Book a call
( Agency Brief )Issue 73 min read

The AI dividend shows up in the P&L, and most agencies can't measure theirs

Holding-company Q3 results turned the AI and consolidation story into reported margin numbers and pitch-free wins. In the same week, agentic AI passed 50% adoption in marketing while two-thirds of organisations still had no way to prove its return.

Published
Last reviewed
By

( 00 )The week in three lines

  1. 01

    Omnicom grew 7.2% organically and said it is on track for $900m of IPG cost synergies.

  2. 02

    Publicis named winning without a pitch as a strategy it wants to repeat.

  3. 03

    51% of marketing organisations use AI agents; 65% don't measure the return.

( 01 )

The AI margin story became a reported number

What changedOn 22 October Omnicom reported Q3 revenue of $6.42bn, up 59% year on year on the IPG merger, with 7.2% organic growth and adjusted EPS up 22% (Omnicom, StockStory). CEO John Wren called the integration "nearly complete" and the group "on track for $900m in cost synergies". Nine days earlier Publicis told investors AI had widened its margin lead over peers from about 270 basis points in 2022 to close to 400 in 2025 (Investing.com).

Why it mattersThis was the quarter "AI will improve margins" became a reported figure. Holding companies now have a growth and margin story clients can read in a press release, and it resets the baseline every independent is measured against in procurement. Independents need their own version of that number: hours removed, cycle times cut, margin reclaimed.

( 02 )

Winning without a pitch is now a stated strategy

What changedOn its 13 October call, Publicis said it had landed three AI-led transformation wins outside traditional pitches this year, and that more than 90% of local wins in the first half went to the three global groups (Investing.com). Omnicom's Wren said on 22 October that clients are "becoming more focused on selecting a single provider". It is the same route that moved Microsoft's media account to Publicis without an open pitch in April (Campaign).

Why it mattersThe front door to new business is moving from the RFP to the paid trial and the partnership. "Help us win more pitches" is becoming the wrong brief. The value is in proof-of-value entry points that make a formal pitch unnecessary, and in protecting existing clients from incumbents who get the first look.

( 03 )

Agentic AI passed 50% adoption, and few can prove the return

What changedA roundup compiled on 22 October by Acalytica put AI agent use at 51% of marketing organisations, up 155% year on year, concentrated in reporting and analytics (72%), campaign briefs (56%), content (46%) and media planning (35%). It also found 65% don't benchmark AI spend against defined metrics, and 61% of agencies treat AI as a cost rather than something they sell.

Why it mattersAdoption no longer sets anyone apart. Measurement does. Agencies that can prove and price AI-driven outcomes will separate from those quietly absorbing the savings, and the gap between using AI and showing what it returns is where most agencies sit today.

( 04 )

Headcount is coming apart from revenue

What changedThe same Acalytica data notes Omnicom and IPG cut around 8,200 roles before their merger while expanding margins, and Publicis is targeting a 25% reduction in manual tasks. Analysis from Piscari argues AI has broken the link between time and value: work once priced at 100 hours now takes about 40, and it expects advertisers to move towards output and outcome-linked pay by 2028.

Why it mattersAgencies still selling hours are visibly on the wrong side of the curve, and procurement can see it. Repricing the service lines most exposed to AI before the next procurement round is far easier than doing it under pressure.

( 05 )

Salesforce buys Braze as the marketing stack consolidates

What changedOn 20 October Salesforce agreed to buy customer engagement platform Braze for about $5.7bn (Bloomberg, Financier Worldwide), folding another messaging and data layer into a single AI marketing suite.

Why it mattersThe platforms agencies build AI services on are concentrating into fewer, deeper suites. That creates integration work, and it also creates dependency risk, because a suite's own agents can commoditise services built on top of it. Platform-neutral orchestration and change management are the layer vendors can't sell.

Which service lines to reprice first, and how to show an agency's own AI return, depends on its clients and its work. That is the conversation I have with agency owners, and it is what the call is for.

( 06 )Get in touch

Want this read on your own agency?

Book a call Start with a 30-minute call about your agency’s new business. No pitch deck. Or take the 4-minute health check →