Customer service statistics a small business can trust, each with who measured it, when and on how many people
Yevhen Kurapov, co-founder · Updated Oct 23, 2026
Fewer numbers than most roundups, each traced to its primary source with year and sample. A one-hour reply delay cut hiring odds by 46% on Fiverr (2026). A quarter to 30% of a US clinic's online bookings came outside staffed hours. A typical small Shopify store converts 1.4% to 1.7% of visits. Several famous figures failed our checks.
Most customer service statistics pages repeat the same numbers, and when we followed them back, several ended at a vendor's blog or at nothing at all. This one is shorter. Every figure below was opened at its source on 23 October 2026, and each comes with who measured it, when, and on how many people or businesses. US figures are labeled US. Where a study was paid for by someone who sells the answer, we say so.
The figures are grouped by the question a small business owner would ask, with a line under each group on what it means for a shop, salon, clinic or trade with one to ten people.
How we picked these
Primary sources only: the study, the dataset or the company's own release, not a roundup quoting a roundup. Fieldwork in 2020 or later, with one exception, the 2011 Harvard Business Review study, labeled old because everyone still quotes it. A stated sample and population; where a vendor published a number with a thin method, we kept it only if nothing better existed and marked it as vendor data. And we read the wording: "46% of frequent online shoppers" is not "46% of customers", and a share of bad experiences is not a share of people.
Reply speed and lead response
| Figure | Who | Year | Sample |
|---|---|---|---|
| A one-hour delay in replying was linked to a 46% drop in the likelihood of being hired; a full day, roughly 90% | Hart, VanEpps, Sezer and Amir, Management Science | 2026 | About 11.6 million buyer and freelancer interactions on Fiverr, plus experiments with more than 3,600 people |
| A 5-minute delay lowered hiring odds by roughly 15% | Same study (Cornell release) | 2026 | As above |
| Replies people believed were automatic or AI-generated lost the speed advantage | Same study (UC San Diego release) | 2026 | As above |
| US companies sent a web lead: 37% replied within an hour, 24% took over a day, 23% never replied; average 42 hours (old) | Oldroyd, McElheran and Elkington, Harvard Business Review | 2011 | 2,241 US companies |
| Firms that tried to reach a lead within an hour were nearly 7 times as likely to have a real conversation with a decision maker as those an hour later, and over 60 times as likely as those that waited a day (old) | Same HBR article, lead data from InsideSales.com | 2011 | 1.25 million leads at 42 US companies |
The Fiverr study is about people hiring freelancers on a marketplace, not customers of a local shop, but it's the best current evidence that minutes matter. The 2011 HBR figures are about B2C and B2B companies following up web leads, where "qualify" meant a phone conversation, not a sale, and they are fifteen years old; we include them because almost every speed-to-lead claim descends from them.
What a small business can take from this: reply within the hour where you can, and make the first reply an answer, since the same study found an instant reply that reads as automatic earned nothing.
After-hours demand
| Figure | Who | Year | Sample |
|---|---|---|---|
| 24.4% of online mammogram self-bookings were made outside Monday to Friday, 7 am to 5 pm | North et al. (Mayo Clinic), JMIR Medical Informatics | 2021 | 13,454 self-scheduled bookings, US |
| 29.5% of well-child visit self-bookings were made outside staffed hours | North et al. (Mayo Clinic), JMIR Medical Informatics | 2021 | 1,201 self-scheduling actions, Minnesota and Wisconsin, US |
| Only 42.5% of website chats arrived between 9 am and 5 pm local time | Remark, chat software vendor | 2025 | More than 25,000 chats in August 2025 at its travel and electronics clients |
| 53% of hair, beauty and wellness appointments booked outside opening hours; 26% during commuting hours | booksalon, salon booking provider | 2023 | More than 460,000 bookings in European markets (the company works with over 5,000 salons) |
The two Mayo Clinic studies are the cleanest measurement we found: real bookings, staffed hours defined. Remark and booksalon are vendor data with thin methods (booksalon doesn't define opening hours), so read them as "more than half" and "about half", not to the decimal. Put together, a small business should expect somewhere between a quarter and half of its enquiries to arrive when nobody is at the desk, and count its own for a month before buying anything.
Messaging and the phone
| Figure | Who | Year | Sample |
|---|---|---|---|
| 73.3% prefer messaging when communicating with a business; 66.8% feel frustrated when it isn't offered; 69% call waiting on hold a waste of time | Kantar, commissioned by Meta | 2025 | 11,056 online adults aged 18 to 64 in 22 markets |
| The phone is the most popular first choice for urgent problems (49%); preference for the phone for complex enquiries rose from 28% in 2018 to 47% in 2024 | ContactBabel | 2024 | Annual surveys of 1,000 US consumers |
| 74% expect 24/7 service because of AI; 86% say responsiveness and accurate resolution strongly influence what they buy | Zendesk CX Trends 2026 | 2025 fieldwork | 6,182 consumers in 22 countries |
Meta sells business messaging and Zendesk sells support software. ContactBabel pulls the other way, and it matters: for an urgent or complicated problem, people still want a voice.
If you can, offer both. Questions, prices and bookings move well in messages; a burst pipe or a cracked tooth gets a phone call.
Online stores: abandonment, unanswered questions and conversion
| Figure | Who | Year | Sample |
|---|---|---|---|
| Average documented cart abandonment rate: 70.22% | Baymard Institute | 2025 | Average of 50 studies, mostly from cart-recovery vendors |
| Reasons, among US shoppers who abandoned (leaving out the 42% just browsing): extra costs 40%, slow delivery 20%, didn't trust the site with card details 19%, forced account 18%, long checkout 17%, returns policy 13%, total cost not shown up front 12% | Baymard Institute | 2025 | US online shoppers; sample not given on the free page |
| 46% left a site ready to buy because a question went unanswered; 53% decided not to buy while waiting for a response | Gladly (sells AI customer service), with Propeller Insights | 2026 | 1,022 US adults who shop online at least a few times a month |
| 50% abandoned a potential purchase in the last six months because they couldn't find enough product information | YouGov for Syndigo (sells product-data software) | 2024 | 6,480 adults in the US, France, Germany and the UK |
| Reasons for leaving an online cart: shipping costs 19.6%, a better deal elsewhere 19.5%, waiting for a sale 14.9% | YouGov for commercetools | 2023 | 4,581 US and UK adults |
| Median Shopify store conversion rate 1.4%; top 20% above 2.7%; median cart abandonment 78% | Littledata | 2026 | 433 Shopify stores, 90 days to 17 October 2026, each with at least 1,000 sessions and 90 purchases |
| Median Shopify store conversion rate 1.74%; middle half 0.83% to 3.32% | Shogun | 2026 | 803 Shopify stores, January to September 2026 |
Baymard's list and commercetools' list asked different questions, so don't set their percentages side by side. The same goes for the two abandonment rates: Baymard's 70.22% averages 50 studies, while Littledata's 78% is a median across 433 Shopify stores. Read together, the store figures say the reasons named most often are cost, trust and checkout friction, which no chat fixes, and a smaller but real share is about questions nobody answered.
A typical small Shopify store turns 1.4% to 1.7% of visits into an order. If yours is below that, fix the shipping surprise and the forced account first, then make sure "will it fit?" and "when will it arrive?" get an answer on the page or in the chat.
No-shows and reminders
| Figure | Who | Year | Sample |
|---|---|---|---|
| Primary-care no-show rates ranged from 5.2% to 38.0%, median 14.2%; about 7% to 15% in large Western datasets | Jad et al., Healthcare (systematic review) | 2026 | 27 studies, 8 countries, more than 13 million appointments |
| Patients who got any reminder were about 11% more likely to attend (RR 1.11); text reminders alone, RR 1.14, not statistically significant | Al-Turbag, Mooney and Corry, Trinity College Dublin (meta-analysis) | 2026 | 10 studies, 8,236 hospital outpatients |
| A second text reminder cut the chance of a no-show by 7% for primary care and 11% for mental-health visits | Ulloa-Pérez et al. (Kaiser Permanente), The Permanente Journal | 2022 | 125,076 primary-care and 33,593 mental-health visits flagged high risk, US |
All three are clinics and hospitals. We looked for salon no-show figures and none passed: salon platforms publish them without a method. So reminders help a little, not dramatically. A deposit or a cancellation rule may do more than a second text; we have no study to prove either.
What customers do after a bad experience
| Figure | Who | Year | Sample |
|---|---|---|---|
| 11% of experiences are bad; 47% of those lead customers to cut spending, 58% after a bad online retail experience | Qualtrics XM Institute | 2025 | Over 20,000 consumers, 14 countries, 18 industries |
| Only 3 in 10 explain why they leave; 30% switch brands without telling anyone | Qualtrics XM Institute | 2025 | 20,000 consumers, 14 countries |
| 63% say they're willing to switch to a competitor after one bad experience | Zendesk CX Trends 2025 | 2024 fieldwork | About 5,100 consumers, 22 countries |
The Qualtrics figures are shares of bad experiences, not of customers. Zendesk's is what people say they would do, not what they did. Still, the direction is clear enough: most unhappy customers won't tell you, and the questions people ask before buying, including the ones that go unanswered, are often the only warning a small business gets.
Price and reviews
| Figure | Who | Year | Sample |
|---|---|---|---|
| 46% compare prices across several retailers; only 9% buy without comparing; 79% have delayed a purchase expecting a lower price; 32% trust retailers to offer a fair price | Dynata for Akeneo (sells product-information software) | 2026 | 1,000 US consumers |
| 97% read online reviews for local businesses; 93% have bought after reading reviews; 27% have spent more than $1,000 after reading them; positive reviews make 85% more likely to use a business, negative ones deter 77% | BrightLocal (sells local marketing software) | 2026 | 1,002 US adults |
Both are US surveys by companies that sell to retailers and local businesses. Read together, they suggest the person asking "how much?" often has two other tabs open, and that reviews and a clear list of what the price includes are worth having ready before anyone asks for a discount.
Instagram and selling in DMs
| Figure | Who | Year | Sample |
|---|---|---|---|
| More than one billion active threads with businesses on WhatsApp, Messenger and Instagram every day | Meta | 2026 | Meta's own count |
| Since September 2025, shops on Facebook and Instagram send buyers to the seller's website to pay | Meta Business Help Center | 2025 | Policy |
| Full shops and product tagging in the US; open beta in 20 countries including Canada, the UK and Spain | Instagram Help Center | 2026 | Policy, as of October 2026 |
| Instagram can reach about 182 million people in the US, 52.3% of the population | DataReportal Digital 2026, from Meta's ad tools | 2025 | Ad reach, not unique users, US |
| Half of US adults use Instagram, up from 40% in 2021 | Pew Research Center | 2025 | 5,022 US adults |
The checkout and tagging rows are Meta's platform rules, not survey figures; they're here because many guides still describe Instagram's old in-app checkout. The DataReportal figure is what Meta's ad system says it can reach, and Pew's is a survey of people, so the two measure different things and shouldn't be compared directly.
For a small brand selling on Instagram, that means the sale starts in a post or a DM and is paid for on its own website.
What we left out and why
"78% of customers buy from the first business that responds." Every trail we followed ended at a vendor's marketing page. No study.
"Respond within 5 minutes and you're 100 times (or 21 times) more likely to reach the lead." It comes from a 2007 phone study of web leads at six companies, presented at a conference; the primary page is gone. That's where the "five minutes" folklore started. The multipliers can't be checked, so we don't repeat them.
"1 in 26 unhappy customers complain." An old TARP figure with no current primary source. The Qualtrics finding above (only 3 in 10 explain why they leave) is the measured replacement.
"55% abandon a purchase if they can't find a quick answer." Attributed to Forrester around 2010. The source no longer opens.
Salon "x% of bookings happen after hours" figures from Treatwell, Fresha, Booksy and Phorest. No method published, numbers that disagree with each other, or older than 2020. We kept booksalon's, labeled as vendor data, because it at least gives its sample.
A 2018 study putting no-shows at 23%, and the Cochrane review of SMS reminders. Both older than our 2020 cutoff, and newer reviews cover the same ground.
Where ChatPal fits
ChatPal is an AI assistant that answers typed messages in a website chat, WhatsApp, Instagram, Messenger and Telegram at any hour, from the owner's own price list, policies and catalog, and books into Google Calendar, Calendly or Altegio. It doesn't answer the phone, which the ContactBabel figures say still matters for urgent problems. It's free for 100 replies a month, with paid plans from $20.