DuckDuckGo Installs +30% After Google I/O 2026: What the AI Search Backlash Actually Means for SEO

On May 26, 2026, TechCrunch's Rebecca Bellan reported that DuckDuckGo's US app installs averaged +18.1% week-over-week over May 20–25 versus the May 13–18 baseline — and peaked at +30.5% on May 25. On iOS specifically, the average was +33% with a single-day peak of +69.9%. Visits to DuckDuckGo's AI opt-out page, noai.duckduckgo.com, averaged +22.7% week-over-week and peaked at +27.7% on May 24.

DDG CEO Gabriel Weinberg told TechCrunch: "Google is force-feeding AI with no way to opt out. As a result, their results are getting worse, not better." Bellan's reporting connects the surge to Google I/O 2026 (May 19–20), which expanded AI Mode and Generative UI across Google Search — the biggest change to the Google SERP since AI Overviews launched two years ago.

The story has been picked up by every major SEO outlet. The most-shared reaction inside the SEO community came from Lily Ray — Senior Director of SEO & Organic Research at Amsive, and one of the most-followed practitioners in the space — who quoted the TechCrunch piece on X with one short line:

Lily Ray 😏

@lilyraynyc

·Follow

Will be interesting to see where this goes (also kinda curious why this didn’t happen earlier with AI Overviews)

TechCrunch

@TechCrunch

DuckDuckGo installs are up 30% as users reject being ‘force-fed’ Google’s AI Search techcrunch.com/2026/05/26/duc…

7:30 AM · May 27, 2026

21

Reply

Copy link

Read 4 replies

That parenthetical is the better story. AI Overviews launched in the US on May 14, 2024 — exactly two years before this TechCrunch piece. If "Google is force-feeding AI" is the trigger for a 30% install week in 2026, the obvious question is: why didn't the original AI Overviews launch trigger the same reaction in 2024? Either it did and nobody noticed at the time, or something is materially different about this cycle.

That's what this post is actually about. We pulled the Google Trends curve for "DuckDuckGo," cross-referenced it with StatCounter market-share data and DDG's own daily-search figures, and worked through what changed between 2024 and 2026 — and what it means for SEO teams who almost certainly haven't been thinking about DuckDuckGo as a serious channel.

Three findings, each unpacked below:

1. Lily's instinct was correct — and the data is sharper than it looks. Search interest in "DuckDuckGo" on Google Trends did spike during the 2023–2024 AI-search wave. But DDG's actual usage — market share, daily searches — peaked back in 2021, before any AI-search launch, and never recovered. So the 2024 reaction happened in curiosity, possibly in app downloads, but not in durable switching.
2. The May 2026 wave is the first time install data is being publicly attributed to a specific Google AI event. Whether it converts to durable use is the open question. The May numbers won't answer it; the June and July ones will.
3. For SEOs, the practical surface is Bing — not DuckDuckGo. Traditional DDG web results come primarily from Bing's index, confirmed by DDG's own help docs. A 30% week on DDG is a Bing-visibility story.

What TechCrunch actually reported (the verifiable numbers)

The headline "+30%" comes from a single-day peak on May 25, not a sustained run-rate. The averages over the six-day window are more modest — but iOS-specific growth is the real outlier.

All three metrics in Bellan's reporting are DDG-reported figures shared with TechCrunch — neither App Store rankings nor third-party install-estimation services have independently confirmed them at this writing. The most reliable comparable is DDG's ~2% US search market share per StatCounter, which puts a +30% install week into context: DDG is gaining share on a small base, not displacing Google.

DDG attributed the surge to Google I/O 2026, where the company expanded AI Mode (Gemini 3.5 Flash as the default model) and rolled out Generative UI — search-results pages built on the fly by the model rather than served from the standard SERP template. Our I/O 2026 walkthrough covers the mechanics.

The Google Trends chart for "DuckDuckGo" (2004 → today)

Pull up Google Trends for "DuckDuckGo," set the range to 2004 – present, leave the geography at worldwide and the search type at web search, and you get this curve:

Source: Google Trends — "DuckDuckGo", Worldwide, Web Search, 2004 → present. Screenshot taken May 27, 2026.

Same curve, with two annotations added so we can talk about specific inflection points:

From near-zero in 2004 to a slow grind through the 2010s, DDG's search interest took off in 2020–2021 (privacy tailwinds, Big Tech antitrust attention), then spiked into a 100-reading peak in late 2023 / early 2024. Then the curve fell. From 100 at the peak to roughly 40 by mid-2025, where it's sat ever since.

The obvious reading is: "there was a 2024 reaction to AI Overviews that didn't convert into installs." That's where I landed first, and it's the wrong reading. Google Trends measures interest in the search term, which is mostly news-cycle driven. To see whether anything actually happened on DDG, you have to overlay the usage data — and the usage data is a different shape entirely.

Interest peaked in 2024. Usage peaked in 2021. The leakage is in between.

Here's the disconnect that makes Lily Ray's question more interesting, not less:

Per StatCounter, DuckDuckGo's US search market share peaked at 2.42% in 2021 and was down to 1.96% in 2024. Daily searches peaked at 111.7 million on January 17, 2022 per DDG's own publicly tracked stats — and then plateaued around 100M/day through 2024. By the time AI Overviews launched in May 2024, DDG had already been flat-to-declining for two years.

Two events worth keeping in mind when looking at this:

1. The May 2022 Microsoft tracker scandal. Security researcher Zach Edwards discovered that DDG's browser allowed Microsoft trackers on third-party sites due to a syndication agreement with Bing. The public backlash was sharp; DDG amended the carve-out in August 2022. The damage to the privacy narrative was a durable headwind on growth.
2. The 2023–2024 AI-search wave. ChatGPT (Nov 2022), DuckAssist (Mar 2023), Google's SGE (May 2023), Bing Chat / Copilot, and finally AI Overviews (May 14, 2024). The Google Trends "DuckDuckGo" line tracks that wave. The DDG market-share line does not.

Did downloads actually happen in 2024?

The honest answer: almost certainly some did — we just can't see exactly how big the spike was. If you Google-search for "DuckDuckGo" in the middle of an AI-search news cycle, the most common next click is the App Store. Public download numbers are too coarse to detect a 2024 bump cleanly:

• April 2024 Android installs are cited at ~1,000,000 in widely-syndicated stats. iOS the same month at ~500K.
• Current 2026 monthly estimates per Sensor Tower are roughly 1M Android + 600K iOS in the US. Roughly steady.
• Month-by-month time-series download data isn't public — Sensor Tower has it behind a paywall.

So the data we can see is consistent with a 2024 install spike that didn't break out of the monthly noise — or with one that happened but churned away. We can't cleanly distinguish those two without Sensor Tower's time series. What we can definitively say is what came next.

The downstream metrics — market share and daily searches — flatlined

Wherever 2024's installs sat in volume, they did not convert to durable usage. Market share went from 1.93% to 1.96%. Daily searches stayed at ~100M against a 2022 peak of 111.7M. The 2024 AI-search anxiety produced enough behavior to keep DDG on the map and not enough to grow it.

That's the actual finding: the leakage point isn't curiosity → install. It's install → durable use. People tried DDG and went back to Google.

Back to Lily's question: why might May 2026 be different?

Reframing her question with the data above: it probably did happen earlier — at the install layer — but it didn't stick. The leakage from install to durable switching has been DDG's structural problem since 2021. The puzzle in 2024 wasn't that people didn't download it; it was that the downloads didn't change which default tab people used.

So what could be different about May 2026? Three hypotheses — none confirmed, all consistent with the data:

1. The product change is sharper. AI Overviews in 2024 sat above the 10 blue links. AI Mode + Generative UI in May 2026 increasingly replace the SERP template entirely on AI-eligible queries — per Google's own I/O announcements. A power user could ignore the AI panel in 2024 and stay on Google; that's harder when the whole page is generated.
2. Weinberg's "no way to opt out" framing has a specific target. In 2024 Google still served standard results below AI Overviews and publishers could think Google-Extended was the opt-out lever. Neither is true for AI Mode in 2026, and the framing lands harder because of it.
3. noai.duckduckgo.com is a frictionless landing page that didn't exist in 2024. The +22.7% week on that subdomain is the most directly behavioral number in the dataset — it's not an install, which a user can ignore. It's a repeat visit to a dedicated AI-free entry point, which means the user is actually searching there.

And one structural reason it might not be different: DDG's install-to-durable-use conversion has been broken since 2021. A six-day measurement window inside a news-cycle peak is the worst possible time to declare that conversion is fixed. The June and July numbers — not the May ones — are where the answer lives. If StatCounter market share moves up off its 1.84% June 2025 reading, the 2026 wave is different. If it doesn't, this is another curiosity-and-churn cycle.

Why this is a Bing story, not a DuckDuckGo story

Here's the part missing from every general-press write-up: DuckDuckGo doesn't have a competitive web index of its own. Per DDG's own help docs, traditional "10 blue links" results come primarily from Bing's index, supplemented by DDG's own crawler (DuckDuckBot), Wikipedia-powered Instant Answers, and partner APIs like Wolfram Alpha and Apple Maps.

This is not theoretical. During the August 2024 Bing API outage, DDG stopped returning results altogether — a public confirmation of how much of the DDG result page is sourced from Bing's index. Per a July 2025 TechCrunch follow-up cited in industry coverage, Microsoft signaled it may rework its Bing search APIs to push more AI-oriented endpoints, though DDG has stated its long-term agreement with Microsoft is not affected.

For SEO teams, that changes the action items dramatically. The question isn't "how do I optimize for DuckDuckGo." The question is: if a measurable slice of US search traffic just shifted onto Bing's index, is your site visible there?

Six checks every SEO team should run this week

Six things to verify if a meaningful slice of US search traffic is about to start flowing through Bing's index. The first four are 10-minute jobs; the last two are ongoing. We've linked our own free tools where they cover the check directly — no signup needed for any of them.

1. Bingbot is allowed in robots.txt

Open yourdomain.com/robots.txt and search for Bingbot and msnbot (the legacy MSN crawler — still in some boilerplate files). A blanket Disallow: / under User-agent: Bingbot is the obvious blocker, but the subtler problem is when a generic User-agent: * block has a path that shouldn't apply to Bingbot. If you have a section in robots.txt that specifically lists Googlebot with allow rules, Bingbot likely doesn't inherit those — it falls back to User-agent: *, which might be more restrictive.

The fastest way to verify: run your URL through our free robots.txt tester. It evaluates the file against multiple user-agent strings including Bingbot, so you can see which paths are crawlable from Bing's perspective without manually parsing the file.

Why it matters: If Bingbot can't crawl, Bing can't index — and DuckDuckGo's traditional results won't surface you. This is the single most common Bing-visibility issue we see on sites that were set up Google-first.

2. Site is verified in Bing Webmaster Tools

Go to bing.com/webmasters and verify the property. Bing accepts the same DNS TXT record method Google uses; you can also import your GSC verification automatically (Bing has a one-click GSC-import flow that saves a few minutes). Submit your sitemap, enable IndexNow at the same time (one toggle), and check the Indexed Pages report.

The Indexed Pages comparison is the diagnostic worth running: pull your indexed-page count from GSC, pull the equivalent number from Bing Webmaster Tools, and look at the delta. If Bing has 60% of what Google has, you have a Bing crawl-coverage problem worth investigating — likely robots.txt, redirect-chain, or canonical issues that Bing parses more strictly.

Why it matters: Without verification you have zero visibility into Bing's view of your site — and that view is exactly the view DDG uses for its "ten blue links."

3. IndexNow is wired up

IndexNow is a single key-protected endpoint that pushes URL changes to participating search engines. As of mid-2026 that list is Bing, Yandex, Naver, Seznam, and Yep — but practically it's the Bing one that matters for DDG visibility. DuckDuckGo hasn't formally adopted the protocol, but because DDG's results come from Bing's index, IndexNow-pushed URLs reach DDG at Bing's indexing speed instead of waiting for the next Bingbot crawl.

Implementation paths, ranked by speed:

• WordPress with Yoast or Rank Math: single toggle. Yoast generates the key file automatically; Rank Math has had IndexNow in the free plan since 2022.
• Microsoft's IndexNow plugin: official plugin if you don't want Yoast's footprint.
• Custom stack: generate a UUID, write it to yourdomain.com/<key>.txt, then POST a JSON body with your URL list to https://api.indexnow.org/indexnow. The spec is one page.

Why it matters: Faster indexation on Bing = faster freshness on DDG. Google does not support IndexNow — it tested the protocol in 2021 and has not adopted it since — so this is one of the few protocols where Bing's ecosystem materially outperforms Google's.

4. Canonical signals are unambiguous

One <link rel="canonical"> per page, self-referential where appropriate, matching the URL in your sitemap. Bing weights canonical signals more literally than Google — Google will often ignore a canonical it disagrees with and pick a different URL based on its own signals; Bing tends to honor what you declare. That sounds good until you have an unintentional canonical to a 404 or to a redirect chain.

Run your top templates through our free canonical checker — it pulls the canonical tag, follows it, and flags conflicts between the canonical, the sitemap entry, the HTTP header canonical (if any), and the meta robots directives. The common issue: a template-level canonical pointing to a parameter-stripped URL while the sitemap entry has parameters intact, or vice versa.

Why it matters: Bing reads canonicals more literally; conflicting or missing canonicals show up as duplicate-content losses in Bing/DDG before they show up in Google.

5. Schema validates without Google-specific dependencies

Google's Rich Results Test is forgiving — it only checks the schema types Google currently uses to render rich results. That can hide structural problems with the markup itself. Bing's parser is stricter on JSON-LD type fields, required properties, and date formats; a schema block that passes Google's test can still fail to enrich Bing/DDG SERPs.

Run your active schema templates through our free schema validator in addition to Google's tool. We test against the schema.org spec (not Google's subset), check JSON syntax, required properties, date formats, and nested entities for Article, Product, Organization, BreadcrumbList, FAQPage, HowTo, and a dozen other types. The schema types worth checking first: Organization (for entity recognition on Bing), Article (for AI-Mode and Bing News surfacing), and Product (for Bing Shopping).

Why it matters: If your structured data only validates against Google's tooling, you're shipping a Google-specific contract — and the rest of the search ecosystem ignores it.

6. Brand and entity signals are reachable

Organization schema with sameAs links to your Wikipedia, Wikidata, Crunchbase, and verified social profiles. A Bing Places listing if you have a physical presence. Consistent NAP (name, address, phone) data across major directories. None of this is new advice — it's standard local/brand SEO — but the surface that consumes it is different in the Bing/DDG ecosystem.

DDG's Instant Answers and Bing's entity panels draw on Wikipedia, Wikidata, and an entity graph that's separate from Google's Knowledge Graph. If your brand has solid Knowledge Graph coverage but no Wikidata entry, you're invisible in the DDG entity surface. Two practical actions: (1) make sure your Organization schema has accurate sameAs URLs (validate it with our schema validator), and (2) if you don't have a Wikidata entry, draft one — it's the lowest-friction entry into the entity graph DDG actually queries.

Why it matters: Brand SEO that only targets Google leaves entity coverage on the table. The DDG/Bing entity graph is a meaningful surface for navigational and brand queries, especially on AI-free search experiences where there's no LLM filling in the blanks.

Prioritization, if you have an hour this week: do checks 1 and 2 first (robots.txt + Bing Webmaster Tools verification — that's 20 minutes and unlocks the Bing data layer). If you publish weekly or more often, do check 3 next (IndexNow). Checks 4–6 are ongoing health items — work them into your normal audit cadence rather than treating them as emergency items.

IndexNow context: who supports it, who doesn't

IndexNow is a 2021 protocol from Microsoft and Yandex that lets you push a URL change to participating search engines instead of waiting for them to recrawl. The current support matrix, as of mid-2026:

• Supports IndexNow: Bing, Yandex, Naver, Seznam, Yep.
• Doesn't formally adopt, but benefits indirectly: DuckDuckGo (via Bing's index).
• Doesn't support: Google. Google publicly tested IndexNow in 2021 and has not committed to adoption since.

For most SEO stacks the integration is one toggle. Yoast, Rank Math, and Microsoft's own IndexNow plugin all ship the protocol; per IndexNow.org's own reporting, the WordPress IndexNow ecosystem crossed 10 million active installs in 2025. If you're on a custom stack, the spec is a single key-protected endpoint — implementation is typically under an hour.

What this story is not

Three things worth saying out loud, because most write-ups of stories like this slide past them:

1. This is not Google losing market share. StatCounter shows DDG at ~2% US share. Even a sustained +30% gain is a small absolute move. The interesting signal is the direction, not the magnitude.
2. The install numbers are DDG-reported. Bellan's TechCrunch piece is sourced from DDG's own measurements. They're plausible (Weinberg has no reason to inflate by a factor that would be obvious in App Store rankings) but not third-party-verified. Treat as "directionally accurate" until comparable App Annie / data.ai numbers land.
3. Google Trends measures search interest in the term, not installs. The 2024 peak is people searching for DuckDuckGo on Google. That's an upper-funnel signal, not a behavior change. The 2026 install surge is a lower-funnel signal — and it can absolutely happen without a corresponding Trends spike, because people downloading the app aren't Googling about it.

What to do this week if you ship SEO for a living

1. Open Bing Webmaster Tools. Verify the site, submit a sitemap, look at the indexed pages report. Compare against your Google Search Console indexed-page count — the delta tells you how much of your content Bing/DDG even knows about.
2. Pull a robots.txt diff. Make sure no inherited template is blocking Bingbot or msnbot. This single check resolves the bulk of Bing-visibility issues we see.
3. Wire up IndexNow if you publish weekly or more. Yoast or Rank Math handle it on WordPress; on custom stacks, write the endpoint. Cost-to-implement is low; the upside on Bing/DDG indexation is real.
4. Add a DDG-specific row to your traffic dashboard. Referrer logs from duckduckgo.com are the only first-party way to track this audience. If you're running plausible/PostHog/server logs, that's a 5-minute filter.
5. Watch the June numbers, not the May ones. A six-day surge after a major news cycle is normal. A sustained June run-rate is the real signal that this isn't a curiosity spike.

Primary sources used in this post

• DuckDuckGo installs are up 30% as users reject being ‘force-fed’ Google's AI Search — Rebecca Bellan, TechCrunch, May 26, 2026 (the primary install-figures source, including the Weinberg quote)
• ‘Google is force-feeding AI with no way to opt’: DuckDuckGo CEO says installs are surging after Google I/O — Tom's Guide secondary coverage
• Where do DuckDuckGo search results come from? — DDG's own help page on result sourcing (the Bing-index confirmation)
• About noai.duckduckgo.com — DDG help page on the AI-free entry point
• AI Overviews expanding to more than 100 countries — Google's October 2024 announcement, with the May 14, 2024 US launch date
• StatCounter Global Stats — DDG market share figures (~2% US)
• IndexNow.org — the protocol's own site, with the participating-engines list
• Google Trends — "DuckDuckGo", Worldwide, 2004 – present, Web Search (chart accessed May 27, 2026)
• Lily Ray (@lilyraynyc) on X, May 27, 2026, quoting TechCrunch

Related reading on CrawlRaven

• Google I/O 2026: Every Major Search Update
• Google May 2026 Core Update: Live Tracking & Recovery
• How to Rank in ChatGPT, Claude, Google AI Overviews
• How to Perform a Technical SEO Audit in 2026