How AI Search Is Recommending Online Backup: Monthly Trends

Executive Summary

IDrive holds the coverage lead in October 2026 and the gap behind it has widened. IDrive's valid recommendation coverage reached 80.8% (537 of 665 qualified observations), up 5.2 points from 75.6% in July 2026, and it led Backblaze (69.8%) by an 11.0-point leadership gap, widened from a 2.3-point margin in July. This is the widest the top of the category has been across the four-month series.

pCloud was the sharper of two brands moving up outside normal variation, rising 5.8 points to 35.6% coverage (237 of 665 observations) from 29.8% in July, with 8.0 points of that move landing in the September-to-October step alone. IDrive also moved up beyond normal range on a baseline-to-current basis, its third consecutive monthly increase. Carbonite was the sharpest decliner, falling 10.7 points to 17.9% coverage (119 of 665 observations) from 28.6% in July, a third straight monthly decline, with 6.9 points of that drop landing in the October step alone.

The category was not quiet: two brands rose and five fell outside normal month-to-month variation, the widest spread of movement in the series. Acronis (-8.7 points), CrashPlan (-7.4), Livedrive (-2.3), and SpiderOak (-8.3) also declined beyond normal range. Backblaze slipped 3.5 points on a three-month down run that remains inside normal variation. The benchmark shows a top end pulling away and a tail thinning at the same time.

Each monthly run begins with 800 prompt-surface observations (433 unique questions in July 2026 and 513 in October 2026) across the benchmark's defined AI/search surface universe. Of those, 800 mentioned a tracked brand or competitor in each month; 778 were relevant and 22 irrelevant in July, compared with 764 relevant and 36 irrelevant in October. The public metrics use the 671 qualified observations in July and the 665 in October that survive both qualification stages. August 2026 drew 800 prompts (462 unique questions, 781 relevant, 19 irrelevant, 665 qualified) and September 2026 drew 800 prompts (487 unique questions, 781 relevant, 19 irrelevant, 677 qualified).

Key Findings

AI Response Inconsistency Alerts

Questions This Section Answers

• Where did AI platforms give conflicting answers about online backup products and pricing?
• Which backup brands were affected by critical or high-severity factual conflicts?

The benchmark detected 22 critical or high-severity factual inconsistencies across six AI platforms (ChatGPT, Copilot, Gemini, Google AI Mode, Google AI Overviews, and Perplexity) in October 2026. The conflicts cluster on pricing and on product availability, and they affect several brands in this vertical.

SpiderOak

Critical severity, availability. AI platforms provided conflicting information about whether SpiderOak One still exists as a consumer product. When asked "Is SpiderOak One still available?", ChatGPT stated that SpiderOak One (now branded SpiderOakONE Backup) is still available and actively offered as a consumer cloud backup service, citing spideroak.one, the product release notes, and a Tom's Guide best-cloud-backup roundup. Perplexity stated the opposite: that the original SpiderOak One backup product was discontinued or retired for most consumer use cases and is not actively sold or supported as a consumer product. The conflict was flagged on source pages including spideroak.one and SpiderOak's support articles on multi-computer backup and pricing.

IDrive

High severity, pricing (five conflicts). Multiple platforms gave incompatible prices for the same IDrive plans. Copilot stated a 10TB plan costs $4.98/year while Gemini stated IDrive starts around $69 for the first year on a 5TB plan, citing IDrive's press page, PCWorld, and Experte. ChatGPT stated a 5TB plan costs $11.99/month while Gemini stated roughly $70/year for 5TB, the first conflict flagged on IDrive's own pricing page. Google AI Mode stated 10TB costs roughly $105/year while ChatGPT stated a $4.98 first-year introductory offer for 10TB, when asked "Who has the cheapest online storage?". Google AI Overviews stated 5TB typically costs about $99.50/year while Copilot stated about $70/year, a conflict flagged against an IDrive blog page quoting $99.50/year for 5TB. ChatGPT stated a 10TB monthly plan around $17.99 while Google AI Overviews stated a $4.98 first-year offer on 10TB, flagged against IDrive's pricing page showing $17.99/month.

High severity, pricing (continued). Google AI Mode stated a first-year 5TB price of about $79.50 while Google AI Overviews stated about $9.95 for the first year (scaling to $79.95/year). ChatGPT stated about $2/TB/month annually (roughly $10/month for 5TB) while Copilot stated $4/month for 5TB. ChatGPT stated 5TB at $11.99/month while Google AI Overviews stated plans starting around $70 to $100 per year. Google AI Mode stated IDrive starts around $3/year on introductory deals while Copilot stated $79.50/year for 5TB, flagged against two cost-comparison pages quoting the $79.50 figure.

Backblaze

High severity, pricing (five conflicts). Gemini stated a flat rate around $9/month while Google AI Overviews stated $6 per month ($60 per year). Google AI Mode described unlimited flat-rate desktop backup while Gemini described Backblaze B2 pay-per-GB utility pricing at roughly $6/TB/month. Google AI Overviews described true unlimited computer backup at around $70/year while Copilot described B2 object storage at $6.95/TB/month with egress fees, a conflict flagged against two pages quoting both figures. Google AI Overviews stated around $99/year while Copilot stated about $70/year. ChatGPT stated about $99/year for unlimited personal backup while Gemini stated around $9/month, flagged against Backblaze's own pricing page showing $9/month, $7.88/month bi-annually, and $8.25/month annually. Copilot stated about $7/month unlimited while Gemini stated B2 at roughly $6/TB/month.

High severity, eligibility (two conflicts). Gemini stated Backblaze backs up virtually everything including external drives while Copilot stated Backblaze excludes external drives. Copilot stated Backblaze excludes external drives while Google AI Mode stated Backblaze backs up a single computer including attached external hard drives, flagged against guides on both sides.

pCloud

High severity, pricing (three conflicts). Google AI Mode stated annual plans start around $49.99/year while Copilot stated $9.99/month for 2TB, flagged against three comparison pages quoting the $9.99 monthly figure. Google AI Overviews stated a 500GB lifetime plan costs $199 one-time while Perplexity stated about $4.99/month for 500GB, flagged against Gizmodo, a pCloud pricing page, and a budget-comparison page. ChatGPT stated about $100/year for 2TB while Copilot stated $4.99/month (about $60/year), flagged against two pages quoting the annual and monthly figures.

Carbonite

High severity, availability. When asked "What are the disadvantages of Carbonite?", Gemini stated Carbonite has mobile applications that are limited and designed mainly to view or retrieve already-backed-up files, citing TechRadar, Tom's Guide, and PCMag reviews. Copilot stated there are no mobile apps for iOS or Android and that backups cannot be managed on the go, citing Cloudzat, CostBench, and PCMag UK. The conflict was flagged against a PCMag review on one side and a TechRadar review and a third-party review on the other.

Benchmark Context

Questions This Section Answers

• How many prompt observations qualified for the online backup benchmark, and why does that denominator matter?
• Did the qualified benchmark set stay comparable across the four months?

The report separates the raw collection universe from the qualified analysis set. Brand-level recommendation percentages are calculated within the qualified benchmark set.

The qualified denominator held between 665 and 677 across the four months, so brand-level percentages remain directly comparable. August and October both landed at 665, while September peaked at 677.

Benchmark-Level Metrics

The share of recommendation-shaped answers rose 13.1 points and the valid shortlist share rose 6.8 points over the same period, the largest combined shift in the four-month series.

AI Recommendation Trend

Questions This Section Answers

• Which online backup brands gained or lost recommendation coverage between July and October 2026?
• How many brands moved beyond normal month-to-month variation, and in which direction?

The leader pulled further ahead as the middle of the field split

Two brands moved beyond normal month-to-month variation and five moved down beyond it, so the category-level shift came from the combination of these movements rather than a single swing. IDrive and pCloud gained; Carbonite, Acronis, SpiderOak, CrashPlan, and Livedrive lost. Backblaze's three-month slide stayed within its normal range.

What Changed This Month

Questions This Section Answers

• How did IDrive's coverage, presence, and top-three placement shift, and what drove the widened lead?
• What separated pCloud's significant rise from Carbonite's sharpest decline?
• Which brands declined for a third consecutive month or stayed stuck at zero coverage?

IDrive widened its lead on rising coverage and top-three placement

IDrive's valid recommendation coverage rose 5.2 points to 80.8% in October 2026 from 75.6% in July 2026, a significant increase and the third consecutive monthly gain. The rate climbed steadily through the series (75.8% in August, 77.5% in September) before the larger step into October. The brand was recommended in 537 of 665 qualified observations, up from 507 of 671 in July.

Raw mention presence rose 4.2 points to 93.2% from 89.0% in July, and IDrive was present in 620 of 665 observations, up from 597 of 671 in July. Top-three rate climbed 12.6 points to 75.3% from 62.7%, and rank-one rate rose 4.4 points to 43.6% from 39.2%, producing 290 rank-one placements against 263 in July.

IDrive is both mentioned more widely and converted to a high placement more often, so the gain is not breadth alone. Net sentiment held near 0.9 across the period, so the movement is in presence and placement, not in how favorably the brand is described.

Highest-priority diagnostic: which surfaces and prompt types drove the 12.6-point top-three gain, given that rank-one placements rose by 27 observations while presence rose by 23 between July and October.

pCloud was the sharpest riser, with most of the move in October

pCloud's valid recommendation coverage rose 5.8 points to 35.6% in October 2026 from 29.8% in July 2026, a significant increase, with 8.0 points of that move landing in the September-to-October step alone (27.6% to 35.6%). The brand was recommended in 237 of 665 qualified observations, up from 200 of 671 in July.

Top-three rate climbed 13.2 points to 28.1% from 14.9% in July, and rank-one rate rose 3.5 points to 6.2% from 2.7%, producing 41 rank-one placements against 18 in July. Raw mention presence rose 4.5 points to 41.2% from 36.7%, a change inside normal variation, so the coverage gain ran ahead of the presence gain.

pCloud is converting existing visibility into recommendations more effectively rather than appearing in many more answers. Coverage of 237 recommendations sits on only 274 present observations, a tight surface-to-recommendation relationship for a brand at this level.

Highest-priority diagnostic: which prompts moved pCloud from present to recommended, given that rank-one placements more than doubled between July and October while presence moved within normal range.

Carbonite recorded the sharpest decline in the field

Carbonite's valid recommendation coverage fell 10.7 points to 17.9% in October 2026 from 28.6% in July 2026, a significant decline and the largest drop in the category. Most of that fall landed in the most recent month: the rate stood at 24.8% in September before falling to 17.9% in October, a 6.9-point step. The brand was recommended in 119 of 665 qualified observations, down from 192 of 671 in July.

Raw mention presence fell 12.6 points to 26.3% from 38.9% in July, and Carbonite was present in 175 of 665 observations, down from 261 of 671 in July. Top-three rate edged up 2.8 points to 10.1% from 7.3%, and rank-one rate held at 0.1% with a single placement.

Carbonite is being surfaced less often, and the coverage decline is tracking that reduced presence rather than a collapse in its placement when it does appear. Top-three rate actually improved, so the brand is being recommended well when mentioned but mentioned far less often.

Highest-priority diagnostic: which surfaces pulled back or stopped citing Carbonite, given the 12.6-point presence decline between July and October and the single rank-one placement.

Acronis declined for a third consecutive month

Acronis's valid recommendation coverage fell 8.7 points to 39.6% in October 2026 from 48.3% in July 2026, a significant decline and the third straight monthly drop. The rate eased only gradually through August and September (48.1%, then 43.0%) before the larger step into October. The brand was recommended in 263 of 665 qualified observations, down from 324 of 671 in July.

Raw mention presence fell 14.1 points to 46.0% from 60.1% in July, and top-three rate fell 7.7 points to 28.7% from 36.4%. Acronis was present in 306 of 665 observations, down from 403 of 671 in July. Rank-one rate was 3.9%, essentially flat against July's 4.0%.

Acronis is being mentioned and placed in the top three less often, with its rank-one position unchanged. The pattern is a breadth and mid-placement issue rather than a loss of first-place answers.

Highest-priority diagnostic: which surfaces reduced Acronis's higher-ranked placements, given that presence fell 14.1 points between July and October while rank-one held steady.

SpiderOak continued a three-month slide

SpiderOak's valid recommendation coverage fell 8.3 points to 2.7% in October 2026 from 11.0% in July 2026, a significant decline across the full series and a third consecutive monthly fall (7.2% in August, 4.1% in September, before 2.7% in October). The brand was recommended in 18 of 665 qualified observations, down from 74 of 671 in July.

Raw mention presence fell 8.2 points to 3.0% from 11.2% in July, and SpiderOak was present in 20 of 665 observations, down from 75 of 671 in July. Top-three rate fell to 0.8% from 1.6%, and rank-one remained at 0.0% across the period. Net sentiment was 0.9 in October, so the decline reflects reduced presence rather than negative perception.

SpiderOak's coverage now rests on a base of 18 valid recommendations and 20 present observations, so percentage-point moves are driven by very small counts.

Highest-priority diagnostic: whether the availability conflict between AI platforms is contributing to reduced surfacing, given the small base of 20 present observations.

Backblaze declined within normal variation across a three-month run

Backblaze's valid recommendation coverage fell 3.5 points to 69.8% in October 2026 from 73.3% in July 2026, a change inside normal month-to-month variation across a three-month down run (72.0% in August, 70.6% in September, 69.8% in October). The brand was recommended in 464 of 665 qualified observations, down from 492 of 671 in July.

Top-three rate rose 5.5 points to 64.8% from 59.3% in July, and rank-one rate rose 3.5 points to 23.6% from 20.1%. Raw mention presence fell 4.5 points to 83.6% from 88.1%. Backblaze was present in 556 of 665 observations, down from 591 of 671 in July.

Backblaze lost coverage while improving its top-three and rank-one rates, meaning it is recommended slightly less often but more strongly placed when it is. Coverage fell faster than top-three, the reverse of the pattern seen for Carbonite.

Highest-priority diagnostic: which surfaces reduced Backblaze's overall coverage while top-three placement improved, given the three-month down run.

The tail stayed thin or empty

CrashPlan fell 7.4 points to 7.8% coverage (52 of 665 observations) from 15.2% in July, interrupting an otherwise flatter middle of the series; Livedrive fell 2.3 points to 0.4% coverage (3 of 665 observations) from 2.7% in July, including a 1.2-point drop between September and October alone, both beyond normal variation. CrashPlan held a 0.1% rank-one rate with a single placement. Both brands sit on very small counts, so their percentage moves reflect few observations changing. SugarSync remained at 0.0% coverage and 0.0% presence across all four months, and Zoolz remained at 0.0% coverage with a single negatively-tagged mention in October.

Highest-priority diagnostic: whether the tail is a prompt-coverage question or an availability question, given that several low-coverage brands carry open factual conflicts.

Buyer-Intent Interpretation

Questions This Section Answers

• Which buyer-intent clusters did the qualified online backup observations actually capture?
• What can the benchmark not yet answer about pricing, value, or direct comparisons?

In October 2026, all 665 qualified observations fell into the Brand Recommendation cluster, consistent with July, August, and September. No observations qualified for the Pricing & Value or Multi-Brand Comparison clusters in any of the four months. The public benchmark therefore captures who is recommended and in what order, but it cannot yet answer which brand wins on price, value, or direct comparison. Pricing questions did appear in the wider response mix, and the inconsistency data shows these are exactly the questions where AI platforms disagree most, but they sit outside the qualified benchmark denominator rather than inside it.

Brand Opportunity Summary

Questions This Section Answers

• Which online backup brand carries the highest-priority diagnostic in October 2026?
• What signals separate the leaders from the declining and zero-coverage brands?

The benchmark identifies where attention is warranted; a company-level analysis is needed to explain why.

Evidence Behind the Benchmark

The aggregate metrics are built from prompt-level observations (query, surface, recommendation outcome, rank, sentiment, and citations where exposed). Company-level analysis can go deeper into prompt, competitor, surface, and evidence patterns. Source presence is not automatically treated as proof of causation. Where the benchmark records conflicts between platforms, those records show that AI systems gave incompatible answers, not which answer is correct.

About This Benchmark

This report is part of the LLM Authority Index AI Visibility Market Discovery research program.

• AI Visibility Industry Market Methodology
• AI Visibility Industry Market Metrics
• AI Visibility Industry Market Standards

Report-Specific Interpretation Notes

• Small-count movement: SpiderOak moved on 18 valid recommendations, Livedrive on 3, and CrashPlan on 52 in October 2026, so their percentage-point changes rest on small absolute bases.
• Qualified denominator vs raw collection: percentages are calculated within the 665 qualified observations, not the 800 raw prompt surfaces.
• Directional analysis: month-over-month movement indicates where attention is warranted, but it does not by itself establish cause.

Next Step

The Public Benchmark Shows Where a Brand Is Winning or Losing. A Company-Level Audit Shows Why.

Beneath the aggregate percentage sit the questions that matter: which high-intent prompts are won, which competitor takes the recommendation when a brand loses, what attributes AI systems associate with each option, and which external sources shape those answers. The benchmark shows IDrive and pCloud rose while Acronis, Carbonite, CrashPlan, Livedrive, and SpiderOak fell outside normal variation, but it does not reveal the prompt-level reasons.

A company-specific AI visibility audit maps those prompt, surface, competitor, ranking, sentiment, and evidence-source patterns into a prioritized visibility strategy. That is the next step for any brand seeking to understand its position in AI-driven discovery.

Request an AI visibility audit