When Phone Follow-Up Makes Sense for an Abandoned Shopify Checkout

Most abandoned checkouts were never a sale you can win back

The pool of abandoned checkouts is much larger than the slice of it a campaign can win back. Baymard Institute, whose running list of abandonment studies goes back to 2006 and was last updated in September 2025, puts the average documented online cart abandonment rate at 70.22 percent [Baymard Institute, cart abandonment rate list, last updated 2025-09-22]. The number that matters for a phone campaign is smaller and more specific: in Baymard's survey of why shoppers abandon, the single largest reason, 42 percent, was that the shopper was "just browsing" or not ready to buy. That group did not lose interest at the payment page. They never intended to check out, and a call does not change that.

The recoverable slice is the shopper who did intend to buy and hit friction. Baymard's same body of research counts 17 percent abandoning specifically because the checkout was "too long or complicated," plus smaller groups who abandoned over payment options, forced account creation, or a delivery that did not look trustworthy. Those shoppers were willing; the store got in their way. The practical point for anyone deciding whether to add phone follow-up is that the addressable audience is narrower than "everyone who left a cart": it is the subset who got far enough to signal intent and then stopped for a reason a human conversation could actually fix.

That distinction is what separates a phone campaign that pays for itself from one that annoys a list. A tool that dials the entire abandoned-checkout segment is mostly calling people who were never going to buy, and the few who convert do not cover the cost or the complaints the rest generate.

What each recovery channel actually costs per contact

Three channels can reach an abandoned checkout, and they are not interchangeable. Email is the cheapest to send and the easiest to ignore, which is why it is the default. SMS sits in between: it is read faster and lands closer to the moment of decision, but it costs money per message and it carries its own consent rules under the TCPA. For copy written to stay inside those rules, the site keeps a set of SMS templates for customer updates, promotions, and reminders.

A phone call is the most expensive of the three per contact and the most intrusive, and that combination is exactly why it only makes sense on a filtered list. The cost runs past the per-minute rate. It includes the human or AI time on every call, the compliance overhead of dialing correctly, and the brand cost of calling someone who did not want to hear from you. On a low-value cart, the arithmetic rarely closes: if one saved order is worth $40 and a call program costs a few dollars per contact, a conversion rate in the low single digits leaves you underwater before the first invoice.

Where the math changes is cart value. A store selling furniture, electronics, or made-to-order goods has carts in the hundreds or thousands of dollars, and a single recovered order can pay for a whole day of calls that went nowhere. That is the profile where a call stops being a rounding error and starts being worth the cost. The same logic that governs which dialer mode a team should run applies here at a smaller scale: match the tool to the volume and value you actually have, not to the demo.

Consent comes before copy, and the bar is higher for a voice call

Consent decides which checkouts a store is allowed to call at all, and it comes before any question about copy or offer. It is also the step most teams skip when they evaluate a cart-recovery tool on features. Under the Telephone Consumer Protection Act, marketing calls and texts that use an automatic dialing system or an artificial or prerecorded voice require the recipient's prior express written consent. The ruling that settles this for AI voice agents is a February 8, 2024 declaratory ruling (FCC-24-17), in which the FCC confirmed that AI-generated voices count as "artificial" under the statute [FCC, Declaratory Ruling, FCC-24-17, February 8, 2024]. That means an AI agent dialing abandoned carts is not treated like a live human making a call. It is treated like the automated marketing call it is, and it needs the same consent a robocall would.

A shopper abandoning a checkout does not, by that act alone, give consent to be phoned. Leaving an email address is not the same as giving a phone number, and giving a phone number for delivery or order updates is not the same as agreeing to marketing calls. A store that wants to call abandoned checkouts needs to have collected that consent at the point of capture, for that purpose, and be able to show it. Skipping this step means the calling list gets built on numbers the store was never allowed to dial, which is a worse problem than the fine it risks.

Two more pieces of the current rulebook matter. First, since April 11, 2025, the FCC requires callers and texters to honor a revocation request made through any reasonable means, including the keywords stop, quit, end, revoke, opt out, cancel, and unsubscribe, and to treat reasonable natural-language requests the same way, under a "totality of the circumstances" test [FCC, TCPA consent revocation rules, effective April 11, 2025]. A caller may not force shoppers to use one specific opt-out method. Second, the FCC's 2023 "one-to-one consent" rule, which would have required consent tied to a single seller and limited to calls "logically and topically related" to where the consent was captured, was vacated by the Eleventh Circuit in Insurance Marketing Coalition Ltd. v. FCC, 127 F.4th 303 (11th Cir. 2025), and the FCC formally removed it in a final rule in September 2025. The practical reading is that federal consent can be broader than the vacated rule implied, but state law still varies, and states such as Florida run their own mini-TCPA statutes with different definitions, a distinction covered in more detail in the site's dialer and call-abandonment compliance guide.

Where a purpose-built voice layer fits, and how Consio runs the outbound campaign

Consio is an AI phone platform built for Shopify merchants, and its outbound side is aimed at exactly this question: it runs call and SMS campaigns that ring abandoned checkouts, VIPs, and repeat buyers, and can text the checkout link to the shopper during the call. It reads live Shopify order, tracking, and catalog data, so an agent knows what was in the cart before the conversation starts, and it lists integrations with Klaviyo, Gorgias, Zendesk, Kustomer, and Richpanel, so a campaign can be built from segments a store already has rather than a list rebuilt from scratch.

Concretely, a campaign runs in the order this article has been describing. You start from a segment, either one built in Consio or one synced from Klaviyo, and the Ecommerce Power Dialer works through it. The agent has the shopper's Shopify order and cart data in front of it, so the conversation opens on what they actually left behind rather than a generic script, and it can text the checkout link during the call so the shopper can finish the order while still on the phone. Each call ends with a summary and transcript attached to the customer record, and Consio reports the revenue back against the specific campaign and agent that produced it. Attributing revenue to a specific phone call is the piece most stores cannot do today, and it is what turns a hunch about calls into a number you can set against the email flow.

The reason to mention it here rather than treat the category generically is that the tool has to do two things a generic dialer does not: pull the checkout data that tells an agent which carts are worth a call, and report revenue back per campaign and per agent so a store can see whether the calls earned their cost. Consio says it attributes recovered revenue to the specific call or campaign that produced it, which is the measurement a before-and-after number cannot give you. On the inbound side, the same platform answers support calls and auto-logs them into the helpdesk, the pattern covered in the site's checklist for adding a voice channel to a Shopify support stack. This piece is about the outbound direction; the two are different jobs for the same line.

The most common objection to calling a shopper is that they will resent it, and it is the reason most stores never test the channel. Cam Stajer, CEO and founder of Kala, frames the call as reassurance rather than an interruption: "Because our products are premium and technical, customers need reassurance. Consio provides that instantly, and the result is clear: more trust and more sales." That is one merchant's read, but it lines up with the logic above: the shopper who reached checkout, entered contact details, and stopped for a fixable reason is not a cold prospect being interrupted, they are someone who wanted the order and stalled. Pavan Bapu, co-founder and CEO of Houndsy, describes the channel in revenue terms rather than service terms: "Consio has become one of our highest performing revenue channels, full stop."

Both quotes come from Consio's own customer page, so treat them as the vendor's chosen testimonials rather than an independent survey. The harder evidence is the case with a control group. Houndsy, a Shopify pet-furniture brand, tested Consio's outbound calling against a holdout group that received no calls, and according to Axel Guiral of Consio the called group converted 23 percent better than the holdout, worth $55,879 in attributed revenue across 3,855 completed calls in a single month. That is one merchant's result over one period, reported by the vendor rather than an independent auditor, so treat it as a prompt for questions on a sales call rather than a figure to plan against. What it does illustrate is the measurement setup, a holdout group, which is the only way to know a call caused an order instead of merely preceding one.

Sequencing a call against the email and SMS flow you already run

A phone campaign layered on top of an existing email or SMS recovery flow creates a coordination problem before it creates revenue. If a shopper gets a reminder email, a reminder text, and a phone call about the same cart, the calls stop feeling like service and start feeling like pursuit. The fix is to treat the call as the last step in a sequence, triggered only after the cheaper channels have run and failed to convert, and only for the carts that clear the value threshold.

In practice that means deciding, before launch, which segment gets a call and which does not, and holding the call for shoppers who ignored an email and a text rather than hitting them all on the same day. Consio's model, reading the existing Klaviyo segments and campaign logic, is built for that sequencing rather than against it, but the sequence itself is a decision the store has to make. The bug to avoid is a call program that runs in parallel with the email flow, double-contacting the same shopper and inflating the call campaign's numbers with orders the email would have recovered anyway.

Deciding whether it applies to your store yet

Whether to add phone follow-up comes down to a short list of conditions, not to store size. If average order value is low, if phone numbers are rarely captured at checkout, or if consent for marketing calls was never collected, the honest answer is no for now, and a documented email and SMS sequence is the better use of the same time. The cost of a call program is real whether or not it converts, and calling a list that was never consented for it is a risk with no upside.

If the opposite holds, high-value carts and a real base of opted-in phone numbers, the entry cost is lower than most teams assume. Consio's published pricing starts with a free plan for account and campaign setup, then $30 a month for 100 connected minutes, $60 for 400, and $120 for 1,000. Extra minutes run $0.10, outbound texts $0.016 each, and AI-handled minutes $0.30, all with unlimited seats and month-to-month terms. That is low enough that budget stops being the deciding factor; what matters is whether the store's cart profile and consent base support calling at all. Run your own numbers against the checklist above before adding a channel that costs more per contact than the two you already run. If most of those answers come back yes, the entry plan is cheap enough to test on one segment for a month rather than settle in a planning meeting.

Frequently asked questions