Compare

Every survey tool solved the first ten minutes.

Then the clock quietly restarts, because the answers landed in someone else's database. These pages follow the whole race — from writing the first question to your team acting on the answer next to the record — using each vendor's own published documentation rather than a benchmark of our own.

Pricing and plan limits verified July 2026

How we write these pages

Each page opens with what the other tool is better at

That is not diplomacy, it is accuracy. AI authoring is commoditised — every tool here drafts a competent survey in under a minute, and several of them do it better than we do. Typeform's respondent experience is the best in the category. SurveyMonkey's templates are the deepest. Google Forms is free and unlimited. Salesforce Feedback Management is first-party and architecturally right.

  • We cite the vendor's own documentation and pricing, or say plainly that we observed something firsthand in a real account.
  • Where a vendor does not publish a number, we print “Not published” instead of estimating one.
  • We re-verify every named figure quarterly and re-stamp the date. Plans drift; a stale number is our problem to fix, so tell us if you find one.
  • We do not claim things we cannot show you. Several claims that did not survive fact-checking are simply absent from these pages.

Not just our opinion

The industry named this problem years ago

63% vs ~1 in 4

In Temkin Group’s 2017 benchmark of 186 large companies, 63% thought they were “good” or “very good” at soliciting customer feedback — but “only about one-quarter feel equally as confident about their ability to make changes to their business based on those insights”.

Temkin Group / XM Institute, State of VoC Programs 2017 — Q3 2017 CX Management Survey, base of 186 organisations with $500M or more in annual revenues. Self-assessment by the practitioners running those programmes. Quoted in the report’s own words; it says “about one-quarter” rather than printing a percentage. Accessed 25 July 2026.

Two-thirds

of those companies named “integration across systems” a significant obstacle — the No. 1 obstacle for the most mature programmes. 74% expected integration of survey data with CRM and operational data to grow more important.

Temkin Group / XM Institute, State of VoC Programs 2017, figures 4 and 10. Verbatim: “Two-thirds of the companies identified ‘integration across systems’ as a significant obstacle to their VoC efforts” and “74% believe that ‘integration of survey data with CRM and operational data’ will become more important over the next year”. Accessed 25 July 2026.

Six years later the picture had not moved much: Forrester’s 2023 State of VoC and CX Measurement Practices Survey reported that “nearly all programs (96%) regularly collect and analyze surveys” while “only 42% of survey respondents said that stakeholders ask for insights”. (To be fair to those programmes, the same research found 75% close the loop in some form — the gap is in whether anyone upstream is waiting for the answer, not in whether the work gets done.)

Pick the tool you are comparing against

Each page opens with what that tool does better than us, then the plumbing evidence, then the same six-stage table.

The short version

The same six stages, without naming anyone

If you would rather not read a page about a specific vendor: here is the pattern in the abstract. The middle column describes the common non-native model, not any one product.

Stage The usual model AutoSurvey
Authordraft the questions Minutes with AI in most current tools — that part is table stakes now. Describe what you want to learn and AI drafts it in about 15–30 seconds, with the right Salesforce field types. AI authoring is included — no separate Salesforce AI license, no Agentforce project.
Publish gatewhat stops you sending Logic, rating types and question count can move you to a paid tier after the survey is built. One per-org price. Logic, scoring and question types are not the upsell. The deploy job runs in 30–60 seconds.
Distributereach known contacts Email sending is often a paid tier; free tiers hand you an anonymous link, so respondent identity becomes your problem. Salesforce campaign send, list or triggered email, public link or QR, or Experience Cloud. The recipient is already a Salesforce record, so “who answered?” is never a matching problem.
Collectwhat happens at the cap Response caps rarely fail loudly: viewing can lock, live forms can go private, or collection can simply stop. Your own org's limits. No per-response meter deciding whether you are allowed to read your own data.
Land in CRMthe expensive stage An integration tier plus an admin-built field mapping per survey, kept in repair as questions change — or the export, match-by-email and import loop, every wave. This stage does not exist. Responses are created as records in your org, linked to the contact and campaign on arrival. Nothing to map, nothing to sync, no export step.
Actdo something about it Rebuild the follow-up logic on the Salesforce side once the data arrives. Score fields are stamped at submission, so the Flows you already run branch on the answer immediately — Tasks, Cases, alerts, reports, Einstein.

“The usual model” describes the common non-process-native pattern, not any one vendor. Comparisons in this table are categorical.

The arithmetic

So what does the plumbing actually cost?

This is the part usually hand-waved with an invented hours figure. We would rather do the arithmetic in public using only numbers the vendors themselves publish, so you can check every row. Below: the cheapest documented path from “response submitted” to “record in Salesforce”.

Path Software floor, per year Source And you still have to…
Google Forms+ Zapier From ~$240 — Forms is free; Zapier Professional starts at $19.99/mo, and Salesforce is a premium-app-gated connector. At volume it meters: the 2,000-task tier is $49/mo billed annually ($588/yr), about 2.45¢ a task at full use — so a Zap with a trigger plus two billable actions is roughly 4.9¢ per survey response (triggers do not bill) Zapier's published list prices, July 2026. The per-response figure is our arithmetic on those prices — formula shown so you can check it build and own the zap. No native Forms→Salesforce path exists at all
Typeformnative ~$1,092 — Business at $91/mo billed annually (the in-product card we were shown says Enterprise) Typeform pricing page hand-build a 1:1 field mapping per form — no Lookups, no Record Types
SurveyMonkeynative Custom-quoted Enterprise. Third-party data points: a median of $16,482/yr across 193 contracts, plus $2,000–$8,000 typical implementation third-party — Vendr (n=193) and dgt27.com. Not SurveyMonkey's own figures; see the note below hand-map every field on every survey, and expect duplicate-match sync failures
Salesforce FMfirst-party $162,000 — $13,500 per org per month, Starter Salesforce all-add-ons price list (PDF) nothing. It is native. That is what the price buys
The CSV treadmillany tool $0 export, clean, match on email, import, dedupe. Every wave. Forever
AutoSurveynative One per-org price, AI included our pricing page nothing — there is no step here to do

Two notes on honesty. The SurveyMonkey row is the only one where the vendor publishes nothing, so those are other people’s figures about SurveyMonkey — a procurement platform’s median and a consultancy’s implementation range — and we label them that way rather than promoting them into a headline. And the Zapier row is genuinely the cheapest, which is worth saying out loud even though it is the row least flattering to our argument.

That range spans nearly three orders of magnitude, and every row but one has something in common: somebody has to build something before the first response lands. The only incumbent row where that is not true is the one that costs $162,000 a year.

Somebody is paid to do this. Salesforce Ben’s 2026 salary survey of 2,316 Salesforce professionals puts the median US senior Salesforce administrator salary at $109,000. We are deliberately not multiplying that by the mapping count below to manufacture a total — deriving a dollar figure from someone else’s estimate is exactly how bad numbers get made. It is context for who owns the work, not a price tag. Salesforce Ben salary survey — published 16 February 2026, updated 9 March 2026; self-selected sample of 2,316 Salesforce professionals; medians. Accessed 25 July 2026.

The recurring work is countable, even when it is not clockable. We will not tell you how many hours your admin spends, because we do not know your admin — but the vendors document what they must do. Where the mapping is one field to one field per survey, the number of hand-configured mappings you own is fields × surveys × (1 + times a survey changes). Four surveys, ten fields, one revision each is 80 mapping configurations a year. Put your own numbers in; the formula is the claim, not the total. In AutoSurvey that count is zero — not an estimate, just a count of a step that does not exist.

Paste your survey link and bring it over

There are working importers for SurveyMonkey, Typeform, Google Forms and GetFeedback. Paste the public link — or connect the account, for SurveyMonkey — and your questions and answer choices are rebuilt as a native Salesforce survey. Nobody else offers to take the survey off you, because for everyone else you are the one who has to rebuild it.

Importers move the survey definition — questions and answer choices — not your response history. Fidelity is still being verified against real exports, so check less common question types before you send live.