Tabellab

Product Feed Quality Audit: Title Length, Missing GTINs, Price Mismatches and the Revenue at Risk

Lesson 1 of 20Lab

All lessons
  1. Product Feed Quality Audit: Title Length, Missing GTINs, Price Mismatches and the Revenue at RiskLabFree
  2. Feed Health by Category: Approval Rate, the Overall Rate from Totals and the Clicks Lost to Disapproved ItemsLab Locked
  3. Reading the Product-Group Report: ROAS, Cost Share, Value Share and the Share GapLab Locked
  4. Break-Even ROAS: Why the Same ROAS Is a Profit in One Category and a Loss in AnotherLab Locked
  5. From ROAS to POAS: Gross Profit after Refunds, Cost of Goods and Fees per Euro of Ad SpendLab Locked
  6. When ROAS and POAS Disagree: Two Rankings with RANK.EQ, Places Lost and the ROAS TrapLab Locked
  7. Product Groups by Performance: Heroes, Villains, Sidekicks and Zombies with Nested IFs and SUMIFSLab Locked
  8. Custom Labels by Margin Band: Approximate XLOOKUP, SUMIFS and SUMPRODUCT for POAS per LabelLab Locked
  9. Setting ROAS Targets from Profit Goals: Target POAS ÷ Margin, Rounded Settings, Gap and ActionLab Locked
  10. Performance Max Asset Groups: ROAS, Spend Share and an Asset Coverage Check with IF and ANDLab Locked
  11. Where Performance Max Spends: Cost on Feed Products vs Other Placements, a Guardrail and the Weekly TrendLab Locked
  12. New vs Returning Customers: New Share of Orders, Cost per New Customer and New-Customer ROASLab Locked
  13. What a New Customer Is Worth: First-Order Profit, 12-Month Profit and the Maximum Acquisition CostLab Locked
  14. Is Performance Max Incremental? A Regional Pause Test with Control Regions and Incremental ROASLab Locked
  15. Splitting the Budget: Marginal ROAS, Marginal POAS and the Last Budget Step Worth FundingLab Locked
  16. Budget Pacing: Cumulative vs Planned Spend, Run Rate, a Linear Trend and the Budget for the Remaining DaysLab Locked
  17. Price Competitiveness: Price Gap to the Benchmark, CORREL with Click Share and the Cost of Being ExpensiveLab Locked
  18. Testing New Product Titles: CTR Before and After, a Control Index, Adjusted Lift and Extra ClicksLab Locked
  19. What-If Grid: POAS for Every Combination of ROAS and Margin with Mixed References and a Heat MapLab Locked
  20. Final Project: Category Dashboard with ROAS, Gross Profit, POAS, New-Customer Share, a Flag and a Combo ChartLab Locked

You are not signed in — the lab works, but your progress is not saved. Sign up

Lesson 1 of 20 is free. Unlock all 20 hands-on labs, the final exam and your certificate.

Buy course — $9.90
Loading the spreadsheet…

Why the product feed decides what your Shopping ads can do

Most people who start with Google Shopping or Performance Max look at bids, budgets and ROAS targets first. Those levers matter, but they sit on top of something more basic. A Shopping ad is not written by hand. It is assembled from the product feed: the list of products with titles, prices, images, availability and identifiers that an online shop sends to Google Merchant Center. If a product is missing from the feed, it cannot be advertised. If its data are wrong, it may be disapproved. If its title says almost nothing, the system has little to match against what people search for.

That makes the feed the cheapest place to improve results, and the most often neglected one. A campaign manager can spend weeks tuning targets while a fifth of the catalogue never enters an auction because of a missing barcode number or a price that differs between feed and shop. This free lesson shows how to run a first feed audit in a spreadsheet: three checks per product, a status, and two summary figures that tell you how much of the range, and how much of the revenue, is affected.

The example is Pebblewick Outdoor, a fictional online shop for hiking and camping gear. The shop, its products and all numbers in this article and in the course are invented for learning. The thresholds are illustrative, not official platform rules, and nothing here is affiliated with or endorsed by Google.

From shop data to a Shopping adShop systemproducts, prices,stock, imagesProduct feedtitle, GTIN, price,availability, labelsMerchant Centerchecks every item:approved or notShopping campaignproduct groupsPerformance Maxasset + listing groupsDisapproved itemscannot be advertisedThe audit of this lesson happens here:title length, GTIN, price match,before the feed is sent.
The feed sits between the shop and every product ad. An error in the feed limits both campaign types, whatever the bids are.

What a product feed contains

A feed is a table with one row per product (or per variant: each size and colour is its own item) and one column per attribute. Shop systems such as Shopify or WooCommerce can export it automatically, and many shops refine it with a feed tool or a spreadsheet. The attributes that matter most for the checks in this lesson are:

  • Title: the most visible text of the ad and the main signal for matching searches. Good titles usually name the brand, the product type and the attributes people search for, such as size, colour, capacity or material.
  • GTIN: the Global Trade Item Number, the number under the barcode. It identifies the exact product across shops, which lets the platform group identical offers and compare them.
  • Price: must be the price a shopper actually sees on the product page. Feed and shop page are compared, and a mismatch can lead to disapproval.
  • Availability and image: out-of-stock items stop showing, and poor or missing images cost clicks.
  • Product type and custom labels: your own grouping fields. They do not affect eligibility but later decide how you can structure campaigns, for example by margin band (lesson 8 of the course).

Notice what is absent: there are no keywords. In a text-ad search campaign you choose the words you want to appear for. In Shopping the system decides, based largely on the title, the product category and the landing page. A short, generic title is therefore not a cosmetic problem; it removes information the system needs.

Three checks that catch most problems

A complete feed specification has dozens of attributes. For a first audit, three checks give a good return on a few minutes of work, and each one is a simple spreadsheet test:

  • Is the title long enough? Count the characters with LEN and compare with a minimum. Pebblewick uses 30 characters as an internal rule of thumb. The point is not the number itself; a title under 30 characters almost never contains brand, product type and a distinguishing attribute.
  • Is a GTIN in the feed? A simple Yes/No column, exported from the shop system.
  • Does the feed price equal the shop price? Compare the two columns with the <> (not equal) operator.
The 30-character minimum and the three checks are a teaching example. Platform requirements and recommendations change, and Merchant Center reports the authoritative list of issues for a real account. The spreadsheet logic stays the same whichever rules you plug in.

Worked example: twelve Pebblewick products

Here is the feed extract used in the lab above, with the three calculated columns already filled in. The last 30 days' revenue per product comes from the shop system and will be used to weight the problems.

SKUTitle in the feedGTINFeed priceShop priceRevenue 30 dTitle lengthIssuesStatus
PW-101Pebblewick Trailhead 2-Person Tent Green 2.4 kgYes€249€249€8,964470Ready
PW-102Tent 3 personYes€189€189€2,268131Fix
PW-103Pebblewick Alder Down Sleeping Bag Regular BlueNo€179€179€5,370471Fix
PW-104Hiking boots menYes€139€149€4,170162Urgent
PW-105Pebblewick Ridge 45 L Trekking Backpack Unisex GreyYes€129€129€7,482510Ready
PW-106Gas stoveNo€39€44€1,13193Urgent
PW-107Pebblewick Ember Titanium Cooking Pot 900 mlYes€34€34€1,836440Ready
PW-108Pebblewick Cairn Merino Hiking Socks 2-Pack Size 39-42Yes€19€19€2,717540Ready
PW-109Headlamp 400 lmYes€29€29€1,653151Fix
PW-110Pebblewick Drift Insulated Sleeping Mat LongYes€99€109€3,960441Fix
PW-111Pebblewick Summit Carbon Trekking Poles Pair 110-130 cmYes€89€89€3,204550Ready
PW-112Rain jacket womenNo€119€119€2,856172Urgent

Three formulas produce the calculated columns. They are written once in row 4 and copied down to row 15. The minimum title length stands in cell L4, so it can be changed in one place.

  1. Title length (G4): =LEN(B4). LEN counts every character including spaces. "Hiking boots men" has 16 characters.
  2. Issues (H4): =IF(G4<$L$4,1,0)+IF(C4="No",1,0)+IF(D4<>E4,1,0). Each IF returns 1 when its check fails and 0 otherwise, and the three results are added. The dollar signs in $L$4 keep the reference on the minimum when the formula is copied down.
  3. Status (I4): =IF(H4=0,"Ready",IF(H4=1,"Fix","Urgent")). The inner IF is only evaluated when the first test is false, so the three outcomes are: no issue, exactly one, two or more.

Walk through a few rows. PW-104 is called "Hiking boots men": 16 characters, below the minimum, so the first IF returns 1. It has a GTIN, so the second returns 0. The feed says €139 while the shop page says €149, so the third returns 1. Two issues: Urgent. PW-106 ("Gas stove") fails all three checks. PW-103 has a descriptive title of 47 characters and matching prices but no GTIN: one issue, Fix. PW-110 looks perfect at first sight, with a 44-character title and a GTIN, but the price was raised in the shop and not in the feed.

Title length per product vs the 30-character minimumReadyFixUrgentMinimum title length (30)47PW-10113PW-10247PW-10316PW-10451PW-1059PW-10644PW-10754PW-10815PW-10944PW-11055PW-11117PW-1120102030405060Characters in the feed title
Five titles stay below the 30-character minimum. Bar colours show the final status, which also depends on the GTIN and the price check: PW-103 and PW-110 have long titles and still need a fix.

From issue counts to a status

Why turn a number into a word? Because a status is something a team can act on. "Ready" means the product can go into the campaign. "Fix" is a to-do with one clear action. "Urgent" marks products with several problems at once, which are the likeliest to be disapproved and the least likely to perform even if they are approved. You could use more levels or other names; the technique of nesting one IF inside another stays the same, and with many levels the IFS function does the same job in a flatter form.

The count treats the three checks as equally serious. In practice they are not: a price mismatch can take a product offline, while a short title only weakens it. A natural refinement is to give each check a weight, for example 3 for the price, 2 for the GTIN and 1 for the title, and to add the weights instead of the ones. The formula barely changes, which is the advantage of building the audit from small, readable parts.

Ready share: how much of the range can go live

The first summary figure is the ready share: products with the status Ready divided by all products. COUNTIF(I4:I15,"Ready") counts the ready rows and COUNTA(A4:A15) counts the SKUs, so the formula is =COUNTIF(I4:I15,"Ready")/COUNTA(A4:A15). For Pebblewick that is 5 ÷ 12 = 41.7 %. Fewer than half of the products in this extract are fit to advertise without a correction.

Using COUNTA for the denominator instead of typing 12 means the formula still works when products are added to the list. It is a small habit that prevents a common spreadsheet error: a hard-coded number that silently becomes wrong.

Revenue at risk: weighting the problem by money

Counting products gives every item the same weight. The tent that sells for €249 and the socks that sell for €19 are one product each. To see what is at stake commercially, weight by revenue. SUMIFS adds a column for the rows that meet a condition, and the condition "<>Ready" means "status is not Ready".

The formula =SUMIFS(F4:F15,I4:I15,"<>Ready")/SUM(F4:F15) divides the revenue of the seven products with issues by the total: €21,408 ÷ €45,611 = 46.9 %. The split by status looks like this:

StatusProductsShare of productsRevenue 30 dShare of revenue
Ready541.7 %€24,20353.1 %
Fix433.3 %€13,25129.1 %
Urgent325.0 %€8,15717.9 %
Total12100.0 %€45,611100.0 %

30-day revenue by feed status€0€5,000€10,000€15,000€20,000€25,000€24,20353.1 % of revenueReady (5 products)€13,25129.1 % of revenueFix (4 products)€8,15717.9 % of revenueUrgent (3 products)Revenue
Seven of twelve products are not ready, and they carry 46.9 % of the last 30 days' revenue (€21,408 of €45,611).

"At risk" does not mean this revenue will disappear tomorrow. The figure includes all sales channels and the products are still on sale in the shop. It means that 46.9 % of recent revenue comes from products whose ads are weaker than they could be, or may be switched off. Use it to set priorities, and avoid presenting it as a forecast of losses.

How to read the result and what to fix first

The two shares tell slightly different stories: 58.3 % of the products are not ready, and they stand for 46.9 % of the revenue. Here the products with issues sell a little less than the ready ones on average. In other shops it is the other way round, typically when bestsellers come from a supplier that delivers poor data. That is exactly why both figures are worth calculating.

For the work plan, sort the products that are not ready by revenue, highest first. For Pebblewick the list is:

  1. PW-103 (€5,370, Fix): no GTIN.
  2. PW-104 (€4,170, Urgent): title of 16 characters, feed price €139, shop price €149.
  3. PW-110 (€3,960, Fix): feed price €99, shop price €109.
  4. PW-112 (€2,856, Urgent): title of 17 characters, no GTIN.
  5. PW-102 (€2,268, Fix): title of 13 characters.
  6. PW-109 (€1,653, Fix): title of 15 characters.
  7. PW-106 (€1,131, Urgent): title of 9 characters, no GTIN, feed price €39, shop price €44.

Two kinds of fix on that list are quick. A GTIN has to be requested from the supplier or read off the packaging, and a price difference usually comes from a feed that is refreshed too rarely; a more frequent or automatic update solves it for all products at once. Rewriting titles takes more thought, and lesson 18 of the course shows how to test whether a new title really earns more clicks than the old one.

Common mistakes in a feed audit

  • Treating the threshold as a goal. Padding a title with filler words to reach 30 characters helps nobody. The length check is a cheap detector for titles that lack information; the fix is to add the information.
  • Reversing the comparison. A title is a problem when it is shorter than the minimum: G4<$L$4, not >.
  • Forgetting the dollar signs. Without $L$4 the reference slides down when the formula is copied, and from row 5 on it points at empty cells.
  • Counting products only. A ready share of 90 % sounds fine until the missing 10 % turn out to be the bestsellers. Always add the revenue view.
  • Auditing once. Prices, stock and assortments change every week. A feed check is a routine, ideally with the same sheet refreshed from a new export.
  • Typing results instead of formulas. A status typed by hand does not update when the data change. In the lab the Coach checks that every answer is a formula that still works with other numbers.

What a spreadsheet audit cannot tell you

A spreadsheet sees the data you export. It does not see how the platform judges the landing page, whether an image violates a policy, or which searches a product actually appeared for. Merchant Center lists the real disapprovals and warnings of an account, and the campaign reports show impressions and clicks per product. The audit in this lesson is a first filter that you control yourself and can run before a feed is ever submitted.

It also cannot tell you whether a product is worth advertising at all. A perfect feed entry for a product with a tiny margin and a high return rate can still lose money with every click. That question needs margins, refunds and ad spend in the same table, which is where the course goes next.

How the feed connects to Shopping and Performance Max

Both campaign types draw on the same feed. A Shopping campaign shows product listings and is organised in product groups, subdivisions of the catalogue by category, brand, item or custom label. A Performance Max campaign combines the feed with creative assets (headlines, images, videos) in asset groups and can show ads across several channels, with the product listings as one important part. An item that is disapproved, or that carries a weak title, is handicapped in both.

This is why the course starts here and not with bidding. Automated bidding can only choose among the products and the information it is given. Feed quality sets the ceiling; bids, targets and budgets decide how close a campaign gets to it.

What the rest of the course covers

The free lesson is the first of twenty labs that follow Pebblewick Outdoor from feed to profit. Each lab is a spreadsheet with real formulas to write, checked cell by cell:

  • Feed and structure: approval rates by category, the product-group report, performance tiers (heroes, villains, sidekicks, zombies) and custom labels by margin band.
  • ROAS versus profit: break-even ROAS from the margin, POAS (profit on ad spend) after refunds, cost of goods and fees, and ROAS targets derived from a profit goal.
  • Performance Max: asset-group results and coverage, and how much of the spend goes to product listings.
  • Customers and incrementality: new versus returning customers, the 12-month value of a new customer, and a regional pause test that measures what a campaign really adds.
  • Budget: marginal ROAS and POAS for the budget split, pacing with a run rate and a linear trend, price competitiveness, a title test, a what-if grid and a final category dashboard.

You do not need an ad account to follow along, only basic spreadsheet knowledge. The function names are the English Excel names and work the same way in Google Sheets. All data are fictional, the benchmarks are illustrative, and the course makes no promise about the results of any real campaign: it teaches the arithmetic and the reasoning that let you judge your own numbers.

Frequently asked questions

What is a product feed in Google Shopping?

A product feed is a table with one row per product and columns such as title, price, image link, availability and GTIN. An online shop sends it to Google Merchant Center, and Shopping ads and the product listings of Performance Max campaigns are built from it.

How do I check title length in Excel?

Use =LEN(B4) to count the characters of the title in B4, including spaces. Compare the result with a minimum kept in a cell, for example =IF(LEN(B4)<$L$4,1,0), so a title that is too short counts as one issue.

Why does a price mismatch between feed and shop matter?

The price in the ad must match the price on the product page. When they differ, the item can be disapproved and stops showing. A feed that is updated automatically and often is the usual way to prevent mismatches.

What is the revenue share at risk?

It is the revenue of all products that are not ready, divided by total revenue: =SUMIFS(F4:F15,I4:I15,"<>Ready")/SUM(F4:F15). In the example it is 46.9 %. It shows how much of the business depends on products with feed issues; it is a priority measure, not a loss forecast.

Is 30 characters an official minimum for product titles?

No. It is an illustrative rule of thumb used by the fictional shop in this lesson. A very short title rarely contains brand, product type and key attributes. Platform limits and recommendations change, so check the current guidance for your own account.

Do I need a Google Ads account for this course?

No. Every lab runs in the browser on fictional shop data. You practise the formulas and the reasoning (feed checks, ROAS, POAS, budget split) and can apply them later to exports from your own shop and ad account.

Keep going: Google Shopping & Performance Max for Online Shops: Feed Quality, Product Groups, ROAS vs POAS, Asset Groups, New Customers & Budget Split

This free lab is lesson 1 of 20. Unlock the other 19 hands-on labs, the final exam and your certificate for a one-time $9.90.

  1. Feed Health by Category: Approval Rate, the Overall Rate from Totals and the Clicks Lost to Disapproved Items
  2. Reading the Product-Group Report: ROAS, Cost Share, Value Share and the Share Gap
  3. Break-Even ROAS: Why the Same ROAS Is a Profit in One Category and a Loss in Another
  4. From ROAS to POAS: Gross Profit after Refunds, Cost of Goods and Fees per Euro of Ad Spend
  5. When ROAS and POAS Disagree: Two Rankings with RANK.EQ, Places Lost and the ROAS Trap
  6. Product Groups by Performance: Heroes, Villains, Sidekicks and Zombies with Nested IFs and SUMIFS
  7. Custom Labels by Margin Band: Approximate XLOOKUP, SUMIFS and SUMPRODUCT for POAS per Label
  8. Setting ROAS Targets from Profit Goals: Target POAS ÷ Margin, Rounded Settings, Gap and Action
  9. Performance Max Asset Groups: ROAS, Spend Share and an Asset Coverage Check with IF and AND
  10. Where Performance Max Spends: Cost on Feed Products vs Other Placements, a Guardrail and the Weekly Trend
  11. New vs Returning Customers: New Share of Orders, Cost per New Customer and New-Customer ROAS
  12. What a New Customer Is Worth: First-Order Profit, 12-Month Profit and the Maximum Acquisition Cost
  13. Is Performance Max Incremental? A Regional Pause Test with Control Regions and Incremental ROAS
  14. Splitting the Budget: Marginal ROAS, Marginal POAS and the Last Budget Step Worth Funding
  15. Budget Pacing: Cumulative vs Planned Spend, Run Rate, a Linear Trend and the Budget for the Remaining Days
  16. Price Competitiveness: Price Gap to the Benchmark, CORREL with Click Share and the Cost of Being Expensive
  17. Testing New Product Titles: CTR Before and After, a Control Index, Adjusted Lift and Extra Clicks
  18. What-If Grid: POAS for Every Combination of ROAS and Margin with Mixed References and a Heat Map
  19. Final Project: Category Dashboard with ROAS, Gross Profit, POAS, New-Customer Share, a Flag and a Combo Chart

Buy course — $9.90 See the full course