How Should a Shopify Ad Agency Find New Clients in 2026?

On October 28, 2026, Google switched on AI Mode ads in ten more countries, and advertisers in those countries cannot opt out of the surface separately from Search. If you run paid media for Shopify brands, that is a dated external change your clients' campaigns are already subject to, whether or not anyone has told them yet. This article does two things with it. First, it walks one campaign you could run in January, from the news to the segment to the conversation. Second, it generalises the shape, so you can run it again next quarter against a different change. There is no email copy here and no promised numbers; the part worth having is the reasoning.

What actually changed in Google AI Mode ads

PPC Land reported on October 28, 2026 that advertisers in ten more countries can't opt out of Google's AI Mode ads. The specifics that matter to the brands you sell to:

• Existing campaigns are automatically eligible. Search, Shopping and Performance Max campaigns can serve in AI Overviews and AI Mode. Google's framing, per that report, is that advertisers need no new campaigns or separate budgets.
• There is no switch. Advertisers cannot target those surfaces on their own, and cannot opt out of them separately from Search.
• There is no separate reporting line. Impressions, clicks, conversions and cost from AI Overviews and AI Mode are folded into standard campaign reports, with no distinct performance segment.
• The map is dated. Ten countries were live on October 28, 2026: Australia, Canada, India, Kenya, Malaysia, New Zealand, Nigeria, Pakistan, the Philippines and Singapore. Brazil, Mexico and the UK are planned for Q4 2026.
• Multiturn ads are narrower. Ads inside follow-up questions are US-only so far, and labelled "Sponsored".

On performance, the one statement from Google in that reporting is that click-through rates are already similar between sponsored and organic product list items within AI Mode. That is the whole of it: not a lift figure, not a return-on-ad-spend claim, and extending it into one would be inventing a number.

For wider context, True Interactive's own reading of Google's Q3 2026 earnings announcement is that Google acknowledged AI Mode's growing popularity was contributing to slower search advertising revenue growth. That is their characterisation, not a quoted executive statement, and it is commentary rather than a disclosed figure.

Why this is a merchant problem, not an industry headline

Three consequences land on the merchant rather than the industry.

Blended performance now mixes a surface with no reporting line. A month-over-month comparison spans a changing surface mix, and nothing in the report separates it out. That does not make the numbers wrong; it makes them harder to attribute, which matters most where someone senior reads the deck.

Feed and product data move from hygiene to placement input. When listings are assembled into an answer rather than a results page, titles, attributes and availability do work they were not doing before. For a large or messy catalogue, that is a real project.

The merchant will hear about this from someone. A newsletter, a board member, a competitor's agency. The open question is who gets there first, and whether they had anything specific to say about this store. That is the whole opening, and it is worth a calm paragraph rather than an alarm.

The five parts of a trigger campaign (and the three everyone skips)

Any campaign built on an external change has the same five parts.

1. The trigger. Dated, external, and not chosen by the buyer. Google turning on a surface qualifies. Your funding round, your service page and your rebrand do not; you chose those, and the reader was not consulted.
2. The segment. Who is measurably worse off because of it. Here: stores running Shopping and Performance Max at real spend, in a country Google has switched on, with enough catalogue complexity that the reporting gap costs them something. Not "Shopify brands".
3. The observable. What you can see from outside that proves a store is in the segment — platform and plan, revenue band, installed apps, current ad activity, paid-media hiring. This is the step that takes longest to build in-house, which is why most teams skip it; it is also the layer we sell as outbound for companies that sell to e-commerce stores.
4. The message. What the store loses, in the store's words. Not what you do, and not a summary of news they could have read themselves.
5. The ask. Matched to the calendar. In November and December, Shopify brands are inside Black Friday and Cyber Monday and will not start a project. Ask for a January slot.

Most campaigns have parts 1 and 5 and skip 2, 3 and 4. A true statement sent to an unsegmented list still reads as a mailshot, because nothing in it could only have been written to that store. The three skipped parts cost time rather than money, which is exactly why they go first.

Working out your segment before you buy any data

The segment is arithmetic, not a persona. Write your own numbers into this chain, in order:

1. Your minimum monthly retainer — the fee below which the engagement is not worth staffing.
2. The monthly ad spend at which that fee is a sensible share of the budget for the client.
3. The store revenue that can sustain that level of spend at the margins in your clients' category.

The output is a revenue floor: the point below which a store cannot be a client of yours, however well it fits otherwise. Apply it before buying a single record.

An observation from our own work, not a statistic: when we build these lists for agencies selling to e-commerce brands, applying the floor usually removes most of what the agency had been calling its addressable market — commonly 70 to 90 percent of it. We have no published figure for that and are not presenting it as research; it is what we see when the rule meets a real list. The useful part is what is left, because that is the set you can work in a quarter.

What survives still has to be described field by field before it is a list: platform, revenue band, category, ad activity, the person who owns paid media, exclusions. That is covered in what belongs in an e-commerce prospect list.

If you have no minimum retainer, stop here. An agency that will take any store at any spend has no segment, and a trigger without a segment is just news. No amount of list spend fixes it, because there is no rule to filter on. Set the floor first and lose the accounts it excludes; that is the point of the exercise.

Where the data for the observables comes from

Each observable in part 3 has a source you can check from outside. None tells you intent, and treating one as proof of a budget is how a good list produces bad messages.

• Platform and plan. Detectable from the storefront itself. Plan tier is a contract fact rather than a need, so it belongs in the fit column, not the reason-to-write column. The detection methods are in how to identify Shopify Plus stores.
• Revenue band. Estimated from third-party store data, labelled as an estimate, with the source recorded. Not inferred from follower counts or how the site looks.
• Apps installed and removed. What a store runs tells you what it committed to; what it removed tells you what it tried and dropped. The second is the more interesting field and the one most lists omit.
• Ad activity. Public ad libraries and storefront tags show whether a store is spending now — the first filter for a campaign built on an ad-surface change.
• Hiring. A paid-media job post is a budget cycle opening, with a date attached, and often names the owner of the problem.

This is the layer we build. For companies selling to e-commerce stores, we assemble store-level signal lists from tech-stack, revenue and app-install data, write per-store opening lines off what is actually visible on each store, and run the campaign on fresh sending infrastructure rather than whatever domain the last agency burned. The reason it is structured that way is the one above: the trigger is worthless until you know which stores it is true for. That is our work with companies that sell to e-commerce stores, and it is deliberately the least glamorous part of the campaign.

What the first touch has to do

The first line has one job: prove you looked at this store before you wrote to it. Everything else is downstream of whether the reader believes that in the first two seconds.

Three things are true of an opening line that does the job:

• It survives the deletion of your company name. If the sentence still makes sense with your agency removed, it was about the store. If it collapses, it was about you.
• It references something observable about this store. Not the category, not the platform, not the news. Something a person could verify by opening the site.
• It names a cost to them. Not a feature of your service — what this change makes harder for this store, this quarter.

We are not publishing example email copy here, and the reason is practical rather than precious. Copy that hundreds of agencies can paste stops being personalisation the week it is published, and your prospects receive every other version of it. The structure above is reusable; a paragraph of finished copy is not.

The January calendar, backwards

If the ask is a January slot, the work happens in November and December. Backwards from a first send in the week of January 7:

• Early January: first sends, with reply handling staffed from day one.
• Mid-December: domains and inboxes warming. This part has a fixed minimum duration, and it is why a December decision cannot produce a January send.
• Week of December 15: list built, floor applied, segments split, exclusions loaded.
• Before the holidays: message briefs agreed, one per segment, each tied to a specific observable, so nobody writes positioning in the last week of December.

Why November is a build month rather than a send month: your prospects' clients are retailers, and from mid-November into early December they are inside their highest-volume trading period. Someone reading a thoughtful message in that window is not going to start a new vendor conversation. That is our judgement from running campaigns into this market, not a measured finding, and it may not apply if your segment is not retail-dependent.

When this doesn't work

Below a certain retainer, the economics do not carry the work. List building, sending infrastructure, reply handling and the sales time on the other end cost the same whether the resulting client pays a little or a lot. We are deliberately not printing a threshold figure, because the honest one depends on your gross margin, your win rate from a first conversation and how long a client stays. Run that arithmetic before you run the campaign; if it does not clear, the answer is a different offer, not a bigger list.

If you cannot articulate who you serve, a trigger will not save you. This is the missing retainer floor seen from the other side. A campaign is only as specific as the segment behind it, and a generalist sending a specific message is making a promise its positioning does not support.

If delivery is already at capacity, more meetings is the wrong project this quarter. Booking conversations you cannot staff is worse than not booking them, and it is visible to the market you just introduced yourself to.

One pre-emptive answer, since the alternative usually proposed is answer-engine optimisation instead of outbound. Ahrefs looked at 10,000 brands and 260 million prompts and reported that ranking Reddit threads correlated at 0.75 with AI mentions, while stating plainly that "we can't say mentions cause AI visibility". Take the caveat seriously: it is a reason to be present where your buyers research, on a timeline you do not control, not a reason to swap a dated campaign for a visibility programme this quarter.

Frequently asked questions

Can advertisers opt out of Google AI Mode ads?

No. Search, Shopping and Performance Max campaigns are automatically eligible to serve in AI Overviews and AI Mode, and advertisers cannot opt out of those surfaces separately from Search. Google's position, as reported by PPC Land on October 28, 2026, is that no new campaigns or separate budgets are required — which also means there is no switch to turn it off.

Is there separate reporting for AI Mode ad performance?

No. Impressions, clicks, conversions and cost from AI Overviews and AI Mode are folded into standard campaign reports, with no distinct performance segment to break out. Practically, that means month-over-month comparisons now span a changing surface mix, and should be read that way.

How should a Shopify ad agency find new clients in 2026?

Pick one dated external change your prospects' clients are affected by, define the narrow segment it is actually true for, and reach only that segment with a first line that proves you looked at their store. The common failure is not volume or channel — it is sending a true-but-generic message to a list that was never segmented. The five-part structure earlier in this article is the version we run.

What are good outbound triggers for a PPC agency selling to e-commerce brands?

The best triggers are external, dated and observable from outside the company: a platform or ad-surface change like Google's AI Mode rollout, a store adding or removing apps in your category, a visible change in ad activity, a paid-media role opening, or a revenue band crossing the level where your retainer makes sense. Avoid triggers that are only about you — your launch, your funding, your new service page.

Should an agency do outbound itself or have it done?

Do it yourself if you have someone who owns list building, sending infrastructure and reply handling as their actual job, and your retainer floor covers the cost of them learning. Have it done if the constraint is that nobody owns it and the founder is the fallback. The honest test is whether anyone could tell you, today, how many stores are in your segment — if nobody can, the problem is the list layer, and that is the part that takes longest to build in-house.

Where to start

If you sell paid media to Shopify brands and you are deciding how to fill the first quarter, the work now is the floor and the segment, in that order; the campaign follows from them. If it would help to talk through your segment, Leon will work the arithmetic with you on a call, and say plainly if the numbers do not support outbound this quarter.

Sources

• PPC Land, "Advertisers in 10 more countries can't opt out of Google's AI Mode ads", October 28, 2026. Checked November 4, 2026. Source for campaign eligibility, the absent opt-out and reporting segment, the country list and rollout dates, multiturn availability, and the click-through-rate statement.
• True Interactive, "Google is bringing more ads to AI search — what should advertisers do?". Checked November 4, 2026. Source for the characterisation of the Q3 2026 earnings announcement above; the page carries no verbatim executive statement, so it is cited as their reading of it.
• Ahrefs, "Which brand mentions track AI visibility", November 3, 2026. Checked November 4, 2026. Source for the 0.75 correlation and its caveat.