Hey — Patrick here from Ad Juice!
Welcome to this week’s edition of The Squeeze, the newsletter created for B2B marketing operators looking for fresh LinkedIn Ads tactics, proven on real ad accounts.
I’ve worked with dozens of B2Bs over the last 7 years, and there’s one trait they all share when it comes to running paid ads programs:
They’re obsessed with launching new creative, hoping they’ll eventually find that one format that suddenly quadruples qualified inbound so they can finally get execs off their case.
But, there’s two huge problems with this approach all those companies share:
There’s no strategy behind the creative output and no system in place to identify what’s working, what isn’t, and scale it.
This leads to companies killing ads that were working, pushing new formats that don’t work, and then getting a frustratingly inconsistent volume of qualified inbound.
Some months it feels like it’s working, other months it doesn’t, and nobody really knows why.
For this reason, I’ve always put a huge emphasis when working with clients on not only identifying winning ad formats to scale, but building a workflow that allows us to record and inject insights back into the broader demand strategy.
This week, I’m going to walk you through the workflow I’ve been developing and refining over the last 7 years to do it.
Let’s get stuck in.
Start with high-quality data
An effective LinkedIn Ads strategy that generates consistent, predictable qualified inbound, lives and dies on good data.
Bad data = inconsistent growth.
Good data = predictable growth.
When it comes to data, you have two types you need to lean into: quantitative and qualitative.
Qualitative data—predominantly sourced from customer interviews, discovery call transcripts, and reading market trend signals from sources like LinkedIn, Reddit, and YouTube—allows you to build an ad messaging strategy built around relevant topics and themes.
Quantitative data—predominantly sourced from your in-platform attribution metrics like campaign, ad-set, and ad-level pipeline volume and acquisition costs—allows you to understand what the market responds best to so you know what to scale and kill.
You could summarise it simply to this: qualitative data to ideate creative direction, quantitative data to validate and scale what works.
Our workflow uses both in 3 micro cycles, and 1 macro cycle every 90 days to keep ad accounts continually improving in relation to the two most important goals for most B2Bs: increasing qualified inbound volume, decreasing cost per acquisition.
Let’s take a look at how it works.
The 90-day ad creative workflow
Foundation
When you’re first building your testing roadmap, the best place to start isn’t: “what ad formats should we run?”
That’s where where most people usually start.
They’ll say: “we need to run thought-leader ads” or “we need to run some video content!”
The best place to start is answering the question: “what should we say that’s going to resonate with the audience?”
To do that, you need to collect qualitative insights, from real prospects.
The way I do that is the same process I’ve run every Ad Juice client through, analysing as many discovery calls from closed/won prospects as possible to find out what’s keeping them up at night.
The reason I go with discovery calls is because that’s when you find out what’s on a prospects mind before they become a customer. That’s important because that’s also the state of mind everybody you’re targeting is going to be in.
Once I have the data, I’ll then run an analysis through Claude with this prompt:
You are an expert sales analyst specializing in qualitative analysis of customer conversations and voice-of-customer research. I am providing you with a set of discovery call transcripts from deals that closed/won. Your task is to identify the most common pain points prospects referenced when they first reached out, and rank them by frequency.
Analyse the transcripts attached using the following approach:
1. **Focus on the early portion of each call** — specifically the moments where prospects describe why they reached out, what prompted them to take the call, what's not working in their current situation, and what they're trying to solve. Prioritize the prospect's own words over the rep's reframing.
2. **Extract pain points** as specific, recurring problems, frustrations, gaps, or triggers. Distinguish genuine pain points from surface-level features they asked about. If a prospect mentions multiple pains, capture each one separately.
3. **Cluster similar pain points together** even when prospects use different language to describe the same underlying problem. Use a normalized label for each cluster.
4. **Count frequency** by the number of distinct calls (not mentions) in which each pain point appears. A pain point mentioned five times in one call still counts as one.
5. **Output a ranked list** from most frequent to least frequent. For each pain point include:
- The pain point (clear, concise label)
- Frequency (e.g., "8 of 12 calls")
- A 1-2 sentence description of how prospects typically express it
- 2-3 representative verbatim quotes pulled directly from the transcripts (with no call identifier if not available, or anonymized if needed)
6. **After the ranked list**, briefly note:
- Any patterns in *who* tends to raise which pain points (role, company size, industry) if discernible from the transcripts
- Any pain points that appeared less frequently but seemed especially urgent or emotionally charged
- Pain points that were implied but rarely stated explicitly
Stay grounded in what is actually said in the transcripts. Do not infer pain points that aren't supported by the text, and do not pad the list — if only a handful of distinct pain points appear, report only those. Be objective; do not editorialize about product fit or sales strategy unless directly asked.
The transcripts are attached below or in the uploaded files.Once I have the pain point analysis, I drop it into an Ad Messaging Matrix. This is a reference file I use when building ads to make sure every piece of content is relevant to prospects. You can view a detailed guide on how I build the matrix here.
From there, the next step is to build the ad concepts and launch your starting campaigns. For ideating concepts, I recently published an ad creative workflow here, and a core LinkedIn Ads creative template guide here.
To keep things simple, I typically start with three main formats, that have historically been the top-performing formats on accounts I’ve managed:
Single image ads
Thought-leader ads
Document ads
CTV and video are also strong performers, but typically less readily available, so I’ll usually plan them in the second cycle, based on proven pain points and hooks.
LinkedIn advises you to run 5 ad creatives, but I typically opt for more, aiming for 10-20 for the first round of creative testing.
The reason I do this is because I’m typically running a high reach and frequency strategy, so I want more creative diversity to account for the higher frequency volume.
From here, we move to setting up our campaigns. You can check a detailed guide, with a basic account structure for LinkedIn Ads accounts, here.
Now, one thing to keep in mind, is that, wherever possible, I change the algo settings from distributing spend based on performance, to distributing spend to all ads evenly.
You can’t do this with the awareness objective, but you can do it with the engagement objective, which is quickly becoming my go-to for thought-leader/single images ads and document ads anyway.
Here’s how it looks in platform:
By rotating the ads evenly, I have a better understanding of what’s working and what isn’t, since the platform isn’t distributing the spend to ads based on what it thinks is good performance, engagement in this case.
Instead, I can look at the numbers in-platform, see what’s actually generating calls and pipeline, and draw my own conclusions about how to scale the account.
In the first month, I typically won’t make any sweeping changes to the account. If you did the first exercise based on a decent data-set, you should start seeing trends about which campaigns are generating calls, and which ones aren’t.
Optimise
After 30 days, we move into the optimisation phase. Here, we transition from collecting qualitative insights, to collecting quantitative insights.
What this means is that, instead of talking to more prospects, we’re using in-platform data to understand which ads are working, and which ads aren’t.
The main metrics I look at are these:
Landing page CTR: Which ads generate the most website clicks. Engagement is mostly a vanity metric, but ads that make users click to the website shows it’s resonating enough to make users want to learn more. That’s meaningful.
Dwell time: Which ads are stopping the scroll? The more time users spend viewing an ad, the more likely it is that we’re building recall for our product and the problems it solves. That’s what generates consistent pipeline in the future.
Conversions: Which ads generate booked calls and pipeline? This is the most important one to track since it’s the most closely correlated with overall business growth.
Here’s a screenshot of the custom view I have built when analysing ad performance:
Based on the data I’m seeing, I’ll then make a decision about which ad concepts to evolve and which ones to kill.
You’ll typically have a reasonable idea of what’s working and what isn’t by this stage. Here’s the prioritisation system for scaling and cutting:
Anything generating consistent booked calls and pipeline warrants further creative testing. These are the most important ads to iterate.
Next, I’ll look at the ads with high website CTRs and dwell times. There’s not as much signal that they’re booking calls, but we’re still only on day 30, so I don’t want to cut them either. I’ll typically leave these live, and only iterate on them if there’s no statistically significant signal with booked calls or pipeline.
Anything that has low CTR and dwell time AND hasn’t generated any inbound I’ll pause. Sometimes, I’ll even pause these 2 weeks into the foundation phase if there’s enough spend behind the ads to determine they’re not landing.
With this data, I’ll then build out new variants around the associated pain points of top-performers, with different formats and concepts and pit them against existing ads to see how they work.
A few tools worth checking out at this point for additional ad creative inspiration that are part of my swipe-file building process:
foreplay.co - Swipe file builder that allows you to view almost every LinkedIn ad that’s live, right now, and based on how long it’s been live.
adfolio.design - Huge bank of proven B2B LinkedIn Ads, which you can filter by angles and formats.
trylapis.com - I’ve only recently started playing around with this, but, so far, I’m really liking the discovery feature which has a curated list of top-performing ads across multiple formats and channels.
I’ll also commonly save interesting ads I see on other channels, like Instagram, Reddit, and YouTube, for additional inspo to see what’s working on other channels and try and apply it to a B2B LinkedIn Ads format.
Scale
Finally, we move into the scale phase at day 60.
Here, I’m running a higher-level analysis to understand macro trends. I’ll still rerun the same process from the optimise phase, but I’m also zooming out further to work out things like:
Are there specific ad formats that generate higher inbound-volume than others?
Are there specific pain points that generate higher inbound-volume than others?
Are there specific ad concepts that generate higher inbound-volume than others?
Are there specific hook formats that consistently outperform others?
Here, I’m not just using in-platform data anymore, instead, I’m also incorporating tools like Fibbler or Dreamdata, to understand the campaign-level influenced pipeline numbers:
Are doc ads outperforming single image ads? Are thought-leader ads outperforming video ads? Are video ads outperforming spotlight ads?
From here, I can then use that information to rebuild the creative strategy for the next quarter.
For example, when I was in my last role at a top US agency, we were able to decrease CAC 14% YoY, even thought CPMs were up, because we refreshed the creative strategy every quarter. Pulling what wasn’t working, doubling-down on what was, repeat, repeat, repeat.
Based on the data we were seeing, we were eventually able to get down to a 3 ad set demand generation structure, that focussed on:
Thought-Leader Ads, Document Ads, CTV Ads formats
2-3 core pain points that consistently outperformed all the others (even ones prospects would talk about consistently on calls)
Problem, Agitation, Solution (PAS) ad formats, that called out ICP job titles in the hooks
A note to add here, is that I’m not just analysing by volume.
For some campaign objectives, like awareness, for example, the platform won’t evenly distribute spend across ads. So, some ads will have higher volume, simply because the algo decided to allocate more spend to them.
Instead, I’m also looking at in-platform CPA, what’s also generating the lowest cost, qualified inbound? Those formats also warrant further development, even if the volume isn’t there yet.
At this point, you then move back into the foundation phase and repeat the process moving into the next quarter or 90 day sprint period.
What to watch out for
The biggest risk for an account when it comes to inconsistent pipeline generation is the spray and pray strategy.
What ends up happening is that companies will launch ads, with no strategy, based on random tactics, and totally overhaul the account every 2-4 weeks because they didn’t see enough pipeline, yet.
This approach always fails, because it’s typically also guided by intuition and guesswork, rather than hard data: “I think we should try this format” or “I think that’ll be a good pain point.”
You’re essentially throwing spaghetti at the wall and hoping something sticks. But, hope isn’t a strategy.
The most effective accounts, that generate consistent pipeline, are more structured in their approach:
Ideate ad concepts based on qualitative insights
Iterate based on quantitative validation
Scale based on influence, attribution, and lift analysis
Unless you have a product nobody wants (another problem entirely), there’s almost no way that structured process can fail.
That’s it from me this week, tune in next week for an in-depth LinkedIn Ads campaign launch checklist you can use to ensure every campaign is set up to maximise pipeline from the get-go.
Thanks for reading and catch ya in the next one 🤘
P.S. Wanna have me run your LinkedIn Ads for you? Hit this link to book a 30 minute strategy call where I’ll ask a few questions about your biz and strategy, then send across an in-depth 90 day roadmap to get your account generating consistent pipeline.