(
)
***********************
Happy Sunday, Everyone!
***********************
I hope you’re taking some well-deserved time to step away from
the inbox, close the ad account, recharge and enjoy the last rays
of summer before the NFL season (Go Birds), Q4 planning and (yes,
really) 2026 arrive.
This summer has been one of the busiest in recent memory for me –
travel, events, new business, client meetings, hiring…it’s felt
like a non-stop, perpetual whirlwind. To bring some order to my
chaotic calendar, one of the things I’ve started to do is bring
my note-taker (shoutout to Otter.ai (
)) to every meeting AND create a process to surface insights from
those meetings.
As I’ve been going through those notes, one thing stood out: the
number of times the following question has occurred in a meeting:
“What’s our AI strategy?”
According to my (well, Otter’s) notes, it’s come up at least 16
times in the last 3 months - and that’s just virtual meetings.
It’s more than a question; it’s THE question that is
front-and-center for every CMO, CEO, investor, owner, operator
and/or marketing leader with whom I’ve had a conversation.
The problem is that it’s the wrong question.
Let’s take a step back: for the past ~2.5 years, the pace of AI
evolution in the marketing sector has been both unrelenting and
overwhelming. It feels like there’s something new - a model, an
application, a use case, a capability, a tool, a workflow -
that’s the next big thing, every single week. The sheer volume of
unsolicited messages I get from people building thinly-veiled GPT
wrappers promising staggering improvements is insane. We've
finally reached an inflection point where the noise has become so
deafening that executives across-the-board are compelled to
respond.
Just in the last few months, I’ve watched this scramble play out
seven ways to Sunday: leadership teams approving AI “pilots” just
to be seen as “doing AI.” Multiple 8/9 figure brands that have
junior employees with individual GPT licenses spinning up AI
content…with no strategy, POV or structure. And agencies are
certainly not immune – just last week, I was passed an agency
deck that had “AI” in it 19 (!!!!) times – and the worst part was
that at least half of them made no sense as a use case/workflow,
let alone something that will materially improve a brand’s P&L.
We’re in the golden age of AI theater.
Everyone wants to be seen “doing” AI (less they fall behind),
which has (rather predictably) led to organizations doing crazy
stuff – from random announcements/company-wide emails, to
assembling “AI teams” to licensing a dozen AI tools for
who-knows-what. And we’re now at the point where all that theater
has bubbled up to the leadership table, and the people that sit
around it are asking: “what’s our AI strategy?” Since we’ve all
been too busy doing AI theater, that question keeps getting
passed down until someone provides an answer that sounds both
smart and plausible.
But here’s the truth: there are very few smart answers to
incoherent questions.
AI is a capability. It’s a productivity enhancer. It’s a force
multiplier. It’s not something that sits off in a little silo
attached to your organization; it’s something with the potential
to help every constituent part of your organization perform at a
higher level.
You don’t need an “AI strategy” any more than you need an
“electricity strategy” or an “Excel strategy” – what you need is
a business strategy. AI only matters insofar as it improves your
ability to execute that business strategy more effectively and/or
efficiently.
That’s what this week’s issue is about.
Let’s start with the framing. Most companies are asking the wrong
question (“what’s our AI strategy”) – so let’s reframe it: AI is
a general-purpose technology that can either (1) reduce costs,
(2) improve choices or (3) enable entirely new capabilities.
That’s it. Every meaningful use case ladders up to one of those
three.
That changes the question to: “How do we create or compound
economic advantage using AI?”
When you approach it that way, AI initiatives start to look a lot
more like capital allocation decisions and not pet projects
designed to garner applause on business Tinder (also known as
LinkedIn). The distinction may sound small, but it matters a
great deal – because if you’re not making investment/allocation
decisions with the P&L in mind, you’re just conducting an
exceedingly expensive science fair project.
With the proper framing in place, the next question is: how do we
make AI investments that create/compound those economic
advantages? Simple: treat them like a portfolio:
* Efficiency Bets (Low Risk, Fast Payback): They’re low-risk,
capped-upside bets that reduce or eliminate the busywork/tedious
tasks that comprise 10% - 30% of your team’s time. Examples
include: workflow automations, mundane task automations, simple
customer support/success processes, research augmentation,
meeting/deliverable summarization tools. At their core, each of
these improves organizational efficiency, reduces knowledge loss
and/or improves the quality/quantity of the service you provide.
None of them are going to transform your organization – but
they’ll quickly fade into the background, quietly becoming
central to your organization in much the same way that fast WiFi
or cloud storage are critical.
* Effectiveness Bets (Medium Risk, High Impact): Every team has a
set of things they *wish* they had time to accomplish.
Effectiveness bets are - at their core - the application of AI to
the things on that list. These things focus on improving decision
quality, customer experience and/or capability
performance/execution. Think: improving your
inventory/performance forecasting, creative ideation/iteration,
developing real-time customer insights, post-click optimization,
etc. Nothing here is going to replace your marketing team, but
done well, it can make that team exponentially more effective and
efficient.
* Expansion Bets (High Risk, Asymmetric Upside): Finally, there’s
the expansion bets. This is where new categories, new AI-native
features and breakthrough products live. I would include anything
from agentic solutions to fully-automated chatbots to automated
ad managers here – these are the things we often hear hyped on
LinkedIn, but (candidly) are rarely ready for prime time. If
you’re going to make bets here, it is absolutely essential to
have both controls and continue/kill criteria in place.
This framework mirrors portfolio theory in finance: core
holdings, growth plays, and speculative bets. If this all sounds
familiar after the last two issues, it should. The principles of
compounding, risk-adjusted return and capital reallocation apply
here, too.
-----------------------------------------------------------------
The Big Takeaway: AI Theater Happens When You Confuse Tools With
Outcomes
-----------------------------------------------------------------
Flashy tools with no connection to operating leverage? That’s
theater. It’s a show that distracts you (and your team) from what
really matters.
Instead of a fancy product demo, your goal should be to craft a
compelling narrative that connects what you’re doing to how your
organization makes money. It should be straightforward, credible
and measurable. For example:
“Our primary focus with AI is improving our operating leverage.
We’ve identified 14 routine tasks that together account for 21%
of our team’s time, and we’re rolling out automated workflows for
each. We expect this to reduce the time spent on those tasks by
80%, with little-to-no impact on the quality of
deliverables/outcomes. Our secondary focus is reinvesting the
time saved above, alongside AI, into three core areas of
opportunity: (1) improving our post-click experience; (2)
creating real-time audience insights and (3) improving creative
efficiency. Based on conservative forecasts, we believe this will
allow us to reduce our customer acquisition costs by 13% in Q4
and 17% in Q1 2026. Finally, we’re also beginning several agentic
pilots that could transform the shopper experience. These are
small, strategic bets that could have a transformative impact on
the business if they hit.
That narrative is infinitely more credible than 99% of what I
hear. That’s what we should all be working toward. Bridging this
chasm requires us to re-focus what we’re doing around what
actually matters to their organization.
The reality is that most marketers aren’t lacking in intent -
they want to use AI more. The gap I see - overwhelmingly - is
infrastructure. They don’t have the tools, systems or support to
go from “I have this GPT license and a way-too-long to-do list”
to “Here’s how I can automate these 19 tasks so I can get more
done.”
Nothing that follows in this section is sexy. It’s not going to
garner the “oohs” and “ahhs” from the LinkedIn crowd. What it
will do is improve productivity and expand margins, while
allowing your team to spend more time on higher-leverage
activities.
AI tools are accelerants: integrate them in a great process, and
you’ll reap exponentially higher returns; apply them to a broken
system, and all they’ll do is multiply the chaos.
Let’s get to it:
1. Get Your Internal Data “AI-Ready”
------------------------------------
Data is the raw material that powers any AI initiative. As the
old saying goes, “garbage in, garbage out.” Most companies still
don’t know what data they have, who owns it or whether it can be
trusted. These organizations are stuck in the 2010s view of
data-as-a-warehouse, vs today’s data-as-an-operating-system.
This is not a fun shift to make. The first step is taking an
inventory of what data you have. This should include everything:
internal docs. User guides. Technical specs/docs. Processes.
Notion boards. Templates. Checklists. Content. All of it.
Odds are, you have reams of data you haven’t used in years (if
not decades) – some of which might be quite valuable, and a great
deal of which might no longer be relevant. If you just set an AI
loose on all of it, the probability is quite high that it’s going
to incorporate some of the wrong stuff in the output. Documenting
everything allows you to exclude the stuff that’s no longer
relevant, which (in turn) improves the quality of the output.
Less garbage in = more good stuff out.
This doesn’t stop there. Once you have the list, the next step is
to build taxonomies and ontologies, standardize naming
conventions, tag entities and map relationships (this process and
this process are both part of this service, etc.). The goal of
this work is to help the system understand the relationships
between all of the content/documents you’ve cataloged – so the AI
is able to grasp that Service 1 includes X, Y & Z, and is
positioned in this way to this audience and that way to that
audience.
If this sounds boring and tedious, it’s because it is. But this
is the boring, tedious work that compounds into a massive
advantage over time.
The second benefit of this is provenance. When you can prove
where your data came from, how it is secured and who touched it,
you don’t simply minimize risk - you create confidence and build
trust. That, in turn, helps you win more business.
2. Brand Taste - Clarify Your Brand, Voice & Tone
-------------------------------------------------
AI has effectively reduced the cost of creation to near-zero.
Anyone with a GPT license and some free time can create a
near-unlimited number of image, video and/or text assets. The
impact of that is that volume is no longer the differentiator;
that honor now belongs to taste (
).
Most brands have not codified their brand, voice or point of
view. That’s a mistake. Without guidelines, the creative outputs
(whether text, images, audio or video) are AI slop. Generic,
boring, undifferentiated sludge.
The smart move is to develop a “brand guide” that includes three
core components:
* The Non-Negotiable Rules - these are the “must follow” elements
for your brand - the things you will NEVER compromise. Your brand
promise, your desired emotional connection with your audience,
your brand values.
* The Guidelines - these are your preferences, but they’re not
written in stone. This might be the types of images you feature,
the kinds of comments/reviews you amplify, the benefits you
highlight.
* Examples/Illustrations - AI models tend to respond quite
favorably to “do this / not that” structures - so illustrate each
of the above.
The scarce resource is not content. It is attention. The question
isn’t whether you can produce at scale - anyone with a GPT
license can. The question is whether people will stop, read and
care – and that’s a question of taste.
3. Workflow Re-architecture
---------------------------
I’ve worked with hundreds of organizations, from solo ventures to
billion-dollar companies. Across every one, there are 20% of the
tasks that take 80% of the time. The exact nature of those tasks
vary, but the rule always holds.
I firmly believe the biggest gains from AI won’t come from
shaving a few minutes off a task. That kind of incremental
efficiency might garner some praise at a weekly standup, but it
doesn’t change how your business runs.
The real unlock is rethinking the task entirely. Not “how can we
do this faster?” but, “Why do we do it this way in the first
place?”
Start by mapping your top 10 recurring workflows based on volume
Ă— value. Go through your daily to-do list / hours log (if you
have one) and identify the routine tasks that take the most time
(and if you have more junior people doing the task, ask them…then
actually listen). Once identified, rebuild those workflows with
AI integrated into the system, not just sprinkled on top.
Your goal should not be to automate people away, but rather to
design processes where:
* Manual tasks are fully automated (think: copy-and-pasting data
from X to Y)
* Handoffs (i.e. marketing drafts copy, design lays it out, dev
implements) are eliminated
* Repetitive tasks (like tagging, summarizing or formatting) are
fully automated
* Human judgment is reserved for edge cases, escalation, or
strategic input
This is the difference between bolting AI onto a broken process
and using it to reduce complexity. When done right, you’re not
just speeding things up. You’re reducing the total number of
steps, decisions and dependencies – that’s what frees up your
people to do more of the work that matters.
4. Real Governance
------------------
Let’s start with this disclaimer: governance isn’t sexy. In the
move-fast-and-break-things AI world, talk of governance is
usually about as well-received as a Chris Rock skit in church.
But….no one wants to be the reason your chatbot hallucinated a
product claim or your “personalized” emails accidentally included
sensitive customer data.
That results in most companies defaulting to one of two extremes
when it comes to AI governance:
* Wild West: Everyone’s using generative tools in their
workflows, but no one knows what data is exposed, what goes into
outputs, how data is being used or what risks are accumulating
behind the scenes.
* Fortress Mode: Legal locks everything down. Teams are forbidden
from using AI tools for even basic, low-risk tasks. Innovation &
experimentation die. The company ends up watching competitors
ship, learn and iterate while they’re still drafting their usage
policy.
Both approaches fail. The first opens you up to real risk +
potential legal exposure. The second guarantees you fall behind.
The alternative - and what high-functioning, forward-looking
companies are doing - is a happy medium. That tends to look like
a lightweight governance layer that enables innovation while
still managing risk.
* Define “safe to try” vs. “needs review.” Make the criteria
explicit. If a use case uses non-sensitive inputs, generates
draft copy and isn’t customer-facing, use it! If it references
regulatory claims, pricing data or customer data, send it for
review (red light). For items that fall in between those
extremes, assign a yellow light, where someone more senior has to
weigh in before deciding on a course of action.
* Document decisions AND procedures – one of the easiest lifts
for any organization is to have a living Notion (or similar)
where decisions, procedures and prompts are stored. This allows
everyone in the company to have full transparency into what’s
being done, all while saving a ton of time (since you won’t have
to draft internal docs and your people won’t have to try to
recreate a prompt that already works).
* Make compliance a feature, not a blocker. The best guardrails
work like lane assist: they help people move faster because they
know where the edges are. I know that compliance has gotten a bad
rap over the years (and most of it is deserved), but there’s also
massive value in ensuring that everything you create/put
out/publish doesn’t pose an existential risk to your company.
Done right, governance isn’t red tape. It’s what allows teams to
test, learn and create with confidence. We’re all operating in
the digital equivalent of the wild west. Hallucinations,
copyright exposure and regulatory missteps happen every day -
what is needed is a way to minimize the risk while maintaining
the upside, which is exactly what this provides.
5. Talent & Training
--------------------
This may be obvious, but paying for a bunch of AI tools doesn't
magically make your people more effective, competent or
productive.
Training your people on how to use those AI tools does.
Your goal should be to make two fundamental changes: (1) shift
from individual instances to scalable use cases and (2) move from
hiring for technical excellence to judgement.
Let’s start with (1). Just the other day, I was talking with one
of our digital team members who was “vibe coding” a JavaScript
setup for landing pages. He built conditional redirects off a
radio button selection, sending users to different thank-you
pages based on their response to a form question. The idea was to
provide both a tailored experience AND better analytics data.
It’s a clever use case. The problem was that it was built for a
single client, when we have at least 5 that would benefit from it
being implemented. We solved that by adding this to our library
of effective prompts, then sharing it with the entire team – so
other account teams who could use it were able to access and
implement it in minutes. And as each of those client teams
implemented the workflow, they uncovered several weird
edge-cases, which we then documented and used to further refine
it.
There’s nothing magical about this – it’s just training. The
difference is that this training transforms one person’s clever
experiment into a scalable capability the entire agency can use,
and that all of our clients can benefit from.
That’s the difference between tinkering and leverage. Individual
hacks live and die inside one client account or team member’s
head. Scalable use cases are documented, shared, refined and
eventually become part of the “operating system”. When you build
that kind of library, every new engagement/project starts a step
ahead of the last one.
This brings me to the second shift (and this is one that’s going
to have HR furious): stop hiring purely for technical excellence.
Tools are closing that gap every day. What you cannot automate is
good judgment. Knowing when an output is good enough. Spotting
when an edge case will break something important. Choosing which
ideas to run with and which ones to put back in the box.
AI makes technical execution cheaper. That, in turn, pushes the
premium to synthesis, taste & judgement. The teams that hire
people with those skills will quickly find themselves at a
massive advantage, because their people will be better equipped
to manage increasingly-more-sophisticated technical tools.
6. Measure The Right Thing
--------------------------
Let’s get clear about one thing: the goal of any/all of this is
NOT “AI adoption” - it’s better business economics. That’s the
whole reason we’re doing this, so we need a
measurement/accountability system that aligns with that goal. I’d
recommend you focus on three core areas:
* Operating Leverage: are our investments in AI tools allowing us
to realize real cost and/or time savings on our operations? Are
we able to automate/expedite routine tasks, thereby freeing up
our team to do higher leverage work? Are these tools reducing the
time required to produce key deliverables (i.e. monthly reports)?
* Effectiveness: Doing something faster or cheaper is only good
IF the quality of the output is as good as (or, ideally, better
than) what was produced before. If you save 3 hours by automating
a client report (good!) but the quality of the automated report
is so poor/disjointed that your team now needs to spend 4 hours
explaining to the client/stakeholder the content and what it
means, you haven’t actually accomplished a damn thing. The same
holds true for AI-generated landers or creatives: they should be
as effective, if not more effective, than what you had before.
The goal must be not only to improve efficiency, but to improve
effectiveness.
* Visibility: This is the one most people forget. How often is
your brand showing up in AI answers? How much traffic is coming
from assistants like ChatGPT or Perplexity? Is the brand visible
in the new discovery surfaces like AI overviews or AI mode, where
customers are starting their journey? This isn’t just about
working smarter inside the walls; it’s about making sure the
brand isn’t invisible as these tools re-write how people find
things.
At the end of the day, the only thing that matters is this: is
your organization more profitable, more effective and more
discoverable than you are today? That’s the thing we’re all
trying to figure out how to do, and the best way to accomplish it
is by having a scorecard that maps to it.
------------------------------------
6 Tactical Moves You Can Execute Now
------------------------------------
If the preceding section is theoretical, this one is exceedingly
practical. The reality is that AI is here to stay, and there are
plenty of ways you can integrate it into your day-to-day right
now.
Here are a few things that we are doing and what I think
marketers should be doing yesterday:
1. Optimize for AI Overviews + Assistants
-----------------------------------------
Search isn’t dead. It’s evolving.
SparkToro’s latest research (
) shows that while just 20% of Americans use AI tools more than
10 times a month, that usage is enough to make a dent,
particularly among high-intent, early-adopter audiences. But
here’s the kicker: traditional search traffic hasn’t meaningfully
declined. Which means Ai tools are NOT a channel replacement;
they’re a visibility expansion opportunity (something I’ve said
for years - AI will make people search more, not less):
The takeaway: it’s not “either/or.” It’s “yes/and.” AI Overviews
are carving out top-of-funnel real estate in addition to classic
rankings. And, for now, the competition for that space is light…
which makes this a rare opportunity.
Fortunately (and despite all the AI bro comments to the
contrary), appearing in AI Overviews is not a super-secret
science; it’s just good, old-fashioned SEO. If you want to appear
in more AI Overviews, here’s the quick playbook:
* Identify which searches are impacted by AI overviews (yes, you
will have to do some searching!). For the ones that are, note
which sites are included (yes, you’ll have to do some clicking).
* Document the commonalities + the weaknesses in the AI Overview
– if an answer has wrong/outdated/partially correct/incomplete
information, that’s an opportunity!
* Once you’ve identified the opportunities, build pages designed
for conversational intent: “Best X for Y in Z context”
* Create layered content: 1-sentence takeaways, 1-paragraph
summaries, and full deep dives, all on the same page
* Ensure your content actually communicates real, legitimate,
unique value – if you’re just saying the same things as everyone
else, there’s little reason for any engine to include your
content in their outputs.
* Structure pages cleanly: H1–H3s, FAQs, comparison tables,
glossary sections
* Add schema (FAQ/Product/How-To)
* Fix page speed + crawl issues
* Publish content with real bios, citations and external
credibility signals
Remember, you’re writing for people, not keywords or language
models. If you create content that adds value to your audience
and is easily intelligible to the LLM, you’ll quickly find
yourself included.
2. Treat AI Assistants as Channels
----------------------------------
Most CMOs I’ve spoken to are still treating referrals from AI
tools (Perplexity, Claude, ChatGPT) as a curiosity, not a true
channel. That’s a massive mistake. We’ve had several clients that
have gone from 5% in the last 90 days. That may not sound huge
(and, even at 5%, it’s the 8th-largest source of traffic), but
the rate of increase is shocking.
This is the SEO goldrush all over again. The numbers are/will be
small at first, until they’re too overwhelming to ignore. The
solution is to begin tracking referrals from ChatGPT, Perplexity,
Claude, et al now. Spend time understanding the differences in
session types/quality between AI referrals and standard search
(this is a great use case for Heatmap.com (
) or Microsoft Clarity).
Once you have clarity on how much you’re appearing now, the next
step is to increase your visibility. Publish explainers and
glossaries in crawlable, indexable formats. Contribute to public
knowledge graphs where appropriate.
When clients ask if this is worth the effort, my answer has been:
“If your target audience is asking questions (doesn’t matter if
it’s on Reddit or to ChatGPT), you want your content to be the
answer.”
That starts by treating these assistants like the channels they
are.
3. Own the Post-Click Experience
--------------------------------
So much of the conversation around AI overviews + their impact on
search right now is on visibility (who is showing up, how much
traffic is it generating, etc.) – but none of that matters if
your post-click experience doesn’t convert. The reality is that
AIOs tend to drive longer, more conversational queries (something
we’ve known for quite some time) – which means that expectations
of your post-click experience are higher.
A generic “contact us” page isn’t going to cut it for a user who
just typed “best CRM for AdTech B2B SaaS companies in US” – that
user expects the lander to match the intent of their query.
The reality is that most brands are still playing catch-up in
this department.
Align your landing experiences with the shape and structure of
long-form, conversational queries (filter your Google Search
Console for queries with >8 words). These queries aren’t
transactional in the traditional sense; they’re layered with
context and constraints.
That means building modular, targeted landing pages that speak to
real-world situations: “best platform for X use case,” “solutions
under Y budget,” “how this compares to Z.” Your goal for each of
these pages should not just be to sell the service/product - it
should be to help your visitor make a smart, confident decision.
This approach creates space for compounding returns. Why? Because
once you find a variant that lifts conversion or AOV, whether
it’s a headline, a section order, a pricing frame or an
objection-handling block that improvement holds across every
future visitor who fits that same profile.
The process for this is simple (and can be easily augmented with
AI):
* Build modular, intent-matched landers that speak to specific
use cases, constraints, and objections
* Run A/B tests that focus on downstream metrics like CVR, AOV,
and lead quality
* Roll learnings back into your core templates so each new page
gets better by default
Then repeat. Iterate. Stack improvements.
This is how post-click becomes a flywheel. Not just a place to
recover wasted traffic but a surface where smarter decisions and
better experiences multiply over time.
4. Ask AI About Your Brand, Then Fix The Gaps
---------------------------------------------
Have you asked ChatGPT what it thinks about your company? Or
Perplexity to describe your product/service? Or Gemini to tell
you what sets your business apart from your competitors? If not,
you should. Chances are, the answers are incomplete, outdated, or
just plain wrong.
That’s a brand risk hiding in plain sight.
The defensive move is to seed the internet with clarifying
content. Publish canonical explainers. Create source-of-truth
pages that models can pull from. For regulated industries, sign
and date them. Give the models something better to reference.
If you don’t, someone else will.
There’s an offensive play here, too. Most of your competitors
haven’t done any of this, which means their AI footprint is thin,
stale or non-existent. Your goal should be to treat every
unanswered question about your competitors as an opportunity to
frame the conversation in your favor. If an assistant generates a
side-by-side comparison, the company with fresh explainers and
recently updated buyer’s guides will look credible, while the one
with a two-year-old press release will look asleep.
You should absolutely defend your own presence. And you should
fill the gaps your competitors leave. Any information vacuum will
be filled; the question is whether that’s by you or by someone
else.
5. Increase Content Velocity Without Going Generic
--------------------------------------------------
AI can (and will) generate endless creative. Without structure,
judgment and taste, most of it will be AI slop. If you want to
avoid publishing reams of generic garbage, the key is to revise
your workflow:
Step #1: Use AI to help uncover patterns in how your audience is
searching for information/solutions (customer insights)
Step #2: Based on that, have AI draft content that is likely to
appeal to your audience as they’re searching for that information
Step #3: Feed all of the content generated through the brand
guardrails (discussed above) AND a taste filter
Step #4: Have a real person with good judgment review, edit and
polish everything that comes out, so everything that gets
published is both valuable and relevant.
The fastest way to scale this is by building a library of proven
content structures. Formats like:
* Myth vs. Fact
* Buyer’s Guides
* Teardowns
* What to Expect
* Comparison Tables
* Checklists
These formats work because they speak to how people actually
evaluate solutions. They break down complexity. They frame
decisions. They let you communicate authoritatively without being
overly sales-y. Once built, each of these content types can be
reused and adapted across product lines, personas, or industries,
without reinventing the wheel each time.
This is how you scale velocity without sacrificing quality. AI
does the heavy lifting. Your team applies the judgment. Together,
that creates content that’s fast, on-brand, and grounded in real
user needs.
6. Start a Regular Portfolio Review
-----------------------------------
One of the benefits of portfolio-style management is that it
provides a structure for you to scale, maintain or deprecate
individual bets without undermining the initiative as a whole. As
you make this transition, one of the best things you can do is to
schedule time to review each set of bets you’re making
(efficiency, effectiveness, expansion) impartially – what’s
working? What’s not? Which processes does it make sense to invest
in automating? Which procedures should we leave alone?
It’s easy to automate something once, then forget about it – even
when that automation ceases to pay dividends or spawns unintended
consequences. The inevitable result of that approach is tech
debt, frustration and patchwork automations/workflows, in which a
new automation is created to fix the deficiencies of an existing
automation.
The superior approach long term is to treat every automation or
AI-driven workflow like a living asset, not a one-off project.
That means revisiting it regularly, assessing whether it’s still
delivering value and making an unbiased, clear-headed decision
whether it should be scaled, refined, or retired.
The reality is that - like any portfolio - some assets will
compound indefinitely. A reporting automation that saves five
hours a week for each of a dozen teams worth doubling down on.
Others will hit diminishing returns. Maybe the workflow only
works under certain conditions, or the maintenance overhead has
grown larger than the time it saves. At that point, it’s better
to deprecate it than keep propping it up.
What you’re building isn’t a collection of hacks. It’s a
portfolio. And like any portfolio, your job is to maximize
returns and minimize drag. That’s why regular (monthly/quarterly)
reviews matter - they keep you honest. They make sure you aren’t
using new workflows/tools as band-aids on top of broken ones.
All that leads to the real advantage: over time, you’ll grow a
portfolio of high-performing, margin-expanding automations,
workflows and tools while your competition winds up with
spaghetti workflows and staggering tech debt.
This week’s issue is brought to you by Optmyzr.
-----------------------------------------------
One of my favorite things about the tool is that - fundamentally
- Optmyzr isn’t about flashy features, it's about giving
marketers agency, clarity and leverage. I’ve demoed a ton of PPC
tools that try to take away control, and none of them work.
Optmyzr provides the best of both worlds: the automations (like
PPC Sidekick + Rule Engine) are powerful enough to run accounts,
while the platform itself is built around PPCers. The end result
is that you are still in control….but you have far more operating
leverage.
Here’s how Optmyzr maps to the portfolio framework we outlined:
* Efficiency (Low-Risk, Fast Payback): Automate repetitive tasks
with one-click optimizations, real-time alerts, rapid audits, and
quick-fire reporting — designed to free up time for strategy, not
just execution.
* Effectiveness (Decision-Support Upside): The AI Sidekick gives
immediate insights on where performance is shifting and why. You
get improved cause analysis and response speed without guesswork.
* Expansion (Upside in New Capabilities): Labs and AI-powered ad
text suggestions let you experiment rapidly with new creative
concepts and structures, with brand safety mechanisms built in.
Optmyzr handles tedious work so you can focus on narrative,
judgment and compounding advantage. If you want to see how
leveraging AI without surrendering control plays out in PPC,
Optmyzr offers a 14-day free trial (no credit card required).
-->Get Optmyzr (
)
Get Optmyzr ( link )
---------------------------------------------------
Your AI Strategy Should Read Like an Operating Plan
---------------------------------------------------
A good AI strategy isn’t a deck or a demo. It’s not a list of
tools, a set of GPT licenses or a pilot program. It’s an
operating plan that explains three things with ruthless clarity:
Where the leverage is. How you’re going after it. How you’ll know
if it’s working.
That plan should read like a portfolio, with efficiency bets that
quietly reduce friction, effectiveness bets that improve outcomes
and expansion bets that explore what might create new revenue
streams. Some will be successes, and some will be failures. The
discipline is in reallocating resources (capital, time, focus)
and attention regularly, so the portfolio as a whole keeps
getting stronger.
Everything else is theater.
So, as the noise ramps up this fall, skip the shiny objects.
Focus on workflows, leverage, distribution and the cadence of
review and reallocation. The winners won’t be the ones with the
most pilots or the flashiest announcements; they’ll be the ones
who quietly went about the boring work of using AI to build a
better, more efficient, more effective business.
And that’s the paradox. In a space obsessed with trends, sexiness
& hype, the real advantage belongs to the people/brands
disciplined enough to ignore it.
And when the inevitable question comes back — “What’s our AI
strategy?” — you’ll have an answer that isn’t just coherent, but
credible. One that moves the business forward, not just the
conversation.
Cheers,
Sam
Loving The Digital Download?
Share this Newsletter with a friend by visiting my public feed.
---------------------------------------------------------------
-->View the Newsletter Feed (
)
View the Newsletter Feed (
link )
Follow Me on my Socials
(
) (
) (
)
1700 South Road, Baltimore, MD 21209 | 410-367-2700
Unsubscribe (
) | Manage Preferences (
)