More people are building automations. Is anyone saving time?
IFTTT says more people are building small automations since its October launch. Before a business pays for one, try a reversible job and count errors, corrections and credits as well as minutes saved. The platform’s growth figures do not tell you whether a client will pay.
Explore this signal
Emerging data
More people made rules. Did the rules do useful work?
A small shop can have a new enquiry copied to an internal practice sheet. Building that rule is one event. Checking that the right row appeared once, on time, is another. IFTTT calls these little rules Applets. On 17 November it reported that weekly builders were more than 50% higher than in early October. The number building Applets that use AI was roughly three times the prelaunch level. Those are counts of builders, not paid subscriptions.
Three numbers with different meanings
IFTTT.ai launched on 28 October. The fresh event was the company’s first-weeks usage report, published on 17 November at 08:23 US Eastern, or 21:23 Singapore time. It says about one in four new Applets was built in chat. Chat is how a person creates the rule; using AI inside an Applet describes what the rule does later. Neither measure establishes hours saved, reliability or revenue.
- More than 50% growthPeople building Applets each week, relative to the earlier baseline.
- Nearly tripledPeople building AI-enabled Applets, not all users.
- About one in fourShare of newly built Applets made through chat, not a share of paying customers.
Start with one reversible job
With the business owner’s permission, try twenty dummy enquiries in a practice sheet, never a customer database. For each one check: did the right row appear, did it appear twice, and how much time did checking or correcting it take? Compare that with copying the row by hand. Record actual credits used, too: IFTTT’s October billing system meters workflow steps. Check current plan limits rather than assuming a free trial represents long-term costs.
Decide who can stop it
Agree who can disable the test and remove access if something goes wrong. Google documents how its account holders can remove third-party connections. NIST recommends multifactor authentication as a protection for business accounts. Zapier’s own workflow history illustrates a useful comparison: it records the outcome of individual runs. These sources explain controls and alternatives; none audits IFTTT’s usage numbers. Do not use this low-risk trial for payroll, payments, medicine or personal customer data.
View evidence links
| Record | What it proves | What it does not prove |
|---|---|---|
| Successful test run | An expected action occurred once. | It works reliably on every input. |
| Matched trial log | Manual work avoided, errors, fixes and credits can be compared. | A paying customer or a profitable service. |
| Signed scope and cleared invoice | A customer actually paid for the delivered work. | Customer retention or earnings after all costs. |
When an experiment becomes paid work
A client might commission a simple reminder or ask someone to monitor errors and credit use each month. A successful demonstration is not an invoice. Compare measured time saved with support time, failed runs and software costs. If there is no agreed price, reliable output or way to stop the workflow, keep the existing process. More builders could simply mean more people trying a free service.
Original forecast • 18 November 2026, 11:03 SGT
Within 30 days, a safe workflow with recorded cost and errors is more likely to earn a modest service invoice than a large undocumented automation package. Look for a customer who has accepted the job and actually paid; the company’s adoption figures do not establish that comparison. If the manual process is faster after corrections, or no customer pays, record that rather than claiming savings.
- 28 Oct 2026IFTTT.ai launch: background, not the new event.
- 17 Nov 2026 · 21:23 SGTCompany published its first-weeks builder metrics.
- Next 30 daysOTR test: compare errors, net time saved and credits in a small trial. Is there a paid client invoice?
01 Money flow
Platform reports rising creator activity; money received from new users, paid plan mix and consultant revenue not verified.
02 Early orders or deposits
More Applets were built; no documented client deposits, signed consulting orders or recurring paid workflow contracts.
03 Bottlenecks
Permission failures, workflow rate limits, credit usage, API changes and incorrect AI outputs are credible bottlenecks.
04 Second- and third-tier beneficiaries
Possible QA testers, trainers and low-code maintainers are inferred beneficiaries; no independent order register.
05 Company finances, stock and valuation
No attributable listed-company exposure established; revenue, margins, debt, valuation, customer concentration and live exchange share price/FX not verified.
06 Unit economics and payback
HYPOTHETICAL: S$300 fee - S$160 labour - S$20 software - S$30 testing - S$40 support = S$50 pre-overhead. No actual paid invoice, payback not verified.
07 Insider and institutional activity
No insider sales, institutional stakes or public financing linked to these user metrics verified.
08 Hiring
No independent hiring records for the small-service niche checked; not verified.
09 Procurement and budgets
No government procurement award or budget linked to the announcement; not verified.
10 Foreign-language warning
US English issuer checked. Foreign-language user failures, account-security warnings or cross-country contract terms: not verified.
11 Behaviour and real payment
>50% more weekly builders and ~3x AI Applet builders are self-reported usage, not cleared client payments or retention data.
12 Contradictory readings
High creation volume may reflect free experimentation. Paid subscriptions could remain flat while usage costs rise.
13 Observed missing expected activity
No observed cancellation, churn trend or abandoned-app figures were supplied; missing expected activity not verified.
14 Commercial pilots and approvals
Public product is live; independent production pilot audits, third-party safety approvals and commercial client results not verified.
15 Competitive weaknesses
Zapier, Make, scripts and manual checklists may fit better; switching costs and permissions weaken one-platform reliance.
16 Cross-border spillovers
Integrations span countries; local personal-data, messaging and currency rules may change deployment. Actual spillover not verified.
17 Prior otr outcomes and failures
The source dossier reviewed OTR learning for the distinction between claim and outcome; no directly matching successful/failed OTR Applet forecast certified.
18 Investor awareness and pricing
User publicity and posts don't establish willingness to pay. Listed-market awareness and priced-in status unproven.
19 Strongest disconfirming evidence
Strongest counter: the new numbers are from IFTTT and do not show subscriptions, profits or fewer real errors.
20 Action clock, cost, horizon, stop rule
Start reversible 20-event test after written permission; keep initial cash spend under S$30 hypothetical ceiling. Horizon 30 days. Stop for error, privacy risk or no paid acceptance.
Sources and limits
The four supplied IFTTT links are one source family. Only its November release supplies the growth figures; no independent customer or revenue dataset was supplied. Google, NIST and Zapier provide separate context on controls and monitoring, not corroboration of IFTTT’s results. The October launch and credit system are background.
IFTTT via PR Newswire • 17 November 2026, 08:23 US Eastern; reported builder counts
IFTTT Help • credit system, October 2026
IFTTT Help • plans at a glance
IFTTT • plan comparison; check current terms
Google Account Help • reviewing third-party connections
NIST Small Business Cybersecurity Corner • multifactor authentication
Zapier Help • viewing workflow history, an alternative platform’s documentation
AI powered research