Klaviyo AI: Ask Your Email Data Questions

Email reporting goes wrong in two predictable places: opens that are not real opens, and revenue that is attributed rather than caused. Here is how to ask questions that avoid both.

You can ask Klaviyo questions in plain English and get useful answers, provided you read two numbers carefully: opens, which are inflated, and attributed revenue, which is credit rather than proof of cause. This post explains what to ask, how to read the answers, and how Quiriz's Klaviyo connector, currently in beta, supports it.

Two traps in email reporting

Opens that are not opens. Some mail apps load messages in the background to protect privacy, which records an open whether or not a person looked. Open rate therefore overstates engagement by an amount you cannot see. A campaign with strong opens, poor clicks and few orders is the classic result. Use clicks, orders and revenue to compare campaigns, and keep opens as a smoke alarm only.

Attribution is credit, not causation. Klaviyo credits an order to an email or text when the customer interacted with it within a window before ordering. A customer who was already on their way to buy and clicked an email along the way is credited. So attributed revenue is best read as "revenue email touched", an upper bound on what email caused. That does not make it useless. It makes the trend and the comparison between campaigns more reliable than the absolute number.

The five questions to ask first

1. What share of our revenue comes from email and SMS, by month? Tells you how much of the business email touches, and whether that is stable.
2. Which email campaigns earned the most revenue in the last 90 days? Ranks by revenue, with recipients beside it so size does not hide quality.
3. Which campaigns have good opens but few clicks or orders? Finds the ones inflated by pre-loaded opens, or whose subject line promised more than the body delivered.
4. Which flows earn the most, and are they growing? Automated flows are often the quiet majority of email revenue.
5. Is our list growing? Subscribes versus unsubscribes by month. Revenue today depends on audience size six months ago.

How the connector answers them

The Klaviyo connector is read-only, refreshes daily, and is in beta. Setup uses a private API key you create in Klaviyo with read access to accounts, campaigns, flows and metrics only. It builds four datasets:

• [Klaviyo] Attributed revenue by day: orders and revenue per day, split by the channel Klaviyo credits: email, SMS or other for orders it does not attribute.
• [Klaviyo] Campaigns: one row per email or SMS campaign sent in the last 12 months, with recipients, delivered, opens, clicks, unsubscribes, bounces, spam complaints, orders and revenue.
• [Klaviyo] Flows by month: one row per flow, channel and month with the same engagement and revenue columns.
• [Klaviyo] Audience activity by day: subscribes, unsubscribes, spam complaints and bounces.

Suggested metrics include Klaviyo revenue, Klaviyo email revenue, Klaviyo SMS revenue, Klaviyo store revenue, Klaviyo attributed share, Klaviyo campaign revenue, Klaviyo click rate and Klaviyo revenue per recipient. Attributed share is Klaviyo revenue divided by all order revenue Klaviyo sees, and click rate is unique clicks per recipient. Both are defined once, so the answer to "what is our click rate" is the same whoever asks.

Reading the answers

A worked example, with made-up numbers. Suppose the share question returns 28 percent attributed for the quarter, up from 24 percent. Before celebrating, ask the second question to see whether the increase came from flows or from a heavy campaign schedule. If campaign revenue per recipient fell while campaign volume rose, you are sending more to earn the same, which is a list-fatigue warning. List growth then shows whether unsubscribes are climbing with the extra volume. Each step is one short question against the same stored definitions.

When you compare campaigns across a period, remember that counts add up but rates do not average. For any period, click rate is total clicks divided by total recipients, not the mean of each campaign's click rate. The connector's note says the same.

Connecting it safely

Klaviyo connects with a private API key rather than a login. Create a custom key, not a preset, and grant read access to Accounts, Campaigns, Flows and Metrics only. The read-only preset in Klaviyo also reads customer profiles, which this analysis never needs, so the narrower custom key is the better choice. Because the key is read-only, nothing can be sent, edited or deleted from the analysis side.

One account maps to one key, so agencies working across several brands connect each account separately and keep the questions per client. Before relying on revenue answers, confirm in Klaviyo that your store integration is active, since the revenue datasets are built from the Placed Order metric that integration creates.

Honest limits

• Twelve-month window. Klaviyo limits the span of a report, so history beyond about a year is not available, and a campaign sent just before the window can show partial totals.
• Revenue needs a store. The revenue figures come from Klaviyo's Placed Order metric. With no store connected to Klaviyo, orders and revenue are zero.
• Per-message results. Flows are by flow and month. Per-step message results are not read.
• SMS. If Klaviyo will not list SMS campaigns for an account, only email campaigns are included, and the connector says so.
• Attribution is Klaviyo's. Quiriz reports the attribution Klaviyo made; it does not re-attribute orders.

For joining email results to store revenue see Klaviyo and Shopify: Revenue by Flow vs Campaign, and for a standing view see Klaviyo Dashboard Without Another Tool. Store-side workflows are at Quiriz for ecommerce.

Frequently asked questions