References & Sources.
Every statistic on this site links here. Real published research, real data — and where we cite our own client averages, we say so explicitly. If you spot a number we didn’t source, that’s a bug. Let us know.
5-minute lead-response window
Cited on the Problem section of the home page — “the window to respond before a lead goes cold”. The original study tracked response times across 2,241 US companies and showed lead qualification odds drop ~80% if the first contact happens after five minutes.
Oldroyd, J. & Elkington, D. (2011). “The Short Life of Online Sales Leads.” Harvard Business Review.
hbr.org/2011/03/the-short-life-of-online-sales-leads ↗78% of buyers go with the first responder
Cited on the Problem section — “78% of buyers go with the first responder”. The general finding across multiple vendor-funded and academic studies: the first vendor to engage meaningfully wins the deal a clear majority of the time.
InsideSales.com / Velocify, “Lead Response Management Study”; corroborated by HBR Lead-Response research and Gartner B2B buying behaviour reports.
HBR (2011) ↗ · Gartner B2B Buying ↗67% of inbound leads go unanswered
Cited on the Lead Counter section — “estimated leads missed globally this hour”. Drift’s methodology: secret-shopper tests across thousands of B2B forms, measuring whether anyone replied within five minutes, an hour, or at all.
Drift, State of Conversational Marketing report. Aggregated across thousands of B2B websites tested for response rate.
drift.com ↗60% recovery rate on lost leads
Cited on the ROI Calculator. Individual results vary widely based on industry, offer fit, the speed of the existing process, and the quality of the lead source. The 60% figure is a conservative operating average — clients with severely manual processes typically see higher recovery; clients already running fast follow-up see lower.
HiveCliq internal data. Average recovery rate across deployed AI follow-up systems on the HiveCliq platform. This is our own number, not a third-party citation.
Average missed-lead value
Cited on the ROI Calculator. The default value is a starting estimate — set the slider to match your own typical deal size for a result that reflects your business.
HiveCliq internal data. Estimate derived from our South African and international client base across professional services, healthcare, retail, automotive and SaaS verticals.
Dealership chatbot agreeing to sell a Tahoe for $1
Cited in The one dollar Tahoe. In December 2023 a software engineer, Chris Bakke, prompted the AI assistant on the website of Chevrolet of Watsonville, California, into agreeing to sell a new Chevrolet Tahoe for one dollar, and to close with “that’s a legally binding offer, no takesies backsies”. The dealership did not honour it, and took the assistant offline. The bot was never authorised to set prices, so the “binding” line carried no legal weight.
Reported December 2023. Summary and screenshots via GM Authority.
gmauthority.com/blog/2023/12/gm-dealer-chat-bot-agrees-to-sell-2024-chevy-tahoe-for-1 ↗Air Canada held liable for its chatbot’s advice
Cited in The one dollar Tahoe. Air Canada’s website chatbot described a bereavement fare refund process that did not match its actual policy. The British Columbia Civil Resolution Tribunal found the airline liable for negligent misrepresentation and awarded the passenger CAD 812.02. The principle that matters for any business: a company remains responsible for what its automated assistant tells a customer.
Moffatt v. Air Canada, 2024 BCCRT 149, decided 14 February 2024. Analysis via the American Bar Association.
americanbar.org, BC Tribunal confirms companies remain liable ↗DPD support chatbot swearing and writing a critical poem
Cited in The one dollar Tahoe. A customer, Ashley Beauchamp, posted an exchange in which the delivery firm’s AI chat agent agreed to swear and composed a poem describing DPD as useless. DPD disabled the AI component the same day. The underlying issue was scope: the agent had no constraint on what it would discuss.
Posted 18 January 2024, reported widely. Coverage via TIME.
time.com/6564726/ai-chatbot-dpd-curses-criticizes-company ↗McDonald’s ending its AI drive-thru trial
Cited in The one dollar Tahoe and The boring AI wins nobody posts about. McDonald’s ran an IBM-built voice ordering trial from 2021 across roughly 100 US restaurants and ended it in June 2024, with the system switched off by 26 July 2024. The decision followed a run of viral videos of ordering errors, the best known showing the system repeatedly adding chicken nuggets to an order while the customers asked it to stop.
Announced 17 June 2024. Coverage via Global News.
globalnews.ca/news/10573231/mcdonalds-ai-drive-thru-end ↗Zillow Offers shut down after algorithmic overpaying
Cited in The one dollar Tahoe. Zillow closed its home-buying business in November 2021 after its pricing model bought homes above what they later sold for. The company recorded an inventory write-down of USD 407.9 million for the year ended 31 December 2021 and cut roughly 25% of its workforce. The failure was one of brief rather than model quality: the algorithm was making the purchase decisions directly, at scale, with no human checkpoint.
Zillow Group, Inc. Form 10-K for FY2021, filed with the SEC. Write-down figure taken from the filing itself.
sec.gov, Zillow Group FY2021 Form 10-K ↗
Talking numbers? Lara’s your first call.
She qualifies your business in 60 seconds, separates service fees from ad spend, and shows you the exact numbers your situation produces — no generic estimates, no fluff.