PPCleaner

Search Terms Report Missing Data and Where the Cost Went

Published November 6, 202613 min read
An ecommerce owner and marketer reviewing search-query cards beside products and customer orders.
The short answer

The search terms report almost never adds up to campaign cost. Part of the gap is real: Google holds back low-volume terms, and Display, Video and App report no search terms at all. But a large share of the gap usually comes from the export: date segments, filters, removed ad groups, the wrong API resource, or a Performance Max campaign whose cost includes inventory that has no query.

Work through it in order. Match the date range and scope, take out non-search cost, pull the report at the coarsest granularity you can, and only then call what's left "hidden." That number is what you're actually missing. You still can't see those terms, but negatives built from the visible patterns block them too.

Pull a month of search terms, sum the cost column, and compare it with the campaign. The search terms total comes up short. Most threads on this stop at one explanation: Google hides terms for privacy. That's true, but it's only part of the story. In the accounts we audit, a real share of the "missing" cost turns out to be something you did, or something that never had a search term in the first place. You can recover that part in twenty minutes. Only what's left over is truly hidden, and you can't measure the hidden part until you've cleared out everything else.

This article is about the reconciliation itself, not the privacy policy. If you want the policy side and what to do about withheld terms, see Google Ads search terms hidden by privacy, and what to do.

Why the totals never match

The search terms report only lists queries Google is willing and able to show. Campaign cost includes everything you paid for. Four kinds of cost sit in the gap between them:

  1. Scope and timing differences. You're comparing two reports that don't cover the same thing. You can fix all of this.
  2. Cost that never had a query. Display, Video and App placements, and the non-search parts of Performance Max. No search term was ever going to appear for this money.
  3. Granularity loss. Terms that would show over a month but disappear when you split the report by day, device or ad group. You can mostly fix this too.
  4. Withheld terms. Searches Google won't list, usually because too few people searched them. This is the only part that's actually hidden.

Most advice jumps straight to the fourth bucket. That's why so many advertisers believe a third or half of their search terms are hidden when the true figure is smaller. It's also why the opposite happens: someone sees a small gap in a Search campaign and assumes PMax has a small one too. The method is the same either way. Remove buckets one to three, then measure bucket four.

Step 1: Make both reports cover the same thing

Before you suspect Google, check that the two reports really match. Most "the totals don't match" problems in shared Google Ads Community threads come down to one of these points.

Date range and time zone. The interface uses the account's time zone. A script, a Looker Studio connector or a third-party tool may pull dates in UTC or in the time zone of whoever set it up. A few hours of offset across a month is small, but on a single day it can move a big share of cost from one date to the next. Today's data also isn't complete. Leave out the current day and, for safety, yesterday.

Filters and saved views. The search terms report remembers filters. A forgotten "Clicks > 0" or "Added/Excluded: None" filter quietly drops rows. Terms you've already added as keywords or excluded can disappear from a filtered view even though you paid for them during the period.

Removed and paused entities. If an ad group, keyword or campaign was removed partway through the month, its spend still counts in the account total. But the search terms report view may be set to show only enabled items. Switch the status filter to "All," including removed.

Campaign selection. At account level, campaign cost includes every campaign type. If the account runs a Display remarketing campaign or a YouTube campaign, that spend is in the total and will never show up as search terms. Compare campaign by campaign, not at account level.

Billing versus campaign cost. If your reference number is the invoice, stop. Invoices include invalid-click credits and other adjustments that never show up as a negative cost on any row. Reconcile search terms against campaign cost, not billed cost. What those credits actually cover is explained in Google Ads invalid clicks refund: what you actually get.

None of this is exotic. It's just where most of the gap comes from in the first pass, and every later step is wrong if you skip it.

Step 2: Take out cost that never had a search term

Some campaign cost has no query attached and never will. Separate it before measuring anything else.

Display, Video and App campaigns report no search terms. That's by design, not a gap. Exclude them from the comparison completely.

Performance Max is the hard case. A PMax campaign spends on Search and Shopping, which do produce search terms, and on placements like YouTube, Display, Discover and Gmail, which don't. Its campaign cost blends the two. So a PMax campaign whose search terms add up to 40% of its cost isn't necessarily hiding 60% of its queries. Much of that 60% may have been spent where nobody typed anything.

Split PMax cost by channel before you draw conclusions. Google has been adding channel-level reporting to Performance Max. If your account shows it, use it to get the search-and-shopping portion of the spend. If it doesn't, you can only estimate, and you should say so in any client report instead of passing off an estimate as a measurement. Listing groups add another blind spot on the Shopping side, covered in how listing groups hide wasted spend in Performance Max.

Here's an illustrative PMax month, with numbers chosen to add up:

LineCost
Campaign cost$10,000
Non-search channels (YouTube, Display, Discover, Gmail)$3,500
Search and Shopping cost$6,500
Search terms listed in the report$4,000
Search and Shopping cost with no listed term$2,500

Check: $3,500 + $4,000 + $2,500 = $10,000. If you compared $4,000 with $10,000, you'd say 60% is hidden. Compared with the right base, it's $2,500 out of $6,500, about 38%. That's still a lot, but it's a different problem with different fixes.

Dynamic Search Ads do produce search terms. But their cost reporting is split across the search terms report and the dynamic ad targets report, and the two don't always line up neatly. If a Search campaign mixes DSA and keyword ad groups, reconcile the DSA ad groups on their own. Dynamic Search Ads negative keywords that actually hold covers the control side.

Step 3: Pull the report at the coarsest granularity

Google decides whether to show a term based on the rows you ask for. If you split the same data into smaller pieces, more of it drops below the threshold. This is the step almost nobody mentions, and in our audits it's often where the biggest recoverable gap is.

Here's how it plays out. Say "emergency plumber open now near me" gets a handful of searches over a month. Pulled as one monthly row, it may be visible. Pulled day by day, each daily row is tiny, and some or all of those days can be withheld. Sum the daily export and the term contributes less, or nothing.

The same thing happens with any segment you add:

  • Date segments (day, week). The most common cause, because most scripts and connectors pull daily by default.
  • Device. One monthly row becomes three smaller ones.
  • Ad group or keyword. A term matched by two ad groups becomes two thinner rows.
  • Hour of day, network or conversion action. Every extra segment thins the rows further.

So for reconciliation, pull the widest view you can: the full date range as one period, campaign level, no extra segments. Use that total as your "visible" figure. Then, if you need daily or device detail for analysis, pull it separately and accept that it will add up to less.

An illustrative Search campaign, one month:

StepVisible search terms costGap to campaign cost ($8,000)
Daily export from a script, summed$5,200$2,800
Same period as one row, no segments$6,100$1,900
Status filter set to All (removed ad group included)$6,500$1,500

The first number makes it look like 35% of spend is hidden. After two changes that have nothing to do with privacy, the hidden share is $1,500 out of $8,000, under 19%. These figures are an example, not a benchmark. But the direction holds in every account we've looked at: coarser pulls show more.

Use the right API resource

If you pull through the Google Ads API, the resource you choose matters. campaign_search_term_view reports search terms for Search, Shopping and Performance Max. The older search_term_view is tied to ad groups and does not cover Performance Max, so a PMax campaign will look like it has no search terms at all. Several "data missing after migration" threads come down to a pipeline that was rebuilt on the wrong resource, or one that kept the ad-group-level resource after the account added PMax.

At campaign level there's also one segment fewer than at ad group level. That alone brings back some terms that were split thin across ad groups.

Step 4: Measure what is really hidden

What's left after steps one to three is the withheld portion. Write it down as a dollar amount and a share of searchable cost. This is the number that belongs in a client report, not the raw gap you started with.

In the interface, the bottom of the search terms report usually has total rows. In many accounts there's one for the listed terms and one for the full selection, and the difference between them is roughly the withheld cost. Use it as a cross-check against your own calculation. If the two disagree by a lot, go back to step one, because a filter is usually still on.

A few things about the hidden portion are worth knowing, because they change how you should read it.

It isn't a random sample. Withheld terms are mostly low-volume, long-tail queries. That's where the oddest matches live: misspellings, very specific job-seeker or DIY queries, other-language variants, competitor-plus-location combinations. So you can't assume the hidden portion converts like the visible one. In our audits it usually converts worse, but how much worse varies too much between accounts to put a number on it. Low volume terms in the search terms report add up explains why the long tail deserves attention even when every single row looks negligible.

It carries conversions too. The campaign's conversion count includes conversions from hidden terms. If you add up conversions from the visible rows and divide visible cost by that, you get a CPA that's neither the campaign's nor the hidden portion's. Calculate three CPAs: campaign, visible, and hidden (hidden cost divided by campaign conversions minus visible conversions). If the hidden CPA is much higher than the visible one, the long tail is where your waste is.

Conversion columns must match. "Conversions" and "All conversions" differ, and so do conversions counted by click time versus conversion time. Recent days also keep picking up conversions as lag catches up. Use the same column and an older date range on both sides, or the conversion side of the reconciliation won't close even when the cost side does.

It changes with match types and broad expansion. Accounts running broad match, AI Max or PMax reach more unusual queries, so more of their spend tends to fall below the threshold. If the hidden share jumped after a match type change, that change is the likely reason. Google Ads AI Max search terms and the controls that work covers that case.

Step 5: Control spend you can't see

You can't add a negative for a term you can't read. But negatives don't work term by term. They work on words and phrases, and the hidden terms share words with the visible ones. That's the lever.

Block the pattern, not the row. Say the visible report shows "plumber salary," "plumber apprenticeship" and "how to become a plumber." Hidden underneath are almost certainly "plumber salary per hour texas," "plumbing apprenticeship near me paid" and a dozen other variants you'll never see. A phrase match negative on "salary" and another on "apprenticeship" block all of them, visible or not. An exact match negative on "plumber salary" blocks one row and leaves the rest running. For the long tail, phrase match does most of the work. Phrase match negative keyword examples you can copy safely has lists that hold up across trades.

Watch out for broad negatives. They seem like the obvious tool for catching unseen variants, but broad negatives behave unpredictably across match types. A broad negative can block more, or less, than you expect, and you won't see what it stopped. Use phrase unless you've tested the broad version.

Read the visible report as a sample of vocabulary, not of volume. Pull n-grams (the single words and two-word pairs) from the visible terms and rank them by cost and conversions. A word that shows up in several non-converting visible terms is very likely in non-converting hidden terms too. That's how adding negative keywords from search terms without guessing works in practice.

Don't block a word that also carries buyers. The cost of a pattern negative is that it's blunt. "Free" is junk for a law firm offering paid consultations and essential for one advertising free case reviews. Check every candidate word against converting terms before adding it.

Tighten where the hidden share is highest. If one ad group or one asset group accounts for most of the withheld cost, that's where the match is loosest. Narrowing the keyword match type there, or splitting the ad group, reduces how much spend reaches the long tail at all.

Know the limits per campaign type. Search and standard Shopping accept negative keywords and shared negative lists. Performance Max accepts up to 10,000 campaign-level negative keywords and no shared lists, so a list you maintain for Search has to be added to each PMax campaign separately. Performance Max negative keywords not working? Here's why covers what those negatives do and don't stop.

What PPCleaner can and can't do here

PPCleaner reads the same campaign_search_term_view data you do, every night, across Search, Shopping and Performance Max. It can only judge terms Google reports. It can't see withheld terms, and neither can any other tool using Google's API. Be suspicious of any tool that claims otherwise.

What it does is pick out the recurring words and phrases in non-converting visible terms and add negatives that also catch the variants underneath. It never blocks a term that converted in the last 90 days, it doesn't touch bids, budgets, ads or keywords, and every block can be undone in one click for 90 days. That lets you act on the vocabulary of the visible report without worrying that a pattern negative quietly cut off a buyer.

A reconciliation you can repeat every month

Run this checklist once a month per campaign. Keep the results in one sheet so the hidden share becomes a trend you track instead of an argument you have every time.

  1. Set the scope. Same campaign, same full date range, account time zone, excluding the last two days. Status filter on All. No other filters.
  2. Record campaign cost. From the campaigns report, not the invoice.
  3. Subtract cost with no queries. For PMax, take out non-search channel cost if your account reports it. Note it as an estimate if it doesn't. Exclude Display, Video and App campaigns completely.
  4. Pull visible search terms at the coarsest level. One row per term for the whole period, campaign level, no segments. Use campaign_search_term_view if pulling through the API.
  5. Calculate hidden cost. Searchable cost minus visible cost. Record it in dollars and as a share.
  6. Split conversions the same way. Same conversion column on both sides. Calculate visible and hidden CPA.
  7. Compare with last month. If the hidden share rose, check for match type changes, new broad keywords, new asset groups or AI Max being switched on.
  8. Act on visible patterns. Pull n-grams from non-converting visible terms, check them against converting terms, and add phrase negatives where the word is safe to block.

The first time takes longer, because you'll find the filters and scripts that have been distorting the numbers. After that it's a fifteen-minute job. For the analysis part of step eight, how to analyze a search terms report in 20 minutes gives the order to work in.

This is worth doing beyond tidy books. If a fifth of your searchable cost sits in the hidden tail and converts poorly, Smart Bidding is still learning from it. The bidding side of that is in Smart Bidding bad data: how junk queries skew your bids. Once you know how big the tail is, you know how much effort the negatives deserve.

search terms reportgoogle ads reportingperformance maxnegative keywordsprivacy thresholds

Frequently asked

Why is my search terms report cost lower than my campaign cost?

Usually for several reasons at once. Your export may cover a different date range, time zone or set of entities than the campaign view. Part of the campaign cost may have no query at all, like the non-search channels in Performance Max. Splitting the report by day or device pushes more rows below Google's threshold. Whatever is left after you correct for those is the cost of terms Google withholds.

Does summing a daily search terms export give the true total?

No. It usually gives a lower total than pulling the same period as one row. Google decides whether to show a term based on the rows you ask for, so splitting a month into days thins each row, and more of them get withheld. For reconciliation, pull the whole period as one row with no extra segments.

Why does my Performance Max campaign show almost no search terms through the API?

Check which resource your pipeline uses. campaign_search_term_view reports search terms for Search, Shopping and Performance Max. The older ad-group-level search_term_view doesn't cover Performance Max, so PMax can look empty. Also keep in mind that PMax cost includes channels like YouTube and Display that have no search terms at all.

Can any tool show me the search terms Google hides?

No. Every tool using the Google Ads API sees the same reported terms you do. What you can do is block the hidden terms indirectly: phrase match negatives built from words that recur in non-converting visible terms also block the hidden variants that contain those words.

Should I reconcile search terms against my Google Ads invoice?

No. Use campaign cost. Invoices include invalid-click credits and other adjustments that never show up on any search term row, so the two will never match exactly even when your reporting is correct.

Do hidden search terms count in my campaign's conversions?

Yes. Campaign conversions include conversions from terms the report doesn't list. Subtract the conversions from visible terms to get conversions from hidden terms, then divide hidden cost by that figure to see whether the hidden tail converts better or worse than what you can see.

About the author

Alain Abulafya

Founder of PPCleaner · Google Ads consultant & media buyer

Alain has worked in digital marketing for more than 15 years, over 10 of them managing and optimizing Google Ads campaigns. He has worked with more than 100 advertisers in e-commerce and lead generation, and he is a Google Partner. He built PPCleaner to automate the search term reviews he used to do by hand for his clients.

Stop the leak

Clean your search terms. Save time and money.

Every night, PPCleaner reviews the searches your campaigns actually paid for, groups them by intent, and excludes the ones that never convert. No spend threshold to wait for, nothing touched but negatives — and one click undoes any of it.

Start 7 days free

Search, Shopping and Performance Max · Cancel in one click

Related articles

Google Ads Search Terms Report Missing Data, Reconciled — PPCleaner