Gevity · 21 September 2026
Mastering the
Algorithm
How to build winning social media ad campaigns.
LEARNBUILDLAB
Mastering the
Meet your host
Tom Nwachuku
A bit more about me
- US Army veteran
- UCLA MBA
- Mandarin speaker
- Founder, Summit X Digital
Meta · Platform-side career experience
Digital campaign strategy for entertainment and media advertisers while working at Meta.
TikTok · Platform-side career experience
Leadership of campaign managers across education, non-profits and service businesses while working at TikTok.
Creator partnerships · Career experience
Creator campaign strategy across consumer categories. This is Tom’s career background, separate from Summit engagements.
THE DESTINATIONTO JACKSONVILLE.
THE REASON TO CARETO MOM’S FOR CHRISTMAS.
The whole job
Sales vs.
Marketing
Same hitchhiker. Same road.
The accurate one
TO JACKSONVILLE. Honest, correct, and ignored.
The one that gets the ride
TO MOM’S FOR CHRISTMAS. The same trip. Now you want to help.
Marketing creates demand. Sales captures it.
How the 90 minutes will work
Session breakdown
| Time | Section | Activity |
|---|---|---|
| 3:30 – 4:00 | Learn. Ads basics, start to finish | About the AlgorithmMarketing FunnelsHow Ads Get PricedFundamentals of Building a Social Campaign |
| 4:00 – 4:30 | Build. Map your own campaign | Hands on practice. A real business that is stuck. Diagnose the funnel, fix it, and leave with the full campaign setup |
| 4:30 – 5:00 | Lab. Your questions, your accounts | Ads Manager, LiveReal Campaign structure reviewedPlanning for Q4 |
You leave with a one page blueprint and everything you need to invest and scale your social media ads.
Hands up · nobody is keeping score
Let’s talk about
social media ads. True or false?
1
Boosting a post is the same as running an ad.
2
Narrow targeting always beats broad.
3
If it is not working after two days, I should change it.
4
I need a bigger budget before this can work.
5
Better targeting will fix a bad ad.
All five are myths. Use the framework to decide what applies to your business.
The actual problem
One campaign.
Endless choices.
ObjectivesPlacementsBid strategies
AudiencesEventsAttribution windows
Hundreds of combinations. Every one of them looks plausible.
The same choices, sorted
Every option. The few that matter.
TikTok
Placements
In-Feed Ad Spark Ads Video Shopping Ads Top Feed TopView Pulse Branded Mission Branded EffectObjectives
Reach Video Views Traffic Community Interaction Lead Generation App Promotion Web Conversion Product SalesTargeting
Broad Location & Language Demographic Interests & Behavior Device Contextual Custom & Lookalike Smart TargetingOptimization
Clicks Landing Page Views Leads Conversion View Content / Add to Cart Initiate Checkout Value-Based Optimization Mobile App InstallMeasurement
Pixel & Events API Attribution Manager Split Test Conversion Lift Brand Lift Study Post Purchase Survey Media Mix Modeling Store Visit AttributionFacebook & Instagram
Placements
Advantage+ (automatic) Instagram Reels Instagram Feed Facebook Feed Stories Marketplace Audience Network MessengerObjectives
Awareness Traffic Engagement Leads App promotion Sales six, since ODAXTargeting
Broad Location & Age Advantage+ Audience Detailed (interests) Custom Audience Lookalike Exclusions Special Ad CategoryOptimization
Link Clicks Landing Page Views Leads Purchase Add to Cart Initiate Checkout Value App InstallsMeasurement
Pixel + Conversions API Attribution Setting A/B Test Conversion Lift Brand Lift Advantage+ reporting Marketing Mix Modeling Offline Conversions
Where you start
Standard
Needs scale or budget you may not have yet
One from each column is a campaign.
Question one
First.
What are your true business goals?
- Not more followers.
- Not getting your name out there.
Question
What has to happen in the business for this to have been worth it?
Return on Ad Spend. ROAS. What came back, divided by what you put in.
Question two
Second.
How do you translate
How do you translate
your business goals into
an advertising campaign?
Everything else in this room is downstream of those two sentences.
Ace the auction
Six things you really control
01
Campaign setup
The objective you pick.
02
Audience targeting
Who you let it reach.
03
Creative messaging
The thing people see.
04
Bidding optimisation
What you pay for.
05
Measure, test, learn
Tests and reporting.
06
Sharing signals
The pixel and events.
Question
Which one of the six is the most important?
03. Creative. Two of the three terms in the auction are whether your ad works. Signals are the precondition. Creative is the lever.
Your goal picks the objective
TikTok
Awareness
→
Reach
Interaction
→
Traffic
Consideration
→
Video Views
Intent
→
Lead Generation
Conversion
→
Web Conversions and App Installs
Facebook & Instagram
Awareness
→
Awareness
Consideration
→
Traffic
Consideration
→
Engagement
Consideration
→
Leads
Consideration
→
App promotion
Conversion
→
Sales
TikTok
One objective for every stage of the funnel. Name the stage and the objective picks itself.Facebook & Instagram
Six objectives, and four of them are the same stage. Meta puts almost everything in consideration.Ads basics 101
The marketing
funnel
Too many brands try to get married on the first date.
- Awareness. Who are you?
- Consideration. Why you?
- Conversion. How do I buy?
At a small budget, start with one campaign and a measurable outcome.
Marketing goal → creative strategy
Awareness
TikTok
Authentic, TikTok first. The brand is present but not aggressive.Facebook & Instagram
Native to Reels, and safe with the sound off. Brand in the first seconds.TikTok
ParticipationViral Sounds & TrendsCommunity Member
Facebook & Instagram
Hook in 3 secondsCaptions always onCreator or UGC
Consideration
TikTok
“Edutainment”. Now that they are listening, lean in.Facebook & Instagram
Show the proof. Now that they are listening, lean in.TikTok
Testimonial / ReviewTips & TricksStop & Watch
Facebook & Instagram
Testimonial / ReviewDemo or before and afterCarousel for detail
Conversion
TikTok
Inspire action to leave the most engaging app in the world.Facebook & Instagram
Make the next step the most obvious thing on the screen.TikTok
UrgencyHow To ApplyWhy Behind The Offering
Facebook & Instagram
Urgency and offerOne clear call to actionWhy you, in one line
Where everything lives
Business
Manager
Manager
Ads Manager
Events Manager
Audience Manager
Creative Center
Commerce
Zooming into Ads Manager
How a campaign is
actually built
CAMPAIGN → what is your goal?
the objective you are buying
Campaign
AD GROUP → who do you want to
reach?
the audience, plus budget and bidding
Ad groupAd group
AD → what do you want to
show them?
the creative they actually see
AdAdAdAd
Answer them in that order and you have built a campaign.
What you control at each level
Three levels, and each one owns different settings. Put a setting at the wrong level and you will spend the session wondering why you cannot find it.
TikTok
Campaign
Buying typeAdvertising objectiveSplit test
Campaign budget optimisationBudget
Ad group
PlacementAudience targetingBudget
Bidding and optimisationBrand safety
Ad
Spark ad deliveryAd detailsCreative
Call to actionDestination page
Facebook & Instagram
Campaign
Buying typeCampaign objectiveBudget
A/B testCatalogSpecial ad category
Ad set
ConversionBudget and scheduleAudience
PlacementsDynamic creative
Ad
Ad setupAd creativeDestination
Tracking
TikTok
Up to 20 ads per ad group.Facebook & Instagram
Conversion at the ad set level holds five settings: conversion location, conversion event, performance goal, dataset, attribution. A generalised view. Individual settings are nuanced and move occasionally.What happens in milliseconds
How your ad gets priced
The feed has to stay worth opening. If money alone decided, it would not be.
The auction scores every ad on whether people will respond, not just what you pay. Highest eCPM wins.
eCPM=
Bid×
estimated
Click‑through Rate× estimated
Conversion Rate
Click‑through Rate× estimated
Conversion Rate
Each one explains itself.
Only one of those three terms is money.
The other two are whether your ad works.
What you pay. Your bid.
What you build. Signals, targeting, creative, the optimisation levers. Measure, test, learn.
Second price auction
What this means for you
You are not bidding
against a price.
You are bidding against how likely everyone else's ad is to work.
That is why $20 a day and a good hook can sit in the same feed as a studio.
How budgeting affects delivery
Budget rules that matter
- Daily budget, not lifetime
- Raise by 20 to 30 percent, never double overnight
- Then a day and fifty conversions before the next change
- Raise once you spend 80 percent of it, consistently
- Campaign budget optimisation from two ad groups up
Every budget change is a bid change, and it can reset learning.
The most expensive mistake in the room
Why ads look
broken at first
A new campaign is exploring, hunting for who converts.
50conversions before
it stabilises
it stabilises
20in ten days, or it
never passes at all
never passes at all
Wild costs on day one are exploring, not failing.
While it learns, do not:
- Lower the bid.
- Move the budget by more than twenty percent.
- Touch the settings. Meta stamps it Learning limited when it never gets there.
Liquidity
The word that explains all five
How much room the system has to work in.
Deep
Broad audience, every placement, enough budget. Millions of chances to find your buyer, so it finds the cheap ones.
Thin
Narrow audience, one placement, ten dollars a day. It takes what it can get and pays more for it.
Every constraint you add drains liquidity.
Each one you tighten starves the auction of signal
Every constraint is a lever
Targeting
Audience too small
Placement
Too few surfaces
Budget
Cannot buy fifty events
Bid
Set too low to win
Optimisation
An event that rarely fires
Feed · Reels · Stories · Messenger · Search · Audience Network
What liquidity costs you
The lowest fruit
is cheapest
Every constraint sends the system further up the tree for the same person. It still finds them. You pay for the ladder.
Start at the bottom. Climb only when the bottom runs out.
Tip one
Avoid frequent edits
Do not touch anything until it exits learning. These edits restart it:
| Campaign | Ad set | Ads |
|---|---|---|
| Budget | Targeting | Any change |
| Bid amount | Optimisation event | |
| Placement | Adding new creative | |
| Bid strategy | Bid strategy, bid amount, budget | |
| Pausing for over seven days |
Set it, then leave it alone for a week. Fiddling on day two is the costliest habit.
Lever 02 · connect audiences
Three ways to find people
Demographic, interest and behaviour
Age, location, device. Interests like beauty or travel. And behaviour, the one nobody uses:
Completed a videoLiked one
Followed a creatorTouched a hashtag
In the last fifteen days.
Custom audiences and re-engagement
People who already acted:
Your customer fileView ContentAdd to Cart
Watched a videoOpened a lead form
This is retargeting, the cheapest audience you will ever buy.
Smart targeting
- Automatic Targeting runs the whole thing
- Targeting Expansion widens a group you built
- Lookalikes copy your best customers
Lookalikes need a useful seed. Automated targeting has different requirements.
Stacking twelve interests is the most common beginner mistake, and it guarantees nothing delivers.
The question everyone asks
What if I want specific targeting?
One ad group
Can be narrow.
All of them together
As broad as possible.
- Test narrow if you have a reason to.
- Widen the moment performance drops. A narrow audience runs out of people before you run out of budget.
At small budgets the creative does the targeting.
Campaign · Discovery calls
Everyone
18+, your city, nothing else
Warm
People who visited your booking page
One interest
Parents of school age children
Three groups, one campaign. Together they are still broad.
Lever 04 · bidding and optimisation
Climb the ladder
Every rung is a bigger ask. Fewer people do it, and each one is worth more.
Engagement
No website and no pixel needed.
Click & landing page view
No pixel on TikTok. Base pixel on Meta.
Conversion
Pixel plus a reliable volume of measured events. Judge it on conversion quality and cost.
Value
Value and currency on every event. Compare return using your own purchase data.
Stand one rung below where you want to be. Climb when the events start coming back.
Two ways to bid
Spend it all,
or cap it
Spend based · Maximum delivery
Meta calls it Highest volume. Spends it all, for the most results.
Goal based · Cost cap
Holds the average cost near your bid, or underspends.
Start on maximum delivery. Cap it once you trust the cost.
Campaign budget optimisation
Hand the budget to the campaign instead of splitting it evenly across ad groups, and it moves money to whatever is working.
| Ad group A | Ad group B | Results | |
|---|---|---|---|
| Split evenly | $10 | $10 | 30 |
| Campaign budget | $15 | $5 | 35 |
Same twenty dollars. Meta calls this Advantage+ campaign budget now.
Lever 03 · creative messaging
Nine moves that make ads work
Tip 1
Frame vertically
Tip 2
Let creators take the lead
Tip 3
Use the power of sound
Tip 4
Overlay concise and informative text
Tip 5
Get your key message out early
Tip 6
Include a strong call to action
Tip 7
Break the fourth wall
Tip 8
Keep it real and entertaining
Tip 9
Ride the latest trends
Three different hooks beat three edits. The auction ignores near copies.
Run the creator's post, not a copy
Standard ad
TikTok
FollowingFor You
Sponsored
@yourbrandThree days in and I am obsessed
Learn more
Facebook & Instagram
Reels
@yourbrand
Sponsored
Three days in and I am obsessed
Original audio · @yourbrand
Learn more
TikTok
Spark adFacebook & Instagram
Partnership adTikTok
FollowingFor You
Sponsored
@examplecreatorThree days in and I am obsessed
Learn more
Facebook & Instagram
Reels
@examplecreatorFollow
Sponsored
Three days in and I am obsessed
Original audio · @examplecreator
Learn more
TikTok
1
It runs from the creator's handle, not yours. Photo, name and taps go to their account.
2
It keeps the likes, comments and shares it already earned, and walks them into the auction.
3
The music disc is live, so the sound page is tappable. It reads as a post, because it is one.
4
The creator generates a code in their account. You pick 7, 30, 60 or 365 days of authorisation.
Facebook & Instagram
1
It runs from the creator's handle, not yours. Meta calls these partnership ads.
2
It keeps the likes, comments and shares it already earned, and walks them into the auction.
3
You can run it as a Reel, in Stories, and in feed from the same authorisation.
4
The creator approves you once in their professional dashboard. No reposting.
If you are already paying creators, you have already bought the ad.
TikTok
SparkFacebook & Instagram
A partnership ad just puts budget behind the post they made.One goal, five settings
Drive purchases on the site.
Campaign objective
What do you want this campaign to achieve?Website Conversions
Conversion location
Where should the action happen?Website
Conversion event
What action should occur?Complete payment
Performance goal
What should the auction bid on?Conversion
Dataset and data source
What sends the signal back?Pixel and Events API
Change the objective and all five change underneath it. Pick the wrong one at the top and every screen after it is wrong.
Four things to settle before you open Ads Manager
One platform
- Not five. Five platforms on one budget is five campaigns that all fail.
- Pick the one your buyer already scrolls.
- Put everything there.
A real daily budget
- Roughly ten times your target cost per result, per day.
- Below ten dollars a day it is not a test.
- It is a donation.
A target cost
- What is a customer worth in ninety days?
- Spend about a third of that to get one.
- That number judges every screen after this.
Where they land
- The exact page, not the homepage.
- One product, one offer, one button.
- No website? Use an instant form.
Every broken campaign you will see broke on one of these, before a single setting was touched.
Web data connection
What is a pixel?
Code on your site that reports when someone did what you wanted.
Why it matters
No signal, and the machine is guessing.
No website at all?
Instant forms collect leads in the app. No pixel.
Selling an app?
The pixel sees nothing. You need the SDK.
TikTok
TikTok Pixel
Browser side. Shares your website visitor events with TikTok and unlocks the advanced features.
Events API
Server to server. Catches what the browser misses. Run both at once, that is the recommendation.
Partner Pixel
Shopify, Squarespace, WooCommerce. The official app wires up both sides for you in about ten minutes.
Facebook & Instagram
Meta Pixel
Browser side. Shares your website visitor events with Meta and unlocks the advanced features.
Conversions API
Server to server. Same idea, different name. Run both at once, that is the recommendation.
Partner integration
Shopify, Squarespace, WooCommerce. The official app installs the pixel and the server side connection for you.
Whiteboard mapping · three groups
One funnel. Three campaigns.
Upper · Video ViewsBuilding an audience
Maximise views at the lowest possible cost.
Mid · TrafficTeaching them what you do
Drive people to a landing page at an efficient cost per click.
Lower · Lead GenerationTurning them into buyers
High quality leads at the lowest possible cost.
Every setting changes depending on which one you picked.
Group 1 · Upper funnel
Video Views setup
| Component | Best practice |
|---|---|
| Objective | Video Views, optimised for 6 second views |
| Budget | Daily budget, lowest cost bidding for maximum reach |
| Bid strategy | Cost per view if you need control, otherwise lowest cost |
| Audience | Broad, plus interest and behaviour targeting |
| Creative | High energy, trend based, strong hook in the first 3 seconds |
| Optimisation | Watch completion rate and CPM. Refresh creative every 7 to 10 days |
Retarget anyone who watched 75 percent or more. That is your mid funnel.
Group 2 · Mid funnel
Traffic setup
| Component | Best practice |
|---|---|
| Objective | Traffic, optimised for clicks or landing page views |
| Budget | Daily budget with lowest cost bidding to start |
| Bid strategy | Highest volume (Meta) / Maximum delivery (TikTok) first. Switch to cost cap if cost per click climbs |
| Audience | A lookalike of a qualified seed, if available, or relevant interests. Meta uses percentages; TikTok uses Narrow, Balanced and Broad. |
| Creative | Strong call to action in the first 5 seconds. Clickable elements |
| Optimisation | Watch click through rate. Adjust creative every 10 to 14 days |
Check the destination on a phone. A slow or confusing page can waste a good click.
Group 3 · Lower funnel
Lead Generation setup
| Component | Best practice |
|---|---|
| Objective | Lead Generation using an instant form |
| Budget | Daily budget high enough to buy fifty leads while it learns |
| Bid strategy | Highest volume (Meta) / Maximum delivery (TikTok) first. Switch to cost cap once the cost per lead is steady |
| Audience | Retarget engaged users, plus high intent interest targeting |
| Creative | Short testimonials, social proof, a clear reason to fill the form |
| Optimisation | Test form length. A/B test the incentive: discount versus free trial |
Keep the form to 3 to 5 fields. Add one qualifying question to filter intent.
Key lesson
Everything shifts by stage
| Upper | Mid | Lower | |
|---|---|---|---|
| Audience | Broad, for exposure | Lookalikes and interests | Retargeting, high intent only |
| Bidding | Highest volume (Meta) / Maximum delivery (TikTok) | Same, then cost cap | Cost cap, then value based |
| Creative | Trendy, fast, entertaining | Interactive, click friendly | Trust building, low friction |
| Watch | Cost to reach a thousand, and how many watched | Taps, and what each tap cost | Cost per lead, and what came back |
A one size fits all approach does not work.
Fix My Funnel
In groupsOn paperTwelve minutes
1The goal
________ a month,
at no more than $______ each
2Where on the funnel
Awareness · Consideration
Conversion
so we are buying ____________
3The offer
A stranger says yes to ____________
and lands on ____________
4Who
One person.
One sentence.
5The ad
Format ____________
First line ____________
6The scoreboard
$____ a day × ____ days
It worked if ____________
Fill the board and you have the campaign. Next block you open Ads Manager and build exactly this.
Handoutsummitxdigital.com/gevity
Creative fatigue and audience overlap
Why it stops working
Creative fatigue
Same people, same ad. Performance decays with nothing broken.
Tell: the same people seeing it over and over, cost rising, taps falling, and nothing changed.
Fix: new creative, not a new audience. Refresh every 7 to 14 days.
Audience overlap
Two ad groups chasing the same person. You are bidding against yourself.
Tell: several small ad groups, all underdelivering, all learning limited.
Fix: consolidate and exclude.
Both are why more ad groups is almost never the answer.
A diagnostic order
They break in the same order
1. Measurement broken it cannot see the outcome
2. Wrong objective you asked for views
3. Over constrained stacked, split, edited daily
4. Creative last the lever that matters most
Work down the list in that order. Most of it is settings you can fix tonight.
A better first question
Is targeting really the problem?
Start with what the platform can measure.
Then check the offer, the creative and the next step before adding another audience filter.
When it IS working
Two ways to scale
Vertical · more money
Raise the budget on the winner. Plus 20 to 30 percent every 48 to 72 hours. Fastest, and it risks resetting learning.
Horizontal · more surfaces
New creative, new audience, new placement, new platform. Slower, and it compounds. Where lookalikes finally earn their place.
Only scale a winner: seven days at target with cost moving under ten percent.
Scaling · right versus wrong
How to not break it
| Area | Correct | Incorrect |
|---|---|---|
| Budget | Gradual 20 to 30 percent every 48 to 72 hours | Doubling overnight |
| Audience | Broaden qualified audiences gradually. Use the platform’s available lookalike sizes. | Jumping to fully broad too fast |
| Creative | Introduce one or two fresh assets at a time, keep proven performers running | Swapping everything at once, or never refreshing |
| Bidding | Adjust cost cap in small increments | Sudden bid jumps, switching strategy repeatedly |
One change at a time, then wait. Everything in the right hand column is a way of sending the ad set back to learning.
How you get better every week
Change one thing
A split test changes one variable.
CreativeAudienceBidding and optimisation
7days minimum,
14 with learning
14 with learning
90%confidence, then
fund the winner
fund the winner
Vary four things at once and you learn nothing.
Before you spend a dollar
Your four green lights
1 · A proven offer
One stranger has already bought. If not, your problem is the offer.
2 · Somewhere measurable
A page with a pixel, or a form.
3 · Creative supply
Three videos this week. Three more next week.
4 · A learning budget
Twenty to thirty dollars a day for thirty days. Information first, customers second.
Have all four and you are ready to spend. Missing one? That is this week's job, and then you are.
It is 21 September · Q4 starts in ten days
Two Q4 strategies
Gifts and year end decisions
Ecommerce, products, anything bought for the holidays.
Be live and out of learning by mid October. Launch inside two weeks.
The mistake is launching on 15 November into the year's most expensive auction with a cold pixel.
Services, coaching, apps
Services, coaching and apps may see different seasonal demand from holiday products.
Build the offer and measurement first. Choose your launch window using buyer demand and an affordable test budget.
Seasonal auction costs vary. Plan with your own history and current demand.
If you remember nothing else
Three things
1
The platform charges you less when people respond. So creative is the primary lever.
2
Optimise for the deepest thing you can measure and afford fifty of in a week. If you cannot, step up a rung.
3
Decide the kill rule before you spend a dollar. Then do not touch it for seven days.
A test without a kill rule is a gamble.
Before you leave
One next
action.
Write it on your blueprint. Put a date on it.
Blueprint · Hooks · Slidessummitxdigital.com/gevity
If you do one thing this week, launch the test you just designed.