Tying 10% of an ETF’s AUM to Marketing Without Seeing the Buyers

Every ETF issuer eventually hears some version of the same question from a founder, a CEO, or a board: is the marketing actually moving flows?

It’s a fair question. Margins are thin, there’s no 12b-1 revenue funding distribution, and every marketing dollar competes with everything else the firm needs. The problem is that the honest answer is usually “we don’t know.” Web sessions, email clicks, and form fills are easy to report. Flows are not, because the issuer never sees the buyer.

Over eight months with one actively managed ETF client, we tied new money equal to roughly 10% of the fund’s AUM to specific marketing efforts. We never saw a holder record. Here’s how we did it, where the evidence is strong, where it isn’t, and a framework any issuer can run with their own capital markets data.

Where the visibility breaks

You know the mechanics, but it’s worth naming exactly where visibility is lost, because that determines what a workable method looks like.

End investors transact in the secondary market, whether they’re self-directed retail buyers or RIAs executing blocks across client accounts. The other side of the trade is almost always a market maker or liquidity provider filling the order from inventory. Even large institutional orders typically route through block desks or RFQ platforms to liquidity providers, not to the fund.

The issuer only enters the picture in the primary market, when an authorized participant creates or redeems shares in creation units. That often happens days or weeks after the underlying purchases, once a market maker needs to replenish inventory. Meanwhile, positions sit in street name and omnibus accounts, so even a holdings snapshot rarely gets you to the advisor who made the decision.

13F filings fill in only part of the picture. Managers with $100 million or more in U.S. equities and ETFs must disclose their holdings each quarter. But a 13F shows only long positions as of the last day of the quarter, and it can be filed up to 45 days later. By the time you see it, the trade may be four months old, which is too late to connect to any campaign. And RIAs under the $100 million threshold, which includes many of the advisors a boutique issuer is courting, never file at all.

That leaves one high-frequency, issuer-side signal of demand: the change in shares outstanding. Everything in this method is built on it.

The approach 

We didn’t try to identify buyers. We identified a few creations we could connect to marketing with high confidence. We used those to calibrate a look-back window, then tested every other creation against that window to see which efforts kept showing up.

Step 1: Treat creations as the conversion event

We pulled the fund’s daily shares outstanding, roughly 250 trading days, and isolated every creation after our campaigns went live. There were 11 creation days.

Each creation carries two pieces of information. The date anchors the look-back. The size and cadence suggest who was behind it. Advisor allocations, placed across many client accounts at once, tend to arrive as large, discrete creations. Self-directed demand tends to show up as smaller creations on a steadier cadence. That’s an inference, not an identification, but it proved useful.

Two cautions belong here. First, creations lag the purchases behind them, since market makers fill from inventory and create only when they need to replenish. Second, not every creation reflects end-investor demand. Market makers build inventory too, and one creation in our data was partly reversed weeks later. Your capital markets team or lead market maker is the best check on both.

Step 2: Calibrate with your highest-confidence flows

The clearest case came in the spring. The fund had gone roughly ten weeks without a creation. Then, over about two weeks:

• Our first large advisor-focused ebook cold email campaign launched
• Retargeting impressions rose 48% the following week as email traffic reached the site.
• We rebuilt the advisor landing page four days before the first creation, and the bounce rate fell from 100% to 17%.
• The day before, an advisor submitted the bottom-of-funnel “ready to invest” form, one of only two advisor submissions all year.

Over the next three days, two creations added new shares equal to about 3.5% of AUM, the largest stretch of new money in the period. The client credited it to our work. A month later, another creation worth nearly 2% of AUM was also credited, though no buyer was named.

Across the credited creations, our activity landed between one day and three weeks beforehand. We set a four-week look-back window for everything else, with the extra week as a buffer for the lag between purchases and creations.

Before running the analysis, our team’s working hypothesis was that advisor email and Google Ads were doing the heavy lifting. The data supported it. That isn’t proof on its own, but it’s a useful check against reading patterns into noise.

The key concession is calibrating on high-confidence flows rather than certain ones. You’ll rarely get certainty. What you need is a handful of cases strong enough to defend to your CEO. 

Step 3: Test every creation against the window

For each of the 11 creations, we mapped everything live in the prior four weeks: advisor email, landing page changes, Google Ads changes, retargeting reach, HubSpot form submissions, to name a few.

What makes this possible is unglamorous: a dated log of every marketing change. Without dates, there’s nothing to line up.

Step 4: Look for what repeats 

A single creation following a campaign proves nothing. Rebalances, model changes, and market moves drive plenty of flow. Repetition is what matters.

• A material Google Ads improvement or addition preceded 7 of the 11 creations.
• Advisor email was in market before 6 of the 11.
• A retargeting peak preceded 6 of the 11.
• No creation occurred without at least one of the three active.
  

The Google Ads results deserve a note for issuers who are skeptical of paid search for an ETF. Over two months, cost per conversion fell roughly 85% and monthly conversions almost tripled. That came from a move to phrase match, a captcha upgrade on the landing pages and new competitor and bottom-of-funnel retargeting campaigns. Each step-up was followed by creations within four weeks. After the June overhaul, the shape of demand changed. Instead of occasional large blocks, smaller creations arrived on a steadier cadence, which fits self-directed investors buying through brokerage platforms.

Step 5: Turn the pattern into the plan 

For most of the summer, advisor email was quiet, apart from a nurture send in August. Creations continued, but they were smaller and steadier, and the large blocks that pointed to advisor allocations stopped. When advisor outreach resumed, the fund saw its first creation in seven weeks within days.

That’s about as close to a control group as real-world distribution gets. It also answers the objection we hear most from issuers: “Advisors buy on platforms, so how does marketing help?” In this data, the large creations consistently followed advisor-facing marketing.

The plan now reflects that:

• Advisor email in market every month, with no gaps. We’re also moving to dedicated cold email infrastructure so sending volume and deliverability don’t limit reach.
• Retargeting always on. It’s inexpensive, every new reach peak came within weeks of a creation, and we’re expanding it to LinkedIn and other platforms.
• A standing roadmap of Google Ads tests. Google Ads improvements preceded more creations than any other effort, so the account is managed as an ongoing program, not a one-time build.
  

What we didn’t count 

Any figure that ties marketing to flows will be scrutinized. It should be. Here’s what we excluded.

First, every dollar is tiered by confidence, and we lead with the client-credited share, nearly 5% of AUM, with roughly 10% as the ceiling.

Second, we excluded flows that clearly weren’t ours. A creation occurred the week before any campaign launched, which signals baseline demand we can’t claim. We also excluded market maker inventory that was later redeemed, since it wasn’t end-investor demand.

Third, we reported gross creations and redemptions separately rather than netting them. We also didn’t assume redemptions came from holders who predated our work.

A smaller number you can defend in front of a board beats a larger one you can’t.

An attribution checklist for ETF issuers 

Any figure that ties marketing to flows will be scrutinized. It should be. Here’s what we excluded.

First, every dollar is tiered by confidence, and we lead with the client-credited share, nearly 5% of AUM, with roughly 10% as the ceiling.

Second, we excluded flows that clearly weren’t ours. A creation occurred the week before any campaign launched, which signals baseline demand we can’t claim. We also excluded market maker inventory that was later redeemed, since it wasn’t end-investor demand.

Third, we reported gross creations and redemptions separately rather than netting them. We also didn’t assume redemptions came from holders who predated our work.

A smaller number you can defend in front of a board beats a larger one you can’t.

None of this requires new tools or vendor data. It requires using the data you already have in the right order. Work through this list before your next board meeting, and you’ll walk in with flows tied to specific efforts instead of web traffic.

Signal

• Pull daily shares outstanding and isolate every creation and redemption
• Review creations with capital markets or your lead market maker to separate inventory builds from end-investor demand
• Check 13F filings each quarter for managers you can match to marketing activity

Calibration

• Identify a few flows you can tie to marketing with high confidence: wholesaler intel, advisor conversations, bottom-of-funnel form fills, 13F matches
• Measure the lag between marketing activity and creation for each, and set a look-back window with a buffer
• Record your team’s hypothesis before analyzing the rest

Alignment

• Maintain a dated log of every marketing change
• Map everything live in the window before each creation
• Read creation size and cadence as a rough indicator of advisor versus self-directed demand

Pattern

• Count how often each effort precedes a creation
• Note which combinations precede your largest and most confident flows
• Study periods when a channel went dark and how creations changed

Reporting

• Tier every flow by confidence
• Exclude pre-launch baseline demand and market maker inventory
• Report gross creations and redemptions separately
• Lead with your most defensible number

The bottom line

Most issuers we talk to have been told their marketing can’t be tied to flows, or have run campaigns that couldn’t show it. Neither has to be the end of the conversation. The holder record isn’t coming. But with disciplined data, a few high-confidence anchors, and honest tiering, you can show leadership which efforts precede new money and where the next dollar should go.

If you’re working to prove that marketing moves AUM, we’d be glad to walk through how this would work with your fund’s data.