£5M is not a bragging number. It's a number that teaches you something. You can't manage that much spend without making every mistake that's possible to make. You can't make that much spend without failing repeatedly, learning painfully, and building systems that work because they have to.

I've run campaigns in 12 industries—fitness, SaaS, e-commerce, finance, coaching. I've managed budgets from £200/day to £200K/day. And after all those campaigns, all those ROAS swings, all those client emergencies at 11 PM, here's what actually matters. Here's what I know now that would have saved me thousands of hours and millions of pounds if I'd known it on day one.

Lesson 1: Creative Is Everything, But Structure Is What Makes Creative Work

Every agency says, "Creative is king." It's true and useless. Of course creative matters. The real insight is this: a 7/10 creative in a 9/10 account structure will outperform a 10/10 creative in a 4/10 account structure.

Here's what I mean by account structure: how you organize campaigns, audiences, and budgets. How you frequency cap. How you sequence audiences. How you separate testing from scaling. Whether you have a creative refresh schedule.

I managed a fitness brand's account that was spending £3K/day. Their best creative was objectively mediocre—workout clips filmed on a phone, low production value, nothing viral about it. But their account structure was obsessive. They rotated creative every 5 days. They had 4 audience tiers. They frequency-capped at 3. They tested new hooks every week.

Meanwhile, a competitor at that same gym brand had genuinely beautiful creative (£10K/month production budget). But their account was a mess: 47 active campaigns, no clear testing protocol, creative rotation happened randomly, audiences overlapped. Their ROAS was 2.1×. The "mediocre creative" account was doing 3.8×.

The lesson stuck with me: You can't fix structure later. You can refresh creative in 48 hours. You can't reorganize a chaotic account without pausing 80% of the spend. Build the structure first. Creative execution follows.

Lesson 2: The Best Audiences Are Built From Your Own Data, Not Meta's Interests

Every marketer believes in interest-based targeting. "Target people interested in fitness, entrepreneurship, and personal development." Facebook lets you stack interests, and it *feels* like you're being precise. You're not. You're guessing.

Here's what I learned: Your best audience is your existing customer list.

In year one, I'd spend weeks building lookalike audiences from interests. Year five, I'd spend 30 minutes uploading customer CSVs and creating custom audiences. The CSV-based audiences converted 40-60% better than interest-based audiences. Always.

Why? Because when you target "people interested in fitness," you're targeting people who clicked one fitness ad, ever. When you upload your customer list, you're targeting people who *already chose your product over alternatives.* That preference signal is gold.

The progression I now recommend:

Tier 1: Customer List (Best Converters)
Upload your email list, phone numbers, website visitors. Build 1% and 5% lookalikes from this. This is your foundation.

Tier 2: Value-Based Lookalikes (High LTV Customers Only)
Don't use your entire customer list. Use the top 30% by lifetime value. Your 1% lookalike from high-value customers outperforms your 1% lookalike from all customers by 25-35%.

Tier 3: Engagement Audiences (Warm Traffic)
People who watched 75% of your last video, clicked your website, engaged with your page. This is your retargeting base.

Tier 4: Competitor Lookalikes (Expansion)
Only after you've maxed out your own data, build lookalikes from competitor websites (via pixel data).

Tier 5: Interest-Based (Last Resort)
Interest targeting is your emergency option, not your starting point. By this point in the funnel, you're trying to reach people with weak intent signals.

A SaaS company I worked with had been running interest-based targeting for 18 months. When we switched to Tier 1-2 (customer lists and value-based lookalikes), their CPA dropped 38% while maintaining the same volume. They'd been throwing away money for a year and a half because they thought "interest targeting was how Facebook worked."

Facebook's interest categories are useful for one thing: telling you what your competitor's marketers are doing. Use the Ad Library to see what audiences competitors are using. Then ignore those audiences and build from your own data.

Lesson 3: There Is No Such Thing As a Campaign That "Just Works" Without Weekly Maintenance

The myth: You set up a campaign, let it run, and collect money.

The reality: Every campaign requires active maintenance or dies. There are no exceptions to this.

I believed the myth for my first two years. I'd set up campaigns, they'd work for 2 weeks, and I'd assume they'd keep working. Then I'd check in after a month and find ROAS had dropped 40%. I'd panic, assume the market changed, and pause the campaign. What actually happened: creative fatigue, frequency creep (users were seeing the ad too many times), and algorithmic drift (the algorithm had moved on from optimizing for my objective).

Now I build "maintenance work" into every campaign. Here's the minimum:

Daily (2 minutes): Check ROAS and cost metrics. If either dropped >15% from yesterday, flag for investigation.

Weekly (30 minutes): Audit creative performance. Check frequency caps. Check CPM trends. Refresh the bottom 20% of creatives (lowest CTR).

Bi-weekly (1 hour): Deep dive. Breakdown reporting by audience, placement, and demographic. Identify which segments are working and which are dragging down ROAS.

Monthly (2 hours): Strategic review. Do a ROAS trend analysis. Identify patterns (seasonality, day-of-week effects). Decide what to scale, pause, or restructure.

This isn't optional. This is the cost of running ads.

A fitness app I worked with had been running a campaign that pulled in £1.2K/day at 2.8× ROAS. They assumed it was set-and-forget. They didn't check for 6 weeks. When I audited the account, ROAS had fallen to 1.4× because frequency was sitting at 4.2 (creatives were fatigued). We refreshed creative (2 hours of work) and ROAS bounced back to 2.6× in 4 days. One person, 2 hours of work, recovering £400/day in lost profit. That's £2,800/week. And it only happened because we finally looked.

The best campaigns I've managed required 30-45 minutes of active maintenance per week. Not because they were broken, but because that's what it costs to keep them working.

Lesson 4: ROAS Is a Lagging Indicator — What to Watch Before ROAS Moves

Everyone obsesses over ROAS. It's the wrong metric to obsess over because it's a lagging indicator. ROAS tells you what happened last week. If ROAS is dropping, you're already losing money.

Here are the leading indicators that ROAS is about to drop:

CPM Rising (+15-20% week-over-week)
This always precedes ROAS drop by 3-7 days. When CPM rises, the market is getting more expensive. Your cost-per-acquisition will rise next. Watch CPM daily. When it spikes, investigate immediately (is it creative fatigue? frequency creep? market saturation? time of day?).

CTR Declining (even if ROAS is stable)
CTR dropping 10-15% is a warning. Your creative is fatiguing. Refresh it before ROAS follows.

Frequency Rising (above 3.5 in a tight audience)
Your audience is seeing your ad too much. Frequency of 4+ correlates with 20-30% ROAS decline in the next 5 days.

Conversion Rate Declining (while CTR is stable)
You're reaching lower-intent users. Either your audience is fatiguing, or you've hit audience saturation. Time to expand your audience or pause this campaign and test new ones.

CPC Rising (while CTR is stable)
Your competition is getting more aggressive. Bid prices are rising. This usually means the market is getting crowded in your niche. Not much you can do, but at least you know what's coming.

A coaching business I managed had ROAS at 3.1× for three weeks straight. Everyone thought it was stable. I was watching CPM (up 22% week-over-week) and frequency (now 3.8). I warned the team: "This ROAS will collapse in the next 7 days." They thought I was being paranoid. 6 days later, ROAS hit 2.1×. We'd already refreshed creative and expanded audiences by then, so the damage was minimal. If we'd waited for ROAS to drop to act, we'd have been reacting instead of preventing.

Start tracking these metrics daily. ROAS is a rear-view mirror. These are your headlights.

Lesson 5: The Client Relationship Determines Account Performance as Much as the Media Buying

Here's something I didn't expect to learn: the quality of a client relationship directly correlates to campaign performance.

My best-performing accounts had something in common: clients who didn't second-guess decisions and gave me authority to make moves quickly.

My worst-performing accounts had something else in common: clients who wanted to optimize for the wrong metrics, changed strategy weekly, or insisted on approving every ad before launch.

The approval problem was the worst. A client once insisted on approving every ad I wanted to run. This meant 2-3 day turnaround between creation and launch. Meanwhile, the market was moving daily. By the time their preferred creative went live, the hook that would have worked on Monday was stale by Thursday. Meanwhile, I'd already identified three better performing angles that were waiting for approval.

Another client optimized for "engagement rate" instead of conversions. They wanted viral-looking ads. We'd build ads that got 8% CTR but 0.8% conversion rate. I'd build ads that got 2.2% CTR but 3.5% conversion rate. Guess which one they preferred? The 8% CTR ads "looked better." We did this for 6 months before they fired me, frustrated that we weren't getting conversions. Of course we weren't—they kept rejecting the ads that converted.

The inverse is also true: I took over an account where the previous agency had full decision-making authority. No daily emails from the client. No questions about creative. No second-guessing the strategy. That account's ROAS was 3.2×. The client trusted me to optimize for conversions, not Instagram vanity metrics. The speed of testing and iteration was 3× faster than my approval-required accounts.

How to have the right client conversation:

1. Agree upfront on the success metric (ROAS, CPA, or CAC). Not engagement. Not CTR. Not "brand awareness." Not what-feels-good.

2. Give authority: "I'll optimize for X metric and report weekly. You don't need to approve each ad."

3. Set review frequency: Weekly reporting is ideal. Daily check-ins kill momentum.

4. Create a "pause trigger": "If ROAS drops below 2×, we'll pause and retest." Not: "Tell me if you don't like the results."

A high-ticket coaching client I worked with was nervous about spending £2K/day. I gave them full authority and committed to weekly performance reports. Three months later, they were spending £8K/day at 2.9× ROAS. They became my biggest client. The difference wasn't better media buying—it was authority to test fast and freedom from second-guessing.

Lesson 6: Speed Matters More Than Perfection in Testing

I used to spend hours perfecting ads. I'd debate headline wording for 30 minutes. I'd do three rounds of creative review before launching. The result: I could test 2-3 variations per week.

Then I worked with a competitor's agency that was killing us. I asked how they were testing so many angles. They said: "We don't perfect. We launch and measure." They were testing 8-10 variations per week. Their winning angles were getting discovered in real time instead of weeks after launch.

Here's the insight: 80% perfect launched immediately beats 95% perfect launched next week.

You learn by testing, not by thinking. I now spend 15 minutes creating an ad (rough, but core message is clear). Launch it. Get 500 impressions of real-world feedback. Make the next version based on actual performance data, not my opinions.

I tested this approach with a fitness app. Their team was used to "one round of approval, then launch." I proposed: "Let's launch 6 rough video variants today. We'll get real CTR data in 2 hours. We'll know which ones work. Then we can refine the winner and scale." They were nervous (the videos weren't "polished"). I launched them anyway. Within 12 hours, we'd identified the angle with the highest CTR (video 4). We refined video 4 based on real performance data. It became their scaling creative. Total time: 24 hours. Total variants: 6. The old approach would have spent 1 week perfecting one variant.

The math is simple: if speed means you test 5× more iterations per month, you'll find winners faster. Finding winners 5× faster compounds into 3-5× better account performance after 90 days.

Lesson 7: Seasonality Will Destroy Your ROAS Twice a Year — Plan for It

Every business has two seasons when ROAS drops for no reason: the "summer slump" and the "post-holiday crash."

Summer (June-August): People are on vacation. Attention is down. Intent is down. CTR drops 20-30%. Conversion rate drops 15-25%. This compounds into ROAS dropping 40-50%.

Post-Holiday (Jan 2-15 and post-Black Friday): Spending fatigue. People are catching up on work. CPA spikes 30-40%.

I used to think this was a problem. I'd panic when ROAS dropped in June, assume something was broken, and pivot strategy. Now I plan for it.

Seasonality Defense Plan:

April-May (4 weeks before summer): Run extra testing budget on new angles. Build a "summer creative bank" of 20+ fresh assets. Start transitioning audience from interest-based to lookalike-based (tighter targeting, less sensitive to low intent).

June-August: Expect 30-40% ROAS drop. Don't panic. Cut budget by 20-30%, tighten audiences, run your best-performing creative only. Treat it as a maintenance period, not a growth period.

September (Refresh): As people return from vacation, ROAS bounces back naturally. You're ready with fresh creative and optimized audiences.

Similar cadence for holidays.

A SaaS app I managed ignored this pattern for years. Every June, they'd panic when ROAS dropped from 3.1× to 1.8×. They'd blame product changes, market shifts, everything. Year 5, I told them: "June is seasonal. It happens every year. We'll cut budget 25%, ROAS will stabilize at 2.1×. September will bounce back to 3.0×." They listened. Sure enough, it happened exactly as predicted. Not a crisis—a pattern. Once you see the pattern, you can plan around it.

Lesson 8: The Best Creative Often Comes From Customer Feedback, Not the Creative Team

Agencies always assume the creative team knows what works. The creative team believes the answer is "better design," "better video production," "cooler graphics."

Here's the uncomfortable truth: customers describe why they bought the product better than anyone else.

I started asking clients to pull 20 customer testimonials every month and give them to me. Not polished testimonials. Raw: "I bought this because my coach kept telling me to use it" or "It's the only app that didn't crash when I had 500 contacts." I'd turn these into ad copy and creative angles.

Result: ads made from customer language outperformed ads made from our creative team's language by 2-3×.

An e-commerce brand was struggling with creative. They had a professional design team producing beautiful ads. I asked their sales team to share actual customer feedback from recent buyers. One line jumped out: "I bought this because I could tell it was built by someone who actually uses it." We made that the hook. "Built by users, for users." ROAS jumped from 2.1× to 3.4× and stayed there for six months.

Where to mine customer language:

• Post-purchase surveys (ask "why did you buy?")
• Sales call recordings (what did customers mention as the decision trigger?)
• Testimonials (record them raw, don't polish)
• Customer support chats (what problems are people solving with your product?)
• Reviews (why are your best customers giving you 5 stars?)

This is a 2-hour-per-month project that outweighs a team of copywriters. The creative team doesn't know why customers buy. Customers do.

Lesson 9: Budget Efficiency Matters More Than Budget Size

Everyone thinks bigger budgets = bigger problems. Wrong. The biggest problems come from inefficient budgets.

I've seen accounts spending £1M/month with horrible ROAS (1.1×). I've seen accounts spending £50K/month with exceptional ROAS (4.8×). Budget size is irrelevant. Efficiency is everything.

Here's what separates efficient budgets from wasteful ones:

Efficient Budgets (3.0×+ ROAS possible):
• Clear separation between testing (15-20% of spend) and scaling (80-85%)
• Audience overlap is audited and eliminated
• Creative refresh happens on schedule, not reactively
• CPM and CTR are monitored daily
• Budget is concentrated in proven audiences (top 3 audiences get 70% of spend)
• Frequency is capped at 3 for cold audiences, 6 for warm
• One person audits performance 30 minutes daily

Wasteful Budgets (0.9-1.5× ROAS typical):
• Testing and scaling budgets are mixed (can't tell which creatives actually work)
• Audience overlap isn't measured (same person reaches 3-4 different campaigns)
• Creative refresh happens randomly or never
• CPM and CTR are checked "when ROAS drops"
• Budget is distributed evenly across 30+ audiences (none get focused spend)
• Frequency cap isn't set (some users see the ad 10+ times)
• Performance is checked monthly

A clothing brand was spending £4K/day with 1.3× ROAS. They thought the solution was hiring a senior media buyer. What they actually needed: ruthlessly cutting the bottom 50% of campaigns (they were burning money), consolidating audiences (reducing overlap), and implementing daily monitoring. Cost: one afternoon of reorganization. Result: ROAS went to 2.8×. Doubled ROAS without spending a pound more, just by redirecting inefficient spend to efficient channels.

Before you increase budget, audit efficiency. A £1K/day budget run efficiently (3.5× ROAS) beats a £10K/day budget run sloppily (1.2× ROAS) every time.

Lesson 10: Facebook Ads Is an Ecosystem, Not a Channel — What Happens Outside the Platform Affects What Happens Inside It

Most media buyers treat Facebook Ads like a closed system. You adjust bids, creative, audiences—and ROAS changes. That's only part of the story.

What actually happens:

Your landing page changes ROAS
A 2-second improvement in page load time improved conversion rate by 18%. A worse headline on your landing page tanked conversion rate by 40% (people clicked the ad based on promise A, landed on page showing promise B). Your ad is only as good as where it sends people.

Your product experience affects repeat customers
If your product is bad, customer LTV is low. If customer LTV is low, your breakeven payback period is high, so you can't afford to pay much for acquisition. If you can't pay much for acquisition, you can only target high-intent audiences, which limits scale. This isn't an ads problem; it's a product problem.

Your email follow-up affects conversion rate
A client was getting 2.3% CTR but only 0.8% conversion rate. I watched their funnel: the email sequence after clicking the ad was terrible. People clicked, got a welcome email 12 hours later (too late), and forgot about it. We changed the email to immediate post-click nurture. Conversion rate jumped to 2.1%. The ads didn't change. The funnel did.

Your sales team affects LTV
A high-ticket fitness coaching company was getting great leads but low close rate (18%). I thought the ads were too broad. Then I shadowed the sales team. They were closing late-night prospects within 30 minutes (low intent, likely to refund). They were slow-playing early-morning prospects (high intent, likely to stay). Just scheduling changes (prioritizing high-intent leads) improved close rate to 31%. This is an ecosystem problem, not an ads problem.

Your product page affects quality of leads
A SaaS company was getting leads with low intent. I thought it was the ads. Then the product team redesigned the product page to emphasize key features instead of benefits. Suddenly, the leads coming through understood the product better. Lead quality improved. Same ads, better conversion rate downstream.

This is the hardest lesson because it's outside your control as a media buyer. But it's the most important: 60% of account performance is determined by what happens outside the ads platform.

A software company I managed was spending £120K/month at 2.1× ROAS. We spent 6 weeks trying every testing angle, audience combination, bid strategy. Marginal improvements only. Then the COO overhauled the onboarding experience (completely unrelated to ads). Suddenly the product was so good that word-of-mouth referrals spiked. Customer LTV increased 22%. With higher LTV, we could increase CAC by 20%. ROAS jumped to 2.9× with zero changes to the ad strategy. The improvement didn't come from the media buyer—it came from product.

Know your ecosystem. Audit your landing page, email, product experience, and sales process before you change ads. That's where your leverage is.

The Bottom Line

£5M in ad spend taught me that scale doesn't come from budget size. It comes from structure, speed, and systems. It comes from listening to customers instead of guessing. It comes from treating performance as a daily maintenance responsibility, not a monthly report. It comes from building accounts so efficient that you can scale 20% and maintain ROAS, instead of scaling fast and collapsing.

Most importantly, it comes from understanding that the Facebook Ads platform is just one piece of a much larger machine. Your ads are only as good as your funnel, your product, your customer LTV, and your sales process. Optimize those first. Then run the ads.