Pér
The customer decision loop¶
Marketing work on customer data follows a shape, whatever the campaign: someone works out what is happening with a group of customers, proposes what to do about it, agrees on a plan, makes it happen in the tools, and carries what they learn into the next round.
In brief: Understand → Recommend → Approve → Act → Learn
This is the customer decision loop, and it is what Pér is built around.
Knowing the loop is the quickest way to find your way around Pér, because every part of Pér serves one of its stages. This article is a close-up of one stage at a time.
Understand¶
Pér works out what is actually happening with your customers.
Nothing further along the loop is worth anything if this stage is wrong, which is why Pér shows its work rather than handing you a conclusion. You can follow what it looked at and check it against what you know.
You do this stage in conversation. Ask a question in plain language and Pér queries your tenant’s own customer data to answer it — the same identity-resolved records your organization already relies on, read under your own Amperity access. As it works, it says in one line what it is about to look at and why, so you can see the direction of the analysis before the answer arrives.
What Pér brings to the question, beyond the data: the company context your tenant has set up, the memories you have given it, and what it can find on the web.
Recommend¶
Pér proposes something worth doing, and makes the argument for it.
A recommendation is a proposal with its evidence attached: the claims it rests on, the numbers behind those claims, and where in your data they came from. It also carries a confidence grade and the reasoning behind that grade, including what Pér could not establish. You are meant to be able to disagree with it on the evidence.
Recommendations gather in the Portfolio, which is where Pér puts the work it thinks is worth your attention. They are drawn from your tenant’s data, your company context and your memories, together with what Pér has already carried out for you and what it has already proposed — so the Portfolio does not keep re-proposing work you have started.
Note
Pér produces recommendations when someone asks for a fresh set, not continuously. You can see when the current set was produced.
Approve¶
Nothing happens in Amperity until a person says so.
Because approval is a real gate rather than a formality, you can let Pér do the work of figuring out what should happen without giving up control of whether it happens.
Acting on a recommendation authors a plan: a titled list of steps, each one a concrete change to your Amperity tenant. You can also ask for a plan directly in conversation, and Pér will author one. Either way, you read the steps before any of them runs.
When you approve and run a whole plan, you approve the plan — not each write inside it one at a time. Pér then re-checks at every step whether it may still go on, stops at any step that needs a person, and records which steps it approved on your behalf.
Important
Read a plan’s steps before you approve it. One approval can kick off a sequence of Amperity writes, and a write that has run cannot be undone from Pér.
Act¶
The approved steps run in Amperity.
This means the audience now exists, the campaign is configured, or the model is trained. Nothing is left for you to go and replicate by hand.
Steps run in order, and some of them start work that takes a while, for example, training a model or running a database. Pér waits for those and carries on when they finish. If a step fails, it says what went wrong rather than leaving the plan stuck, and you can fix the cause and run that step again.
Learn¶
What happened feeds what Pér does next.
This stage is what makes the second round better than the first — both for you, in having a record to look back at, and for Pér, in not starting cold.
Three things carry forward:
What you told Pér to remember. Memories persist between sessions, so a preference, a rule or a correction only has to be given once.
What Pér did. The Activity log keeps a record of its actions and its recommendations.
What Pér produced. Reports it writes and files you give it are kept as artifacts, which you can come back to or share with your team.
The next set of recommendations is drawn with all of that in view, including the steps you have already carried out and the recommendations already in the Portfolio.