Profound, Peec, ReFiBuy, and Us: What Each One Actually Does

Some version of this comes up on nearly every call. We’re looking at Profound. How are you different? What about ReFiBuy?

So here’s the map, written down once. I’ll say up front that I run one of the companies on it, which you should weigh accordingly. What I can do is describe the others accurately, using their own material, and be specific about where we’re the wrong call.

Four groups. They’re converging, which is part of the story.

1. The SEO suite you already pay for

Semrush AI Toolkit. Ahrefs Brand Radar. Conductor.

Every established SEO platform has bolted on an AI visibility module. They track how often your brand shows up in AI answers, usually alongside your existing rank tracking.

Worth knowing: when we measured which firms ChatGPT actually names on AI-visibility questions, Semrush came up in 12.8% of answers and Ahrefs in 10.5% — more often than Profound at 8.1% or Peec at 7.0%. The incumbents have the brand recognition even where they’re not the specialists.

Buy this if you want to know whether this matters for your category and you’re not ready to spend real money finding out. It’s usually a small add-on to a subscription you already have. The data is shallower than a dedicated tool, and for a first look that’s fine.

2. The dedicated trackers

Peec AI. Otterly. Scrunch. Athena. Evertune. Brandlight.

Purpose-built AI search analytics. Peec is the one we get asked about most: it tracks brand visibility across ChatGPT, Perplexity, Gemini, AI Mode, AI Overviews and Copilot, does competitive benchmarking, exports to Looker and an API, and says it serves 3,000+ brands and agencies. Published tiers from Starter to Enterprise, self-serve, free trial.

This category does measurement and recommendation. You act on what it tells you.

Buy this if you have a team who will actually act on the output — an in-house SEO or content lead with time and budget to make changes. These tools are good and they’re cheap relative to services. The failure mode is buying one and having nobody funded to do anything with it, which is the single most common waste I see in this market.

3. The platform moving into execution

Profound.

Worth being precise here, because the old shorthand about Profound is out of date. They raised a $96M Series C led by Lightspeed at a $1 billion valuation, bringing total funding past $155M. And they’re no longer a measurement product: Profound Agents are autonomous workers that generate content and optimize it, with more than 500 customers running content, monitoring, AEO and PR agents. In July they launched Aim, an always-on agent that prioritizes what to work on. They describe themselves as “the full stack marketing platform for the marketer of the future.”

So the tidy story where the dashboards measure and someone else does the work isn’t true anymore, and anyone still selling against Profound on that basis is selling against a 2025 version of them.

Buy this if you’re an enterprise marketing team that wants to cover a lot of surfaces with software and has the people to run it. It’s a platform, and platforms pay off for teams who actually operate them.

4. The optimization engines

ReFiBuy.

Scot Wingo’s company — he founded ChannelAdvisor, took it public in 2013, and it sold to private equity in 2022. ReFiBuy launched in 2025 and raised a $13.6M seed in May 2026 led by NewRoad Capital Partners. They coined the term ACO, Agentic Commerce Optimization.

Their Commerce Intelligence Engine runs six steps on your catalog: ingest, evaluate, enrich, sync, distribute, monitor. Read that list and you’ll notice it’s the closest thing on this page to what we do — it’s catalog-first, SKU-level, and it ends in monitoring rather than starting there. They sell to brands, retailers and agencies, and in July they added ACO Everywhere and an Amazon Alexa shopping program.

Calling this “just software” would be wrong, and it’s the mistake I’d have made a few months ago. Their own material describes human oversight in the loop — approvals, client guardrails, quality standards — and they sell ACO to agencies as a managed, repeatable capability. From talking to people at brands running pilots with them, there’s a service layer with named KPIs attached. So this is a hybrid: an engine, with people around it.

The distinction that does hold is what the people are for. In an engine-led model the team exists to run the system well across many accounts, and that scale is the point — it’s how you serve a 20,000-SKU catalog without 20,000 decisions being made by hand.

Buy this if you have a large catalog and want throughput with humans checking the work, rather than humans doing the work.

5. Where we fit

Vektor10. We’re an agency, not a platform. A senior team measures your shelf at the SKU level, does the optimization work, and reports what moved — the same people across all three, in your account, every week.

We’re also the smallest thing on this page by a wide margin. Profound is a billion-dollar company. ReFiBuy raised more money last spring than we’ve ever had. If you want the biggest vendor, that’s a completely defensible way to buy and it isn’t us.

What we’re actually built for: a brand between roughly $5M and $100M that doesn’t have a person to hand a dashboard to. The tools assume you have an operator; we’re the operator. Which is why we cost more than software and less than an enterprise platform plus the headcount to run it.

Since Profound and ReFiBuy both have people attached to accounts now, “we have humans” is not the distinction and I won’t pretend it is. The difference is what the humans are there to do. In an engine-led model they run the system across many accounts, and standardization is the point — it’s what makes it work at catalog scale. We carry few accounts and the same senior people make the calls on yours, which is better judgment per SKU and worse economics per SKU. That trade is genuinely better in one direction and genuinely worse in the other, depending on how many SKUs you have.

The table

	SEO suites	Trackers	Profound	ReFiBuy	Vektor10
Tracks AI visibility	Yes	Yes	Yes	Yes	Yes
SKU-level, product by product	No	Partial	Partial	Yes	Yes
Does the optimization work	No	No	Yes, via agents	Yes, via the engine	Yes, by hand
Custom prompt set per account	Limited	Yes	Yes	Yes	Yes
Humans attached to your account	No	No	Support	Yes — oversight and managed delivery	Yes — they do the work
Accountable to a named KPI	No	No	Varies	Yes, in managed engagements	Yes
Self-serve	Yes	Yes	Sales-led	Sales-led	No
Roughly what it costs	Add-on to a subscription	Software pricing	Enterprise software	Enterprise software	Agency retainer

How I’d actually decide

Does anyone on your team have time to act on a dashboard? If yes, buy a tracker and skip the services conversation entirely. You’ll spend a fraction of the money. If no, a tool will produce a monthly report that makes everyone feel informed and changes nothing.

How big is your catalog? Under a few hundred SKUs, humans can do the work well and the judgment matters more than the throughput. In the thousands, you want a system, and that points at an engine.

Do you want capability, or do you want the work taken off your desk? A platform gives your team capability. An agency or a managed engagement takes the work. Those are different purchases and people conflate them constantly — and note that the engines now sell both, so ask which one you’re actually signing.

Is your problem measurement or is it your catalog? If your feed is broken, no amount of measurement fixes it — you’ll just watch the number stay flat with better resolution.

The conflict all of us have

Everyone in groups three, four and five now measures and executes. That means every one of us reports on our own work. Including us.

I’d rather say that plainly than pretend we’ve solved it. What we do about it: your question set is fixed at the start and handed to you, so it can’t be tuned mid-engagement to make our reporting look better; you get the raw answers and citations, not just the summary; and we report the questions we lost. When we say we proved something in revenue, we mean modeled revenue from AI-sourced traffic that mostly arrives without a referrer — a real estimate with a real method, not a measured checkout.

If you’re already running an execution vendor and want someone unaffiliated to check whether their work moved anything, that’s a fair thing to want, and it’s a job we’ll take. It’s also a job someone other than us can do.

Where we’re the wrong answer

Under about $5M in revenue. If you sell B2B services rather than products — our whole method is catalog-shaped. If what you want is the lowest-cost way to watch the number, buy a tracker. If you want software your own team operates at scale, buy the platform. And if you’ve got 20,000 SKUs and need throughput more than judgment, an engine will beat us on that specific job.

We audited our own agency last week and published the result, which was that ChatGPT recommends us in zero of 86 buyer questions. So take the above as a map from someone with an obvious stake and a demonstrated willingness to publish the number when it’s bad.

---

Sources: each company’s own site and public funding announcements, checked 6 August 2026. Firm-mention percentages come from our own study — 16 discovery questions, five repeats each, ChatGPT™ only, 86 answers, probed 5–6 August 2026. Competitors’ offerings change fast; where this is out of date, theirs is the accurate version.