Your agency is probably already using AI somewhere. Even though 86.4% of marketing teams report using AI in 2026, according to HubSpot’s State of Marketing Report, agencies are still unsure how to turn AI into actual revenue.
That gap was the starting point for our webinar with Mitchell Leiman, CEO of iPromote, and Rachel Nulman-Schapiro, CMO of inTandem by vcita. Both work closely with agencies, and both have watched the conversation move from AI enthusiasm to a harder question: how does AI drive real advertising ROI for real SMB clients? Here’s what they shared.
Why AI adoption isn’t showing up in advertising ROI
Adoption isn’t the problem. For most agencies, the question is no longer whether to use AI but where it actually moves the numbers.
US AI-powered ad spend is projected to reach approximately $26.7 billion in 2026, roughly 12% of total marketing spend, according to eMarketer, and 67% of agencies and brand marketers are prioritizing agentic AI for ad buying and campaign execution, per the IAB’s 2026 Outlook Study.
What’s missing is AI across the full revenue path. As Mitchell put it, success doesn’t come from one tool bolted onto the campaign, but it comes from AI running through the whole funnel: who sees the ad, what the ad says, what happens when someone responds, and what happens after that.
1. Smarter targeting built for small budgets
The large ad platforms are trained to perform with large budgets, so a tool like Google Performance Max works best when an advertiser spends tens of thousands of dollars a month. Most SMB clients spend a few hundred or a few thousand. That’s the core challenge for agencies serving local businesses, and it’s where AI earns its place.
There are three layers, as Mitchell describes:
- Audience modeling. When a client has no first-party data, intent-based targeting answers the question directly: given a plumber in Cleveland, Ohio, who’s the right audience for this message?
- Omnichannel reach with pre-campaign intelligence. Instead of defaulting to one channel, the campaign starts from what has worked before across search, social, display, and video, with the right media mix and ad types from day one.
- Predictive bid and budget optimization. Spend gets allocated at launch, then reallocated as real results come in. Budget leaks, whether a weak placement or a channel that isn’t working, get closed continuously instead of at the monthly review.
Mitchell calls the approach “launch and learn”, where every campaign retrains the model, so small budgets get the kind of learning that used to need enterprise spend.
2. Personalized creative at the volume platforms now expect
Creative has always been the bottleneck for SMB advertising. Small businesses don’t have design teams, and producing multiple ads for a small budget rarely made economic sense for an agency.
That’s changed, and the platforms are pushing it; as a result, Meta and Google now reward advertisers who supply many creative variants and help identify which performs best. And generative AI makes that possible at SMB scale with dozens of image and video variants from a single campaign brief, matched to the audience the targeting model just selected. 79% of marketers plan to increase spend on generative AI content in 2026, up from 70% in 2023, per eMarketer.
As Rachel noted, the result is enterprise-grade capability on a small business budget. Mitchell’s caveat: it’s more accessible than ever, but it’s still not easy, which is exactly why agencies matter.
In practice: iPromote worked with an agency whose strength was SEO and web presence rather than paid media. When a large dental group client asked for help with digital advertising, the agency didn’t have the capacity to run it alone. Working with iPromote, it launched the campaigns within weeks and over roughly a year, leads increased 400%, and cost per lead fell from about $53 at the start of the year to about $37 in recent months, against a national average of about $73.
3. Capturing every lead the campaign generates
Campaign ROI is often lost after the lead has been delivered to the client. The leads arrive, but the plumber is under a sink, the roofer is on a roof or the dentist’s office is closed when the patient tries to call after hours. The phone rings, nobody answers, and the caller moves on to the next result in Google.
The numbers are stark: the average first response to a new lead takes 42 hours, fewer than 10% of businesses respond within an hour, and many never respond at all, even though 78% of customers buy from the business that responds first. A recent inTandem survey of SMBs also found that many still rely on phone calls as their main way to convert, even though calls are the easiest leads to miss.
When those leads die, the client blames the agency. It’s nobody’s fault: the agency delivered the leads, and the business owner was doing their job, but there is something missing in between.
An AI receptionist fills that gap. inTandem’s voice and chat receptionists share the same “brain,” so they’re trained on the same business data and can answer questions, triage leads, and capture the details that matter on calls, social messages, and website chat, at midnight or whenever the lead arrives.
Rachel shared three lessons from building them:
- Make training simple. SMB owners often set things up from their phone, so the agent interviews them the way a caller would (“Do you have parking?” “Do you serve my area?”). Setup takes about half an hour, and the agent should be delivering value within the hour.
- Unify the channels. Every lead, whatever the source, should get the same experience and land in one inbox.
- Let owners keep teaching the agent. One of the most used features lets owners review chats and improve an answer for next time. It builds trust, and trust drives adoption.
4. Follow-up that converts and retains
Capture is only the beginning. The bottom of the funnel is the stage agencies have the least influence over, and it’s where SMBs have the least support.
inTandem’s lead conversion agent closes that second gap:
- A prioritized feed that shows which leads have the highest potential, filters out spam, and flags what’s urgent.
- Interaction summaries of every call and chat, so follow-up always has context.
- Next best action: call back, send a calendar invite, a price quote, or an invoice, or reply with a message that’s already drafted.
- Voice notes: owners can record a quick note after a real-world conversation, and the agent turns it into action items so the deal keeps moving.
So, for example, Tolt Home Services, a roof and gutter cleaning business in Bothell, Washington, runs the chat receptionist on its website around the clock, with permission to book appointments straight into the owner’s calendar.
For the agency, the payoff comes twice. Clients see faster response, shorter time to deal, and more leads converting into revenue. That shows up as higher average revenue per account for the agency, as clients stay longer and often increase their budgets.
Watch all four AI advertising ROI plays in the replay
The full session includes the case study, the product walkthroughs, and the complete Q&A. If your agency is ready to move from experimenting with AI to profiting from it, watch the recording.
