This guide starts from a phone call we get often. “We tried Meta for three months, spent a few thousand euros and nothing came of it.” When we open the account we almost always find the same scene: six campaigns with different names, fifteen ad sets on a few euros a day, all in Learning limited. A pixel installed two years ago that counts purchases twice. Ads changed every week because they “weren’t working”. And nobody who had written down, before starting, what result would count as a success.
Meta Ads can work very well for an SME, but it has strict rules: the algorithm needs data to learn, the data only arrives if tracking is done properly, and every significant change sends it back to square one. The pages that follow are written for anyone who has to decide how much to invest, set up or check an account and read the results without being led only by the platform’s reports. Every chapter ends with something to do. At the end you will find four appendices to fill in.
Before you spend: what role to give Meta
Meta does not do the same job for every business. For an e-commerce store it can be the main sales channel; for a B2B company it is more often a supporting channel. The role has to be decided before the budget, because it changes the campaign objective, the benchmark metric and the amount needed.
Four possible roles
| Role | For whom | Objective in Ads Manager | Benchmark metric |
|---|---|---|---|
| Direct sales | E-commerce with a catalogue and a low to medium average order value | Sales | Cost per purchase, margin, share of new customers |
| Low-cost leads | Local B2C services, training, events, property | Leads | Cost per qualified lead, contact rate |
| B2B leads | Companies with a long sales cycle and high-value contracts | Leads | Cost per qualified lead and per opportunity in the CRM |
| Awareness and recall | Small markets, few decision-makers, trade fairs, launches | Awareness or Traffic, plus remarketing | Reach on the right audience, brand searches, direct enquiries |
The fourth role is often the best fit for B2B companies with few potential customers. It does not bring measurable leads inside Meta, but it makes the name known to people who will later search for the company on Google or meet it at a trade fair. It needs judging with different metrics, and management needs to be told that up front.
What you need before starting
Whatever the role, there are three conditions without which it is better to wait:
- A clear offer for people who do not know you. On Meta people are not searching for your product: they come across it while looking at something else. The ad has to explain in a few seconds why they should stop.
- A page or form that converts. If the site is slow on mobile or the checkout has problems, that is where the budget ends up.
- Tracking of the events that matter. Purchases, leads and, for B2B, the ability to know which leads became customers. Without it, the algorithm optimises in the dark and you judge in the dark.
Self-assessment question. If a colleague asked you tomorrow, “What does Meta need to do for us in the next three months, and how will we know whether it has?”, could you answer in two sentences? If not, the first job is to write those two sentences.
To do. Choose one of the four roles in the table and write next to it the metric you will judge Meta by. Share it with whoever decides the budget before launching the first campaign.
The budget: from margin to the learning phase
The Meta Ads budget is calculated from three numbers: how much you can spend on each result without losing margin, how many results the algorithm needs to learn, and how many results you need to reach your commercial goal. The full method, with two examples, is in the article how much to invest in Meta Ads. Here we go over the steps needed to fill in Appendix A.
The maximum cost per result
The starting point is margin, not selling price. If a product costs 80 euros and, after the product, shipping, returns and fees, you are left with 32 euros, those 32 euros are the ceiling above which every sale loses you money. Decide what share of that margin to put towards acquisition: at 75%, the maximum cost per purchase is 24 euros.
For a B2B company there is one more step. From the first-year margin of a contract you work out how much you can spend on a customer; then, using the close rate of leads coming from campaigns, you arrive at the maximum cost per lead. If a customer leaves you 4,800 euros of margin, you want to spend at most a quarter of that to acquire them and one lead in twenty becomes a customer, the maximum cost per lead is 60 euros.
These are example figures. Yours should come from your management system and CRM, and if the CRM does not have them, retrieving them is the first job to do.
The 50-result threshold
Meta explains on its page about the learning phase that an ad set usually stabilises after around 50 results in the week following the last significant edit. During learning, performance is less stable and the cost per result is generally higher.
Which leads to the most useful formula in the whole guide:
minimum weekly budget per ad set = 50 × expected cost per result
With an expected cost per purchase of 24 euros you need 1,200 euros a week; with a cost per lead of 60 euros, 3,000 euros. If the ad set does not get there, Ads Manager shows the status “Learning limited”. Meta makes clear on its page about Learning limited that it is not a penalty: it indicates that the budget is not being spent efficiently with that setup.
What to do if the calculated budget is too high
For many SMEs the figure the formula produces is higher than what is available. The levers, in order:
- reduce the number of ad sets, because the threshold applies to each one;
- choose a more frequent event: add-to-baskets instead of purchases, a lead instead of an appointment, checking that the chosen event then leads to sales;
- change Meta’s role, moving from direct sales to awareness and recall;
- concentrate the budget on a shorter period or a single product, instead of spreading it across the whole catalogue all year.
Meta’s technical minimums
Technical minimums exist, but they are low and vary by country, currency and objective. On its page of best practices for minimum budgets Meta adds two useful rules: with a cost per result goal, the daily budget must be at least 5 times the goal, and optimising for purchases requires more budget than optimising for landing page views. Ads Manager shows a warning when the budget is below the minimum or too low for optimal results: it is worth reading.
To do. Fill in the first six rows of Appendix A with the real figures from the last year. If the minimum budget for the learning phase is more than you can invest, decide now which of the four levers to use.
Account structure for an SME
For an SME the best structure is almost always the simplest one: few campaigns, few ad sets, budget at campaign level. Every extra split halves the data each part learns from.
Two or three campaigns are enough
| Campaign | What it is for | Ad sets | Indicative share of budget |
|---|---|---|---|
| Acquisition | Finding new people | One or two, broad audience | 60-80% |
| Remarketing | Retargeting visitors, unclosed leads, customers | One, if the audience is large enough | 10-25% |
| Creative testing | Trying new messages and formats | One, with a few ads at a time | 10-15% |
These percentages are the ones we start from on the projects we manage, and they get adjusted after the first month. In B2B, remarketing carries more weight because the sales cycle is long and people come back several times before deciding. The limit on remarketing is audience size: with a few thousand visitors a month, a high budget means seeing the same ad too many times.
Meta itself, on its learning phase page, advises avoiding high ad volumes and combining similar ad sets, because combining them also combines their learnings.
Campaign budget or ad set budget
With Advantage+ campaign budget (what used to be called campaign budget optimisation), Meta shifts spend towards the best-performing ad sets, and the risk of under-delivery is lower. It is the default choice for always-on campaigns.
Ad set budget is useful when a segment needs a set level of spend: a country, an audience of existing customers, a product to push. In that case you can turn on ad set budget sharing, which lets Meta move up to 20% of each ad set’s daily budget to the others. Campaign spend stays the same.
How much Meta can spend in a day
With a daily budget, on any single day Meta can spend up to 75% more than the budget set if it sees better opportunities; over the week, spend stays within seven times the daily budget. The same page on budget sharing explains this. Knowing it avoids plenty of Monday-morning phone calls: one day spends more, another less, and the week evens out.
Names you can read
A naming convention seems like a detail, but it is what lets you read the data six months from now and match it against the CRM. The one we use is simple:
- campaign:
objective_role_market, for examplesales_acquisition_uk; - ad set:
audience_placements, for examplebroad_advantage; - ad:
format_message_date, for examplevideo9x16_delivery_2610.
The same names go into the UTM parameters of the links, so that in the CRM and GA4 every contact can be traced back to the campaign and ad that brought it in.
Watch out. Reorganising an account with many campaigns into two or three new campaigns restarts learning everywhere. It is best done once, in a period without promotions, allowing for two weeks of unstable results.
Example. A hypothetical online homeware shop has eight campaigns, one per product category, each on 20-40 euros a day. None of them exits learning. By merging them into one acquisition campaign with the full catalogue and one remarketing campaign, on the same total budget, the main ad set passes 50 purchases a week and the cost per purchase stabilises.
Objectives, optimisation events and forms
The campaign objective tells Meta what type of result to look for; the optimisation event tells it which precise action to count. For an SME the right choice is almost always Sales for e-commerce and Leads for anyone collecting enquiries.
Choosing the right event
The ideal event is as close as possible to revenue, but frequent enough to reach 50 results a week. It is a compromise, and it should be reviewed as volumes grow.
| Situation | Recommended event | Why |
|---|---|---|
| E-commerce with more than 50 purchases a week | Purchase, with value | It is the signal closest to margin |
| E-commerce with few purchases | Initiate checkout or add to basket, for a while | More frequent; check that it leads to purchases |
| B2C leads with a phone follow-up | Lead from a Higher intent instant form | Fewer careless contacts |
| B2B leads with a long cycle | Lead, with qualified leads sent back from the CRM | The algorithm gets a pointer on who to look for |
| Appointment booking | Appointment or lead, depending on volume | Appointments are few, leads more plentiful |
Instant form or website form
Meta’s instant forms are filled in without leaving Facebook or Instagram, with the details already pre-filled. They bring in more contacts at a lower cost, but some people fill them in and have forgotten about it a day later. The website form brings in fewer contacts, usually more considered ones, because the person has seen the page.
Meta offers several instant form types. The default type, More volume, is designed for quick submission on mobile. The Higher intent type adds text warning that someone will be in touch and a review screen where the person confirms their details. Meta recommends it when every contact needs a follow-up; it is only available in the Facebook and Instagram feeds on mobile.
On B2B projects we almost always start with the Higher intent type, with two or three qualifying questions (role, company size, timescales), and compare it with the website form after the first month on cost per qualified lead, not cost per lead.
Sending lead quality back to Meta
A completed form is a weak signal. To teach Meta which leads are worth something, you send the lead status back from the CRM. For instant form leads there is the Conversions API for CRM integration: you store Meta’s lead ID in the CRM and send the later stages, such as qualified lead or opportunity. We explain how to set up the stages and values in the article from lead to contract.
To do. List the events you currently send to Meta and, next to each one, how many arrived in the last week. The one you optimise for should be close to 50 per ad set, or the account will stay in learning.
Audiences: how narrow to go
With today’s tools, an SME is better off starting from broad audiences and letting the creative select the people, rather than building lots of narrow interest-based audiences. The controls that still matter are few: where to show the ads, who to exclude and who to retarget.
Broad does not mean random
A broad audience works if the ad clearly speaks to a specific person. A video showing a workshop owner wrestling with a spare-parts problem selects the people who have that problem better than a list of interests does. That is why audience work shifts to the message, which we look at in the next chapter.
Three settings still need attention:
- locations: only the areas you serve, excluding those where you do not deliver or have no sales staff;
- language and age, when they are a real constraint of the product;
- exclusions: recent purchasers, employees, suppliers, if you can upload their lists.
Remarketing with judgement
Remarketing on Meta is useful when the audience to retarget is large enough and when you have something new to say: a proof point, a customer case, an answer to a common doubt. Repeating the same ad to people who have already seen it produces high frequency and results that would have come anyway. Our glossary page on remarketing explains how to build the audiences.
Existing customers and new customers
In sales campaigns, some of the attributed purchases come from people who were already customers. That is not a problem in itself, but it needs measuring: if that share grows, you are paying for sales you would have made anyway. Separating new and returning customers, with uploaded customer lists and data from your management system, is one of the checks in the monthly report.
Self-assessment question. Do you know what percentage of the purchases attributed to Meta last month came from new customers? If you do not, you also do not know how much Meta is actually growing the business.
Creative: the work that matters most
With broad audiences and automated bidding, creative is the main lever left in the hands of whoever manages the account. It decides who stops, who clicks and, to a large extent, who the algorithm goes looking for.
The formats to start with
For an SME it is better to cover three formats rather than ten:
- 9:16 vertical video of 15-30 seconds, for Reels and Stories, even shot on a smartphone, with the message in the first three seconds and subtitles;
- 4:5 image or carousel, for the feed, with one idea per card;
- a format with real people: the owner, a technician, a customer telling their story, because these are the ads that stand out most from the competition.
For practical tips on shooting and editing vertical videos, see the guide to Reels. Up-to-date format sizes are in the guide to the brand kit for campaigns.
One message per ad
Each ad should answer just one of the viewer’s questions. The most common: how much does it cost, how long does it take, why should I trust you, what difference does it make to me. An ad that tries to say everything is not remembered. Before producing anything, write down for each ad the question it answers: if two ads answer the same question in the same way, one of them is surplus.
How many ads to keep live
Meta advises avoiding high ad volumes, because with lots of ads and lots of ad sets the system learns less about each one. In our accounts we keep three to six ads live per ad set and add new ones in batches every two weeks, removing those that have stopped performing.
Spotting fatigue
An ad gets tired when the people seeing it have already seen it too many times. The signs, to be read together:
- frequency rises week after week;
- CTR falls;
- cost per result rises with no change in budget or season.
When all three appear, you need a new ad before a bigger budget. The plan in Appendix C includes a new test every two weeks precisely so you do not reach that point with nothing in reserve.
Watch out. Changing the copy or image of a live ad is a significant edit and can restart learning. To try a new message it is better to add a new ad and leave the old one running until the new one has data.
To do. Open the last 30 days’ report at ad level and sort by spend. If the top two ads take more than 80% of the budget, they are your benchmark ads: the next test should try to beat them, not replace them blindly.
Tracking: pixel, Conversions API and consent
Without complete tracking, Meta optimises on partial data and the budget works less hard. There are four parts: the pixel in the browser, the Conversions API from the server, deduplication between the two and the connection with the CRM. Visitor consent decides what can be sent.
Pixel and Conversions API together
In its Conversions API best practices, Meta recommends a redundant setup: sending the same events through both the pixel and the server. The Conversions API recovers events the pixel loses to network issues, loading errors or browser blocking. Meta suggests aiming for a 75% event coverage ratio between the Conversions API and the pixel, which you can check in Events Manager, and sending events in real time or as close to it as possible.
For an e-commerce store on a popular platform, the Conversions API is often switched on through the partner integration. For a custom-built site you need server-side development or a gateway; that is the work we do in our development and integrations service.
Deduplication
If the same purchase arrives from the pixel and from the server, Meta needs to know it is just one. The deduplication documentation explains how: the pixel and the Conversions API must send the same event name and the same event ID (eventID in the pixel, event_id on the server). Events are deduplicated if they arrive within 48 hours of the first one. Without deduplication, conversions double and reports overstate the campaigns.
Event match quality
Events Manager shows an event match quality score for each event: it indicates how far the customer data sent (email, phone, name, city, browser identifiers) lets Meta link the event to an account. The more events carry these parameters, the more conversions get recorded. The data must only be sent with the necessary consent and in hashed form, as Meta’s documentation requires.
Consent
In European Union countries, the pixel and advertising cookies only fire after the visitor has given consent. A banner that loads the pixel before the choice is made is a legal problem before it is a measurement problem. The flip side is that some events will never reach Meta: that is normal, and it is why results should always be compared with your management system. We cover consent on the Google side in the article on Enhanced Conversions and Consent Mode; the banner rules apply to every platform.
The connection with the CRM
For businesses that sell off the website, tracking does not stop at the form. The CRM must store the contact’s source (UTM, fbclid parameter or _fbc cookie, lead ID from instant forms) and send the later stages back to Meta. For the Conversions API for CRM, Meta sets minimum requirements on volume and frequency of uploads, listed in the integration documentation: check them before starting, because with few leads a month the algorithm will not be able to optimise directly on CRM stages, even though the measurement is still useful.
To do. Make a test purchase or submit a test form, then open Events Manager. Check that the event appears only once, with both browser and server as sources, and that it carries the customer parameters. It is the quickest check in Appendix B.
The first 90 days, week by week
The first three months are for setting up tracking, getting through the learning phase, finding two or three ads that work and making the first budget decision. This is the sequence we follow.
| Weeks | What you do | What you do not do | Expected result |
|---|---|---|---|
| 1-2 | Tracking verified (Appendix B), structure with two or three campaigns, first 3-6 ads, budget calculated (Appendix A) | Launch before the events have been verified | Account ready, campaigns in learning |
| 3-4 | Daily check of spend, event checks, quick responses to leads | Change budget, audience or ads | First ad set out of learning |
| 5-6 | First creative test (Appendix C), first comparison with management system or CRM | Judge the channel on the first two weeks’ data | Stable cost per result, first data on quality |
| 7-8 | Second test, first budget increase if cost is below the maximum | Increase budget and change ads in the same week | Benchmark ads identified |
| 9-10 | Third test, possible remarketing campaign if the audience has grown | Open a new campaign for every idea | Final structure |
| 11-12 | Quarterly report (Appendix D), decision: increase, hold, change Meta’s role | Renew the budget without comparing against the CRM | Documented decision for the next quarter |
How to increase the budget
You increase it when the ad set has exited learning and the cost per result is below the calculated maximum. Meta makes clear on its page about the last significant edit that budget changes can restart learning, depending on how large they are. That is why we increase in moderate steps, one at a time, waiting a few days between steps, and always read the data from the date of the last significant edit.
More budget does not bring proportionally more results: beyond a certain threshold Meta has to look for less interested people and each result costs more. When cost rises along with frequency, it is time to broaden the audience, refresh the creative or open a new market.
Example. A hypothetical company running training courses for professionals starts with a maximum cost per qualified lead of 45 euros. In weeks 3-4 the cost per lead is high and unstable. In week 5 the ad set exits learning and the cost per qualified lead falls below 40 euros. The first budget increase comes in week 7, the second in week 9, after checking in the CRM that the extra leads sign up for courses at the same rate as the first ones.
Reading results without being fooled
The results in Ads Manager are the ones Meta attributes to itself, under its own rules. They should always be compared with your management system or CRM, and read knowing how those rules have changed.
What has changed in attribution
Since 2026 Meta has counted clicks differently. Following the announcement on click attribution, since the end of March 2026, for campaigns optimised for website or in-store conversions, only link clicks count as click-through; likes, shares, saves and videos watched for at least 5 seconds have moved to a separate category. Anyone comparing 2026 with 2025 without knowing this sees a drop that is partly just a change in the rules. The topic is covered in more depth in the article on misleading ROAS.
The metrics and where to find them
| Metric | Where | What it tells you | Decision it triggers |
|---|---|---|---|
| Cost per result since the last significant edit | Ads Manager | Whether the campaign is staying below the maximum cost | Above the maximum for two weeks: review creative or event |
| Delivery status | Ads Manager | Whether ad sets are in learning or limited | Learning limited: reduce ad sets or change event |
| Frequency and CTR per ad | Ads Manager | Creative fatigue | Frequency up and CTR down: new ad |
| Share of new customers | Management system and customer lists | Whether Meta brings growth or sales that were already yours | Share falling: review exclusions and remarketing |
| Qualified leads and opportunities | CRM | Contact quality | Cost per qualified lead above the maximum: review form and message |
| Total sales or MER | Management system | The overall effect on revenue | Spend up and revenue flat: incrementality test |
When you need an incrementality test
If the doubt is serious, meaning it is unclear how much of the attributed sales would have come anyway, the only way to find out is a test with a control group. Meta offers Conversion Lift tests, with access on request; alternatively you can run geographic tests, switching campaigns off in some areas for a period. For an SME this is not a monthly exercise, but once a year, on the campaigns that take the most budget, it gives an answer no report can give.
To do. Fill in Appendix D with last month’s data. The rows you cannot fill in show you exactly which piece of tracking or CRM is missing.
The 12 most common mistakes
These are the mistakes we find most often when auditing SME accounts. Almost all of them cost budget without the report showing it.
- Starting without deciding Meta’s role. The channel gets judged on the wrong metric.
- A budget worked out backwards. “Let’s put in 500 euros” instead of starting from cost per result and the 50-result threshold.
- Too many campaigns and too many ad sets. None of them exits learning.
- Constant changes. Budgets, audiences and ads changed every few days reset learning to zero.
- Judging after two weeks. The most unstable period becomes the basis for switching everything off.
- Optimising on leads and measuring on leads. In B2B, a low cost per lead hides contacts that never close.
- More volume instant forms for complex services. Lots of contacts, few you can reach.
- A pixel without the Conversions API, or both without deduplication. Events lost on one side, doubled on the other.
- The pixel firing before consent. A legal problem that also makes the data less reliable.
- Oversized remarketing. Very high frequency on a small audience and sales that would have come anyway.
- Ads that are all the same. Same message, same format, same person: the algorithm has nothing to choose from.
- No comparison with the management system or CRM. Next quarter’s budget gets decided on the platform’s numbers.
Quick self-assessment
Count how many of the twelve mistakes you recognise in your account:
- 0 to 2: the account is in good shape; the work is on creative testing and gradually increasing the budget;
- 3 to 5: it is worth reviewing the structure and tracking before increasing spend;
- 6 or more: before spending any more, go back through the chapters from role to tracking with Appendices A and B.
Watch out. Fixing mistakes 3, 4 and 8 together changes the numbers Meta reports all at once: fewer campaigns, fewer double-counted conversions, a learning period. It is the expected effect, and it should be explained in advance to whoever reads the reports.
Appendix A: Meta Ads budget calculator
Fill in the “Your value” column with the last year’s data. The rows with a formula are calculated from the previous ones. The examples are made up.
| Row | Item | How to calculate it | E-commerce example | B2B example | Your value |
|---|---|---|---|---|---|
| 1 | Average value of a sale or contract | From your management system: average order value or first year of contract | €80 | €12,000 | |
| 2 | Margin after variable costs | Row 1 minus product, shipping, returns, fees | €32 | €4,800 | |
| 3 | Share of margin for acquisition | Company decision | 75% | 25% | |
| 4 | Maximum cost per customer | Row 2 × row 3 | €24 | €1,200 | |
| 5 | Close rate of leads from campaigns | From the CRM (B2B or leads only) | - | 5% | |
| 6 | Maximum cost per result | Row 4, or row 4 × row 5 for leads | €24 | €60 | |
| 7 | Minimum weekly budget per ad set | Row 6 × 50 | €1,200 | €3,000 | |
| 8 | Number of ad sets planned | Account structure | 1 | 1 | |
| 9 | Minimum monthly budget for acquisition | Row 7 × row 8 × 4.3 | €5,160 | €12,900 | |
| 10 | Total monthly budget | Row 9 plus remarketing and testing (10-25% each) | about €6,500 | about €16,000 |
If row 10 is more than the available budget, choose a lever and write it here:
- reduce the number of ad sets (row 8)
- optimise for a more frequent event (lower row 6)
- change Meta’s role: awareness and remarketing
- concentrate the budget on one period or one product
Appendix B: Tracking checklist
Pixel and Conversions API
- The Meta pixel is installed on every page of the site, once only
- The pixel only fires after consent to marketing cookies
- The Conversions API is live and sends the same events as the pixel
- The event coverage ratio between the Conversions API and the pixel is checked in Events Manager
- Each event has the same name and the same ID in the pixel and on the server, for deduplication
- A test event appears only once in Events Manager
- Events arrive in real time or close to it
Events
- The standard events used are the right ones (Purchase, Lead, InitiateCheckout, AddToCart, CompleteRegistration)
- The purchase carries the correct value and currency, with VAT and shipping always treated the same way
- The lead is only recorded when the form is successfully submitted, not on the button click
- Test events and staff events are excluded or identifiable
- The event being optimised for arrives around 50 times a week per ad set
Customer parameters
- Email and phone are sent hashed, with the necessary consent
- The event match quality score is checked for the main events
- The
fbpandfbcparameters are passed from the server when available - The recommended actions in Events Manager have been read and assessed
Instant forms
- The form type is chosen according to the follow-up (Higher intent for services with a call-back)
- The form has two or three qualifying questions
- The privacy notice is linked and up to date
- Leads arrive in the CRM automatically, without manual downloads
- Meta’s lead ID is stored in the CRM
CRM and contact source
- Website forms store UTMs,
fbclidand landing page in hidden fields - The CRM keeps the source on the contact and on the opportunity
- CRM stages (qualified lead, opportunity, customer) are defined and kept up to date by the sales team
- Stages are sent back to Meta with the Conversions API for CRM, if volumes allow
- There is a monthly check between leads in Meta and leads in the CRM
Reports and naming
- Campaigns, ad sets and ads follow a naming convention
- The same names are used in the UTMs
- The “Last significant edit” column is added to the Ads Manager view
- The monthly report (Appendix D) has an owner and a fixed date
Appendix C: 12-week creative testing plan
One test at a time, with a precise question. Each test runs for at least two weeks or until each ad has enough results for a comparison. Winning ads stay live, losing ones are paused, without editing the ones that are running.
| Weeks | Test question | What you compare | Metric | When to stop it |
|---|---|---|---|---|
| 1-4 | No test: learning | 3-6 starting ads with different messages | Cost per result | Nothing is stopped before week 5 |
| 5-6 | Which format? | Vertical video versus image or carousel on the same message | Cost per result, CTR | After two weeks or when one is spending much less |
| 7-8 | Which message? | Price or saving versus time or convenience | Cost per qualified result | After two weeks |
| 9-10 | Who is speaking? | Someone from the company or a customer versus the product alone | Cost per result, watch time | After two weeks |
| 11-12 | Which opening? | Three versions of the first three seconds of the best video | 3-second video plays, cost per result | After two weeks |
Rules for every test:
- only one variable changes between the ads being compared
- new ads are added; live ads are not edited
- the test result is written in a shared document, with the date and the decision taken
- the winner becomes the basis for the next test
Appendix D: Ten-row monthly report
| Row | Metric | Where | This month’s value | Previous month | Decision |
|---|---|---|---|---|---|
| 1 | Total Meta spend | Ads Manager | |||
| 2 | Results (purchases or leads) | Ads Manager | |||
| 3 | Cost per result since the last significant edit | Ads Manager | |||
| 4 | Ad sets in Learning limited | Ads Manager | |||
| 5 | Average frequency and CTR of the top three ads | Ads Manager | |||
| 6 | Share of new customers in sales | Management system | |||
| 7 | Qualified leads and opportunities from Meta | CRM | |||
| 8 | Cost per qualified lead or per opportunity | CRM and Ads Manager | |||
| 9 | Total sales or MER (revenue over total advertising spend) | Management system | |||
| 10 | This month’s creative test and result | Test log |
How to use it: fill in the “Decision” column at the end of the meeting, with a single action for each row that needs one. A row with no decision for three months running is a metric you can stop looking at.
