How to Use Klaviyo: Set the Billing Preference Before You Send Anything

Most Klaviyo tutorials start with the campaign builder. This one starts with the billing page, because the single most expensive mistake new accounts make is a setting, not a send.

What follows is the order we would do it in: the configuration that decides your bill, then the free-tier work that proves the product fits, then the flows that make it earn its price.

Before anything: set the billing preference

Klaviyo ships with auto-upgrade switched on. "When you first purchase a subscription for the Profile & Email plan or the Mobile Messaging plan, auto-upgrade is toggled on by default" ^[11]. That means crossing a profile limit moves you up a tier and leaves you there, because "Your plan does not automatically downgrade when you reduce profiles" ^[5].

The alternative is flexible sending plus auto-downgrade, which charges a one-time fee when you exceed a limit and returns the plan on its own. The catch is that the two cannot coexist: "auto-downgrade can be used with flexible sending but not with auto-upgrade" ^[11]. You are choosing between them, not adding one.

For a seasonal business, meaning anyone whose list swells for a campaign and shrinks afterwards, flexible sending plus auto-downgrade is almost always the cheaper configuration. For a business growing steadily in one direction, auto-upgrade is fine and simpler.

While you are on that page, note your billing cycle date. Every manual downgrade has to be made more than twenty-four hours before it, and one made later "will not take effect" ^[5]. For self-serve accounts the cycle runs on Eastern Time.

## Start on the free tier, and make it do real work

The free plan is not a demo. It carries 250 active profiles, 500 email sends a month, five dollars of mobile messaging and five dollars of Composer usage, renewing every month ^[2]. That is enough to build a genuine flow against genuine data and see what it does.

It is also the only part of the product where the bill cannot move. Exceeding the limit stops sending rather than charging you: "If you are on the free Profiles & Email plan and are over your profile limit, you won't be upgraded but you also cannot send emails (or set flow emails live) until you are under your active profile limit" ^[5].

Use that deliberately. Import a segment of your list rather than all of it, build the flow you most want to validate, and measure. An account that has proved one flow works is in a much better position to argue about a price than one that has proved the software installs.

Connect the store first, not the list

Klaviyo's value comes from what it knows about a shopper, and it knows nothing until the store is connected. The vendor publishes individual integration pages for Shopify, BigCommerce, Magento, WooCommerce, PrestaShop, Wix, Square Online and Salesforce Commerce Cloud.

On Shopify the path is shortest. The listing describes a guided setup that produces a sign-up form, a welcome email and an abandoned-checkout flow without leaving Shopify Admin, and the account arrives with 60+ pre-built flows and 150+ templates available ^[19]. For a new store that is a working programme in an afternoon.

Connect the store before importing a list, in that order. A list imported against an empty store has no purchase history attached, so the segments you build from it are list segments rather than behavioural ones, and rebuilding them later is more work than waiting an hour now.

## Understand what a profile is, because it is what you pay for

A profile is billable when it can still receive at least one kind of marketing. Not when it opens. Not when it buys. When it is reachable.

Two consequences follow immediately. The first is that a long tail of addresses collected at checkout years ago is on your invoice whether or not anyone mails them. The second is that test and development traffic counts: one reviewer reports being "charged almost three times my normal monthly bill because of profiles that were obviously generated during development/testing" ^[20]. If you have a staging environment writing to the same account, separate them now.

The lever is suppression or deletion, and the timing matters more than the action, because the count that bills you is the one at the end of the billing cycle ^[5]. Clean before the cycle closes, not after.

Build the four flows that pay for the account

Every ecommerce account earns its price on a small number of automations, and they are the same ones everywhere: welcome, abandoned checkout, post-purchase and win-back.

Klaviyo ships pre-built versions of all four. The value of building them yourself is not the structure but the conditions: which products trigger what, how long the delays are, who is excluded. Start from the pre-built flow, send it to yourself, then change one thing at a time.

Abandoned checkout is the one to get right first, because it has the shortest path to revenue and the clearest signal. Welcome is second, because it runs against your highest-intent audience. Post-purchase and win-back can wait a month.

Resist building twenty flows. An account with four flows that are measured beats one with twenty that are not, and the four are what you will point at when you justify the cost.

## Use segments, not lists

This is the habit that separates people who get value from Klaviyo from people who pay for it.

A list is who signed up. A segment is who matches a condition right now: bought twice in ninety days, browsed a category and did not buy, has not opened in six months. Segments update themselves; lists do not.

Klaviyo's data platform resolves returning visitors to known profiles and carries purchase history, browsing behaviour and predicted value on the profile ^[12], which is what makes behavioural segments possible at all. If you are sending to lists, you are using a platform at the price of a platform and getting the output of a mailing tool.

Build three segments before your second campaign: engaged, lapsed, and never purchased. Most accounts need little else for months.

Add channels deliberately, one at a time

Email, SMS, RCS, mobile push, WhatsApp and social all run against the same profile. That is the architectural advantage and also the way bills grow without anyone deciding.

Mobile messaging has its own meter and its own five-dollar monthly allowance on every account ^[10]. The vendor publishes only that the rate is "the list (standard) price per message for a given channel, country, and number type", with no rate card, so the only reliable forecast is your own first two cycles of usage.

Add SMS when you have an email programme that works and a reason the message belongs in a text rather than an inbox: order status, back in stock, a genuinely time-bound offer. Adding it because it is available is how accounts end up with two channels performing like one.

## Know what the AI costs before you turn it on

Composer is twenty dollars a month, includes five dollars of usage even on the free plan, and bills beyond that at ten dollars per thousand credits ^[2]. The vendor describes it as taking an objective and building "the audience, draft the content, and map out the send strategy" ^[14].

Customer Agent is fifty dollars a month for seventy-five resolved shopper conversations ^[5]. Its published example is practical rather than magical: "When a customer asks where their order is, Customer Agent can find the order, check the status, and send an update automatically" ^[14].

One asymmetry to know before enabling the second: Customer Agent cannot be capped. Using it requires enrolment in flexible overages or auto-upgrade ^[5]. Composer can be left to its allowance; Customer Agent cannot.

Set up review collection early, if you sell physical goods

This is the piece most accounts find a year late, and it compounds.

Klaviyo's review product requests reviews "from eligible customers through email, text, push, and WhatsApp with pre-built flows", lets you ask custom questions that capture "shopper preferences, sizes, and habits", writes those answers back to the profile, and syncs the resulting reviews to Google Shopping, Shopify, the Shop app and Meta stores ^[12].

The reason to do it early is not the reviews. It is the second sentence. Custom review questions are the cheapest structured data you will ever collect about a customer, they arrive attached to a real purchase, and they feed segments you cannot otherwise build. An account that has been asking size and preference questions for six months can personalise in ways one that started last week cannot, and no amount of budget closes that gap.

One operational warning from the review corpus: a reviewer reports the review feature sending live messages during what they understood to be a trial run. The vendor's own documentation notes that review requests queued while you were at a plan limit "will be sent out immediately after upgrading" ^[8]. Check what is queued before you change plans.

Reviews bill on orders placed within a billing cycle, which is a fifth denominator alongside profiles, sends, credits, tickets and conversations ^[5].

## Measure attribution, not opens

The number that justifies this platform is revenue attributed to a flow. It is also the number that does not transfer if you ever leave, which makes it worth establishing properly from the start.

Set the attribution window deliberately rather than accepting the default, and write down what you chose. Most disagreements about whether email "works" are disagreements about the window, conducted by people who do not know they are having that argument.

Then report on three things monthly: revenue per recipient by flow, the share of total revenue attributed to flows against campaigns, and the trend of both. Opens have been unreliable since privacy-protecting mail clients began prefetching images, and clicks tell you about the subject line rather than the programme.

If flows are earning less than campaigns after three months, the flows are misconfigured. That is the single most useful diagnostic the platform gives you, and it is visible from month one.

Check your usage every cycle, not every quarter

The account usage page shows consumption per product for the current cycle and the previous three, six or twelve ^[5]. It is the only place where the shape of your bill is visible before it arrives.

Two numbers are worth a monthly glance. The first is your profile count against your plan limit, because that one drives an automatic upgrade. The second is which side of eighty-five percent of the next tier you are sitting on, because the vendor publishes that as its own break-even: "Once you're using 85% or more of the next tier, it's better to upgrade" ^[5].

Below eighty-five percent, flexing is cheaper. Above it, you are paying a premium to stay small on paper.

## What to do in your first month, in order

Set the billing preference and note the cycle date. Connect the store. Import a segment of your list rather than all of it. Build the abandoned-checkout flow from the pre-built version and measure it for two weeks. Build welcome. Create three segments. Then, and only then, look at the campaign calendar.

If at the end of that month the flows have not earned more than the plan costs, the problem is not the configuration and no further feature will fix it. That is a useful thing to know in month one rather than month nine.