> ## Documentation Index > Fetch the complete documentation index at: https://docs.forepost.ai/llms.txt > Use this file to discover all available pages before exploring further. # Industry benchmarks > Where Forepost's 'above median' / 'bottom quartile' framing comes from. When Forepost says your CSAT sits "above median for SaaS", that comparison is grounded in public industry reports. We never invent benchmark numbers, never claim they're proprietary, and never compare you against other Forepost customers (that comes later, when we have enough data for the percentiles to be meaningful). **Last refreshed:** May 2026, against the 2026 editions of the major reports. ## Sources | Report | What it covers | | -------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------ | | [Intercom 2026 Customer Service Transformation Report](https://www.intercom.com/customer-transformation-report) (n=2,400 support pros) | AI deployment depth, CSAT improvement, response time targets | | [Zendesk CX Trends 2026](https://cxtrends.zendesk.com) | Channel mix, response time targets, "CX trendsetter" CSAT averages | | [Zendesk QA / Klaus Customer Service Quality Benchmark](https://www.zendesk.com/qa/) | CSAT, IQS, agent performance — narrow CX-led cohort | | [HubSpot State of Service](https://blog.hubspot.com/service/state-of-service-report) | Customer expectations, response time targets | | [Lorikeet First Response Time Benchmarks 2026](https://www.lorikeetcx.ai/articles/first-response-time-benchmark-customer-service) | FRT by channel, percentile guidance | | [SurveySparrow CSAT Benchmarks 2026](https://surveysparrow.com/blog/csat-benchmarks/) | CSAT by industry, broad-market median | | [Contentsquare 2026 Benchmark Report](https://contentsquare.com/blog/conversation-resolution-rate/) | Resolution rates across industries | Where reports disagree, we round toward the more demanding figure. "At median" should feel like a real bar, not a generous one. ## Cohort definition (read this if a band feels harsh) The bands describe the **CX-led cohort**: companies that take support seriously enough to track CSAT systematically and run support as a discipline. They're not the broad market. The distinction matters. The broad-market SaaS CSAT median sits around 78% (HubSpot, SurveySparrow 2026). The CX-led cohort — companies running quarterly QA programs, channel-specific SLAs, dedicated CX leadership — sits closer to 85% (Intercom 2026 high-performer subset, Zendesk QA / Klaus narrow-cohort 2023). Forepost users tend to be in the second cohort. We benchmark you against your actual peer group, not against teams that don't measure CSAT at all. If you'd prefer the broad-market frame, the report URLs above are public. ## What we benchmark today Forepost compares your numbers against an industry bucket on five metrics. Volume, open-tickets-now, and agent utilisation aren't benchmarked — those are too dependent on your team size and ARR for a public benchmark to be honest. ### SaaS bucket | Metric | p25 | p50 | p75 | Lower is better | | -------------------- | --- | --- | --- | --------------- | | CSAT | 78% | 85% | 90% | No | | First response | 1h | 4h | 12h | Yes | | AI deflection | 25% | 41% | 58% | No | | One-touch resolution | 55% | 70% | 80% | No | | Backlog age | 1d | 3d | 7d | Yes | ### E-commerce bucket | Metric | p25 | p50 | p75 | Lower is better | | -------------------- | --- | --- | --- | --------------- | | CSAT | 75% | 82% | 88% | No | | First response | 15m | 45m | 3h | Yes | | AI deflection | 30% | 50% | 65% | No | | One-touch resolution | 60% | 72% | 82% | No | | Backlog age | 1d | 2d | 5d | Yes | E-commerce ticket mix skews heavily toward order-status and refunds, so the cadence is faster and deflection is higher (retail and travel see deflection rates above 50%, per multiple 2026 sources). CSAT runs lower than SaaS because shipping issues land in support's queue regardless of fault. ## What changed in the May 2026 refresh Worth knowing — these are the deltas from the v1 (2024) bands: * **SaaS CSAT median: 90 → 85.** v1 anchored to the narrow Klaus B2B cohort. The cross-source 2026 read is harder. * **SaaS AI deflection median: 25 → 41.** The market has materially moved. Median tier-1 deflection is 41.2%, top-quartile 58.7% (Gartner, Notch, Builts.ai 2026). What was "above median" in v1 is now "below median" — without your performance changing. * **SaaS first-response p75: 8h → 12h.** The top end has tightened (40s for chat is the new strong-performance bar) but the long tail is genuinely longer than v1 admitted. * **E-commerce CSAT median: 88 → 82.** Public 2026 data puts e-comm CSAT meaningfully below SaaS. v1 was anchored to a B2B-skewed sample. * **Backlog age:** unchanged in both buckets. Public data on this remains thin and we hold rather than guess. If your benchmark badge changed bands without your numbers moving, this is why. ## Bands For each metric you have data for, the Watchlist card shows a band: * **Top quartile** — green dot, you're in the top 25% * **Above median** — green dot * **At median** — neutral * **Below median** — amber dot * **Bottom quartile** — red dot The Daily Brief receives the same comparison and uses it where it changes the read. We avoid the phrase "industry average" — it's lazy framing. "Bottom quartile" is honest; "industry average" is a hedge. ## Bucket selection Forepost picks the bucket from the **Industry** field on your workspace (Settings → Workspace). The mapping is deliberately conservative — only buckets where we have full public data get their own bands. | Industry | Bucket used | | ----------- | --------------- | | B2B SaaS | SaaS | | B2C SaaS | SaaS | | E-commerce | E-commerce | | Fintech | SaaS (fallback) | | Marketplace | SaaS (fallback) | | Other | SaaS (fallback) | If you're in fintech or a marketplace, take the band as directional. The Watchlist tooltip and the Daily Brief always name the actual bucket used ("SaaS"), never a label that pretends we have data we don't. The **ARR band** field is captured for future bucket refinement (smaller teams have meaningfully slower first-response distributions at the same headcount density). It isn't yet used in selection — public benchmark data isn't sliced finely enough by ARR to justify it. ## Why we don't aggregate Forepost customer data yet Anonymous peer benchmarking is on the roadmap, but it only makes sense when we have enough customers in each segment for the percentiles to be statistically meaningful. With fewer than \~30 SaaS customers in a single ARR / industry bucket, "the median Forepost customer" would just be the loudest few. When we do flip on customer-data benchmarks, it'll be opt-in and clearly labelled. The public-data benchmarks above will stay as a fallback so you always know what the comparison ground is. ## Refresh cadence We re-cut these numbers when major sources publish their next edition (typically Q1 each year). The dataset's `BENCHMARK_LAST_REFRESHED` constant is the canonical date. If the gap between today and that date exceeds 12 months, treat the bands as directional rather than literal.