixed windows. Under 7 days, or with too few conversions, it is too early to judge. At 7 days the portal gives a preliminary conclusion.

At 14 days it gives a verdict. The change worked if ROAS after it is more than 1.2× the level before and conversions are at least 0.8× the level before. It harmed if ROAS falls below 0.8× or conversions fall below half. Anything in between is neutral.

After 30 days the portal stops judging a change, because too many other causes pile up. The portal’s change log covers settings inside Google Ads: budgets, bids, conversion goals, product sets and campaigns. A price lives in your feed, so apply the same windows to your price test yourself.

Then check margin, which ROAS leaves out. In the tableware store example, a 10% raise that leaves 240 of 300 orders brings ROAS to 0.88× and orders to 0.8× of the level before. By the verdict rule, that is neutral, yet gross profit grows by 1,800 a month. The 5% cut with 316 orders is also neutral by that rule, while gross profit falls by 6,120 a month.

Price is not just a feed field: Google uses it to decide where and when to show a product. Products priced above the market: your price, the market one, the difference and Google’s recommendation. The portal changes nothing without your consent.

Check your feed →

Sources

• Price [price] — the price in product data must match the landing page and checkout; any user in the target country must be able to buy at the submitted price; products that don’t meet the requirements are disapproved; automatic item updates happen only when Google crawls the landing page; the sale price attribute for sales. Checked 2 October 2026.
• About price drop annotations — the “Price drop” badge for a price lower than the average of the past 60 days; designed for a significant drop from a fairly stable price. Checked 2 October 2026.
• About Pricing in Merchant Center Analytics — the benchmark price comes from retailers selling the same GTIN and needs a valid GTIN; sale price suggestions are directional guidance and predictions don’t guarantee results. Checked 2 October 2026.
• REST Resource: accounts.reports — Merchant API — the benchmark price, the suggested price and predicted changes in impressions, clicks and conversions per product. Checked 2 October 2026.
• Duration of the learning period for campaigns and what affects it — the reasons for the “Learning” status; algorithms keep learning after it ends; up to 3 weeks or 1–2 conversion cycles after a bid strategy change. Checked 2 October 2026.
• Create seasonality adjustments — adjustments suit short events of 1–7 days and may not work as well over more than 14 days. Checked 2 October 2026.
• GetProfit study: 213,913 products across 51 accounts, 13 months — share with conversions and average ROAS by price against the market; r = −0.075; 17.6% of advertised products with conversions.
• GetProfit data: 1,360 store-months, June 2025 – June 2026 — how ROAS moved with conversion rate, average order value and cost per click within a store.
• GetProfit portal rules — verdict windows of 7, 14 and 30 days and the verdict rule; Target ROAS steps of up to 15%; price cuts of about 5% every two or three weeks.