Marketing automation product comparison: how to assess

What to take away

This marketing automation product comparison uses one buyer scenario and a consistent evidence standard to show how to assess HubSpot, Mailchimp and Microsoft Customer Insights; it sets out a method, not independent test results.
No winner is selected, because each product record rests on vendor statements rather than independent tests.
Score observed workflow evidence separately from contractual coverage, so a passing test cannot hide a weak exit or support term.
Apply the non-product gates firstICO processor contracts, NCSC cloud principles, WCAG 2.2 and PECR consent.
Normalise currency, VAT, term, users and volumes before any total is compared.

Products used to illustrate the method

Three products illustrate the method, in alphabetical order and unranked: HubSpot Marketing Hub, Mailchimp automation and Microsoft Customer Insights. Each had a live supplier page on 5 September 2026, the inclusion cut-off used here.

No account, licence, data import, integration, security setting or quote was tested. The evidence base is vendor material, so every product cell carries a vendor statement label. The marketing automation reviews and comparison methods page explains how the evidence labels are written.

A named shortlist helps a buyer. A named winner does not, unless the buyer's own test produced it.

One common scenario for collecting evidence

The scenario is one England buyer task, frozen so that every candidate faces identical inputs. Freeze an audience file with ordinary, duplicate, incomplete and suppressed records.

One common scenario

Scenario checks

HubSpot Marketing Hub: Import reconciliation, conditional branch, audit history
Mailchimp automation: Suppression, audience counting, preference propagation
Microsoft Customer Insights: Role separation, failed connector, export

Evidence label

HubSpot Marketing Hub: Vendor statement
Mailchimp automation: Vendor statement
Microsoft Customer Insights: Vendor statement

Result on buyer data

HubSpot Marketing Hub: Untested
Mailchimp automation: Untested
Microsoft Customer Insights: Untested

Ask every supplier for the same seven demonstrations:

One common scenario

Import reconciliation across the frozen file.
A conditional branch with a suppressed record.
A failed connector and its alert.
Preference propagation into the suppression list.
Audit history for one contact.
Role separation between author and approver.
A full export of the audience and its consent state.

Keep the version, configuration, input, expected and actual output, and reviewer for each run.

A polished demonstration on supplier data does not count. A criterion the buyer cannot run on its own data is labelled untested, not passed.

Example: a normalised commercial boundary

These figures are illustrative, not a real quote. Suppose one supplier quotes 900 pounds a month and another quotes 10,000 pounds a year for a similar scope.

Compare totals only after the boundary matches:

Normalised commercial boundary

Currency and tax: pounds sterling, excluding VAT, with 20% VAT shown separately.
Period: a 12-month term, not a rolling monthly price.
Users and volumes: 5 seats, 20,000 contacts, 150,000 emails a month.
Services: onboarding, migration, support tier and exit, as separate lines.

On that boundary the annual 10,000 pound figure is 833 pounds a month excluding VAT, or 1,000 pounds with VAT added.

A public pricing page explains a charging unit, not a payable amount, so ask for a written quote at your own volumes. A marketing automation budget template puts each charging unit into a signed-off sheet before negotiation.

Non-product gates that no score covers

The ICO processor-contract guidance lists the topics a processing contract must cover, but it does not settle commercial terms. The NCSC cloud security principles frame configured-security evidence, and WCAG 2.2 defines relevant tests for authoring and recipient routes. None of the three certifies these products.

Processing contractroles, data categories, subprocessors, support, liability and exit.
PECR consent and soft opt-in for every email and SMS journey.
CAP Code compliance for claims made inside automated sends; the Advertising Standards Authority (ASA) administers the Code.
Configured security evidence mapped to the NCSC principles.
WCAG 2.2 checks on authoring and recipient routes.

Deciding without a ranking

Build a decision matrix with one row per requirement and one column per candidate. Link each cell to a supplier response, contract clause or test record, then ask a second reviewer to sample the links before weights are applied.

Stop a candidate when a mandatory requirement fails or cannot be investigated inside the decision window. When evidence conflicts, report both records and their dates rather than the preferred score.

A signed quote may control the purchase while a public page remains what a reader can verify. The commercial contract guide covers processing instructions, liability and exit terms that a matrix cell cannot resolve alone.

Reopen the comparison after an edition, integration or contract change.

Common questions

Why is no product recommended?

The three vendor records rest on supplier claims; no product was tested on a buyer account. A ranking without first-party results would only restate supplier claims, so the output is a shortlist for a named task.

What happens if the buyer cannot run the scenario?

Mark the criterion untested rather than accepting a supplier demonstration. If a mandatory requirement fails or cannot be investigated in time, stop that candidate.

How are conflicting sources handled?

Report both records and their dates, and escalate the mismatch. A signed quotation and a public page can legitimately differ in what each is able to show.

Which rules sit outside the product score?

ICO processor contracts, the NCSC cloud security principles, WCAG 2.2, PECR consent and the advertising codes. They gate a purchase but do not rank products against each other.