Running an engagement

This guide covers the mechanics of running a client engagement once the workspace exists and the client has an owner. It explains what you and the client can see, which templates fit consultancy work, how co-branded ballots look, and what verifiable evidence you can hand over at the end.

#What you can see, and what nobody can

While you're on an engagement you hold a copy of the client's key, so you read what the client reads: every result, every cohort that clears its floor, and written comments word for word, shuffled and undated. Where the poll allows it, anyone who'd rather not be read can send just the topics they pick instead.

What neither of you can do is find out who wrote any of it. An answer is stored with no name, email address, IP address or time beside it, and the personal link someone answered through isn't kept with what they said. There's no record to look up, for you, for the client, or for us, and no setting anyone can switch on later to create one.

That's worth saying to the client's staff in exactly those terms, because it's the thing they're actually asking. Something like:

"[Your firm] is running this survey for [client]. We'll both see the results, including what you write. Nobody, including us, can see which answers are yours."

One caveat belongs in the same message. A written comment can still give its writer away by what it says: a name, a project only one person ran, a meeting only three people were in. Ask people to leave out names and details only they would know, and they get the full benefit of everything above.

It also changes what you're carrying. You don't hold an identifiable dataset about somebody else's employees, so there's nothing of that kind to protect, lose or be asked to hand over. What we can and can't see says the same thing for a general reader.

#Templates built for this work

Six of the ready-made surveys were written for consultancy engagements rather than for a company polling itself. Each one opens with the questions in full so you can see what you're sending before you send it.

Start from one and edit. They're copies, so changing yours changes nothing for anyone else.

For leavers, run an Exit pool: one standing poll released in groups. You send it on someone's last day. Its minimum group size starts at five, so the first results appear at eight answers and show five; after that, answers arrive in random groups of three to five. A few answers are always held in reserve until the poll closes, so no one's answer can be picked out. Because an Exit pool never closes, it never shows department or tenure breakdowns.

#The ballot names both firms

While the workspace is sponsored by your licence, a respondent sees your firm on behalf of the client above everything else on the response page. Their logo and colours stay where they are; yours is a second, distinct slot. Once the client takes over its own subscription, the ballot carries only the client's brand.

This is co-branding, not white-label, and there's deliberately no way to turn the InviziPoll footer off. A survey carrying only a consultant's brand invites the reading that some third party is collecting identified answers for the employer, which is precisely the fear that stops people answering honestly. Naming both parties, in that order, is what makes the arrangement legible instead of suspicious.

#Cohorts and the floor under them

Tag questions with a demographic dimension (department, tenure) and results come back as cohorts, which is what lets you write "the problem is in one team" without writing "here is who said it". Cohorts appear once the poll closes, not while it's collecting, because comparing a group before and after new answers arrive could reveal one person's answers.

Every cohort has a floor: five, or the poll's minimum group size if that's higher. A group smaller than the floor stays sealed rather than being shown with a caution, because a caution beside a number is still a number. If a department has three people in it, you won't see its slice, and neither will the client. A cohort that could be worked out by subtracting the others is withheld too, and the page doesn't say which or why.

While a poll is open, its first results appear once the minimum group size plus three have answered, and then grow in random groups of three to five, so no change you see can be pinned on one person. A few answers are always held in reserve until the poll closes, so no one's answer can be picked out. At close, everything shows. How batched release works.

#What you can hand over at the end

Two things beyond the report you write yourself.

A signed results document. The counts, the cohorts and the anonymity floors that were actually in force when the survey ran, signed with the workspace's own key. Publish it after the poll closes if you want the cohorts in it: a document published while the poll is open carries no breakdowns. Your client's security team can check that signature against the workspace's public key, which anyone can fetch, without asking us for anything and without trusting your PDF. See Publishing results as a signed document.

An evidence packet. Offered on a shared slice that carries a signature: the same document plus an anonymity attestation naming this survey's real floors rather than generic ones. An unsigned slice has no packet.

That's the difference between telling a board the survey was anonymous and letting them verify it.

#A handout for the kickoff

How this protects you is written for the employees, not for the buyer. Print it, attach it to the kickoff email, pin it to a noticeboard. Participation is the whole game in this work, and the fastest way to raise it is for people to understand what is and isn't being collected before they're asked to answer.

#Related

• Creating a client workspace
• Inviting the client's owner
• Batched release and the exit pool
• What a sponsored workspace means