Google Search Partner Network in 2026: Where Your Ads Actually Run
You open the placement report and see domains you have never heard of: aggregators, directories, small search engines, apps. Some of those clicks convert, some do not, and in your blended CPA all of it disappears into one number.
This is the Google Search Partner Network (SPN) — properties that show Google ads without being Google. Across 2025–2026 Google added a lot of transparency here: full placement reporting for Search, Shopping and App campaigns, an invalid-activity credits report with adjusted metrics, a third-party pre-screen programme run by independent vendors, and, from 10 February 2026, the removal of parked domains as an SPN surface. This guide covers where the data actually lives, how to evaluate the network’s contribution honestly, and when opting out is genuinely the right call.
What the Search Partner Network is
Search partners are third-party sites and services that have embedded Google search or Google ad units and show your ads on them. That includes:
- sites running an internal site search powered by Google;
- directories, aggregators and vertical marketplaces;
- certain Google-owned surfaces beyond the main search results page;
- apps that use Google search.
The key difference from the Display Network: on search partners your ad appears in response to a query, not next to content. Intent is therefore higher than in banner placements, while the surrounding context is less controllable than on google.com.
Search partners are neither “junk traffic” by definition nor “free extra reach”. They are a distinct source with distinct behaviour, and it has to be measured before its fate is decided.
Isolating search partner traffic in reports
Until you separate this traffic, every opinion about it is guesswork. The sequence:
- Open Campaigns and apply the Network (with search partners) segment. Rows split into Google Search, Search Partners and Display.
- Use a range of at least 30 days — 90 is better, or the partner sample may be too thin to read.
- Record spend, clicks, CTR, conversions, CPA and conversion value for each row.
- Calculate the partner share of spend separately. Everything you do next depends on that number.
- Repeat per campaign — the share is almost never distributed evenly.
Build this once as a saved view in custom columns and the report editor and the analysis takes minutes rather than an afternoon.
Full placement reporting: what to look for
The placement report lists the specific domains where ads appeared, with impression data. It is available for Search, Shopping and App campaigns; for Performance Max, channel segmentation includes a separate search partners line.
What matters when you read it:
- Concentration. Typically 70–80% of partner impressions come from 10–20 domains. Those are the ones to work with; the rest is a long tail.
- Topical fit. Is the property related to your category at all, or is it a general-purpose aggregator of everything?
- CTR anomalies. Extremely high CTR with zero conversions usually means accidental clicks on a badly placed unit, or low-quality traffic.
- Persistence. A one-off spike from a domain is noise. Systematic presence is worth investigating.
What changed in the Search Partner Network in 2025–2026
| Change | What it means | Advertiser action |
|---|---|---|
| Parked domains removed (from 10 February 2026) | Parked domains are no longer an SPN surface and the option was removed from content suitability settings | Check historical reports: if that category used to take spend, compare before and after |
| Full placement reporting | Available for Search, Shopping and App campaigns with per-domain impression data | Make a monthly domain review part of the routine |
| Invalid activity credits report | The report editor now surfaces issued credits with adjusted cost and adjusted clicks metrics | Judge efficiency on adjusted metrics rather than raw ones |
| Performance Max channel segmentation | Channel reporting splits out search partners | Evaluate SPN contribution separately from Google Search inside PMax |
| Third-party pre-screen programme | DoubleVerify, Integral Ad Science and Zefr receive the SPN domain list weekly, classify sites by risk level and provide curated exclusion lists | Relevant for brands with formal brand-safety requirements; set up directly with the chosen vendor |
| Geo-targeting refinements | Location targeting on the network was brought closer to advertiser expectations for area of presence and area of interest | Re-check geo reports if you previously saw spend from non-target regions |
Google also publishes performance uplift figures for automated bidding when the network reaches roughly 5% or more of spend. Treat any such number as a vendor-stated reference point rather than a forecast for your account, and validate it on your own data — ideally through incrementality and geo experiments.
Can you opt out of search partners?
The answer depends on campaign type, and this is where most of the confusion lives.
- Search and Shopping campaigns: yes — uncheck search partners in the campaign’s network settings.
- Performance Max: there is no generally available toggle. During 2026 Google has been testing partner-network selection and channel-level opt-outs with a limited set of advertisers, but that is not a setting you can plan around.
- Account-level placement exclusions: these apply more broadly — across the Search Partner Network, YouTube and display placements. This is the main lever when a full opt-out is unavailable.
- Content suitability settings and account-level negative keywords add a further layer of control.
A method for deciding: opt out or not
Step 1. Quantify the contribution
| Metric | What it tells you | Concern threshold (reference) |
|---|---|---|
| Partner share of spend | How big the bet is | Under 3% — low stakes; over 15% — must be investigated |
| Partner CPA ÷ Google Search CPA | Relative efficiency | Above 1.5 with a material spend share |
| Partner conversion rate | Quality of intent | Less than half of search conversion rate |
| Partner CTR | Placement anomalies | Far above search CTR with no conversions |
| Share of adjusted clicks | Invalid activity | Consistently visible credit volume |
| Segment conversion lag | Whether you are underrating it | Apply the correction method from the conversion window guide |
One caveat that changes conclusions: compare segment CPAs on cold data only. Look at the last three days and partners will almost always look worse simply because their conversions have not matured yet. The correction method is in conversion windows and conversion lag.
Step 2. Prune instead of switching off
- Sort the placement report by spend.
- Flag domains with meaningful spend and zero conversions over a period no shorter than two of your conversion lags.
- Add them to account-level exclusions.
- Clean the query side in parallel — some poor partner traffic arrives on off-target phrasing and is removed by routine work with negative keywords and the search terms report.
- Wait two weeks and re-run the numbers.
Step 3. Prove the effect with a test
If the numbers still disappoint after pruning, do not flip the network off account-wide in one move. Run an experiment on a share of traffic, or opt out in one or two campaigns and keep the rest as control. Watch total conversion volume alongside CPA: an 8% CPA improvement that costs 20% of conversions is usually a bad trade for a growing business.
Also account for the algorithmic side effect — changing network settings is the kind of edit that sends a campaign back into learning. Plan it using the Smart Bidding learning period guide.
What healthy and unhealthy partner traffic looks like
To keep the decision out of the realm of opinion, it helps to have two reference patterns. Below is a segmented 90-day view of two accounts — figures are illustrative, but the proportions come from typical audits.
| Metric | Account A — healthy | Account B — problem |
|---|---|---|
| Partner share of spend | 6% | 23% |
| CTR search / partners | 7.1% / 5.4% | 6.3% / 19.8% |
| Conversion rate search / partners | 4.2% / 3.6% | 3.9% / 0.4% |
| CPA search / partners | $41 / $47 | $44 / $210 |
| Domain concentration | Top 20 = 62% of impressions, on-topic | Top 5 = 71% of impressions, unrelated |
| Invalid activity credits | Occasional | Visible every month |
In account A partners run slightly more expensive than search while contributing real volume — opting out costs conversions and buys almost nothing in economics. Account B is a different story: anomalous CTR against a near-zero conversion rate, an enormous CPA, concentration in five unrelated domains and recurring credits. Even there, though, the first move is not switching off the network — it is excluding those five domains and re-measuring two weeks later. Very often the remaining network then behaves like account A.
Check one more source of bias before concluding: partner traffic frequently carries a different device mix, and a head-to-head comparison then exaggerates the gap. Break both segments down by device first.
Five misconceptions about search partners
- “It’s the same as the Display Network.” It is not. Display places ads next to content; partners place them against a query. Different metrics, different evaluation logic.
- “Always opt out.” That advice is a decade old. You now have placement reporting, account-level exclusions, adjusted metrics and third-party domain screening — use those first.
- “You can opt out in any campaign.” Performance Max has no generally available toggle; extended channel controls have been limited to a test group of advertisers during 2026.
- “A bad partner CPA is obvious immediately.” It is not — the segment can carry a different lag, and fresh-data reads systematically penalise partner traffic.
- “Invalid clicks mean the money is gone.” Part of it comes back as credits, which is exactly why performance should be judged on adjusted metrics rather than raw ones.
When opting out makes sense — and when it does not
| Opt out when | Do not opt out when |
|---|---|
| Spend share is large and CPA is consistently 1.5–2× worse on cold data | Spend share is 1–3% — the decision lands inside the noise |
| Formal brand-safety requirements cannot be satisfied with a placement list | The problem is concentrated in a handful of domains that can simply be excluded |
| Adjusted metrics show systematic invalid activity | The account is already short of conversions for the bidding strategy |
| A narrow niche where partner properties deliver irrelevant audiences | You are reading fresh, unmatured data |
| A test confirms volume barely moves while economics improve | The decision is being made on instinct with no segment and no placement report |
Reducing risk without switching the network off
- Maintain a living account-level exclusion list. Review it monthly — the network’s composition changes.
- Use content suitability settings and, for strict brand requirements, third-party pre-screening through your chosen vendor.
- Report on adjusted metrics. The invalid-activity credits report exists precisely for this: it shows what was already refunded and makes evaluation fairer.
- Keep measurement clean. Some “bad” segments are simply under-counted; verify tagging before passing judgement. Related reading: traffic anti-fraud.
- Separate campaigns by requirement. Keep strict-context campaigns apart from volume-oriented ones instead of compromising both.
The monthly process: who looks at what, and when
A one-off audit produces a one-off result. Only cadence works, and cadence needs a four-point process.
1. A fixed day and a fixed report
Pick one day a month — the first working Monday is convenient — and one saved report: network segment, 90-day placement report, adjusted metrics. “I’ll look when there’s time” always loses to how fast the network changes.
2. Thresholds that trigger action
Agree in advance what counts as a reason to act. For example: a domain spending more than 1% of the monthly budget with zero conversions across two conversion lags goes to exclusions; a partner spend share growing more than 1.5× month over month triggers an investigation; invalid-activity credits above their usual level trigger a concentration check.
3. One owner for the exclusion list
The account-level exclusion list needs a single owner, with a date and a reason on every row. Otherwise, six months later nobody can explain why a domain is on the list and removing anything feels risky.
4. A quarterly reverse review
Once a quarter, review the list in the other direction: which exclusions can come off? The network’s composition shifts, and a list that has grown to hundreds of unreviewed rows starts costing reach for no benefit.
The same principle applies account-wide — recurring checks against a single list are laid out in the Google Ads account audit checklist.
A 30-minute checklist
- Apply the network segment; record partner share of spend and conversions over 90 days.
- Export the top 30 domains by spend from the placement report.
- Flag domains with spend and zero conversions over at least two conversion lags.
- Check adjusted metrics and any invalid-activity credits.
- Add account-level exclusions and log the date in change history.
- Re-measure after two weeks and decide whether an opt-out test is warranted.
- Put the review in the calendar as a monthly task.
If you run several projects at once and need stable ad accounts with network settings and exclusions configured correctly from day one, see Google Ads agency accounts from PPC Rebels.
FAQ: the Google Search Partner Network
What are search partners in Google Ads?
Third-party sites and services that display Google ads in response to a user query — directories, aggregators, sites with Google-powered internal search, and certain Google surfaces beyond the main results page.
How do I see how much traffic comes from search partners?
Apply the “Network (with search partners)” segment in your reports; spend, clicks and conversions split into Google Search, Search Partners and Display.
Can search partners be turned off?
In Search and Shopping campaigns, yes — via network settings. Performance Max has no generally available toggle; you use account-level placement exclusions, while extended channel controls have been limited to a test group of advertisers during 2026.
Do placement exclusions apply to search partners?
Yes. Account-level exclusions apply across the Search Partner Network as well as YouTube and display placements.
What happened to parked domains?
As of 10 February 2026 they are no longer used as an SPN surface, and the corresponding option was removed from content suitability settings.
What are invalid activity credits?
Refunds for clicks identified as invalid. The report editor provides a campaign-level view of issued credits with adjusted cost and adjusted click metrics — those are the numbers to evaluate performance on.
What does the third-party pre-screen programme give me?
Independent vendors — DoubleVerify, Integral Ad Science and Zefr — receive the network’s domain list weekly, classify sites by risk level and supply curated exclusion lists. You engage the vendor directly.
Is search partner traffic worse than Google Search traffic?
Not necessarily. In some accounts it is cheaper at a comparable conversion rate; in others it is clearly worse. Only a segmented report over a sufficient period answers this for your account.
Why do partners show high CTR but no conversions?
Usually a poorly positioned ad unit on the property, or an irrelevant audience. Those domains are the first candidates for the exclusion list.
Will opting out reset my bid strategy learning?
Changing network settings is a material edit, and campaigns typically re-enter the learning period afterwards. Schedule it as a standalone change rather than bundling it with others.
How often should the exclusion list be reviewed?
Monthly for active accounts. The network’s composition shifts, so a list built six months ago is already stale.
Do I still need manual exclusions if a third-party vendor is connected?
Vendor lists solve brand suitability and content risk, not efficiency. A domain can be perfectly safe and still deliver clicks without conversions — only your own placement report catches that.
How do I tell a network problem from a campaign problem?
Compare the same segment across campaigns. If partner traffic is poor everywhere, the source is the issue. If it is poor only in broadly targeted campaigns while brand campaigns are fine, the question is which queries are taking you there.
Does network share affect automated bidding?
Google states that automated strategies account for network performance and cites uplift when the network reaches around 5% of spend or more. That is a vendor claim — validate it with a test on your own data rather than adopting it as fact.
The right sequence is almost always the same: segment, read the placement report, exclude specific domains — and only if that fails, discuss turning the network off entirely.
Related reading: conversion windows and conversion lag so you never judge partners on raw data, the Smart Bidding learning period so the change is scheduled properly, and impression share and auction insights to tell a source problem apart from rising competition.
