What a social media campaign actually consists of
Many companies start social media campaigns exactly backwards: they pick the budget, then look for an objective to justify it. The result is a campaign with no essential pieces set beforehand — who it's aimed at, what "it worked" means, how long it runs before you draw a conclusion. A social media campaign, in the sense that's worth a company's budget, always has the same components: a single, measurable objective, ad sets with distinct audiences, a few creative variants for each set, a budget with a clear timeline, and a measurement plan set before launch, not invented after the first numbers come in.
The starting point matters too: according to Eurostat, in 2025 only 48.07% of Romanian companies with at least 10 employees used social media, against an EU average of 63.57% — a gap that has stayed almost unchanged since 2016, when Romania was at 30.03% and the EU average at 42.49%. A significant part of the local market hasn't yet moved past the basic stage, let alone structured paid campaigns.
Reach, traffic, leads or sales: four objectives, four rules
Every major platform — Meta, TikTok, LinkedIn — requires an explicit objective at ad-set level, and that objective changes how the system bids and who it shows the ad to, not just what the final report looks like. Four objectives come up constantly: reach, traffic, leads and sales. Mixing them into a single set — optimised for shares, but expected to produce sales — is the most common reason a campaign "looks good" but doesn't bring customers.
Reach builds a new brand's presence, but should rarely be the only objective for a small business with a limited budget — it costs little per impression, but produces nothing measurable in sales. Traffic brings people to the site and feeds retargeting, but a click isn't buying intent. Leads — through forms, messages or a landing page with a clear step — are the first objective with a cost per result that's comparable from one month to the next. Direct sales require enough conversion history for the platform to have something to learn from; below that threshold, the campaign runs essentially "blind".
How you split the budget between testing and scaling
A campaign's budget shouldn't be split according to how much you want to spend, but according to what you do with the money at each stage. At the start, the objective isn't the volume of results, but information: which audience, message and format produce an acceptable cost per result. Only after that answer do you move the budget towards scaling — gradually increasing spend on the combinations that have already proven to work, without changing the audience, message and creative all at the same time.
Even the minimum testing budget has to clear the platforms' technical thresholds. On TikTok, an ad group requires a minimum of 20 dollars a day, and a campaign, a minimum of 50 dollars a day or in total. On LinkedIn, the minimum budget is 10 dollars a day, with a minimum of 100 dollars for the total budget of new campaigns — enough to get delivery started, but too little for a real B2B test. If you split a small budget across too many variants, none of them clears the technical threshold, let alone the threshold of significance.
| Stage | What you're tracking | How you know you can move on |
|---|---|---|
| Testing | Which combination of audience, message and format produces an acceptable cost per result | You have enough results on each variant that it isn't chance, not just one isolated good day |
| Validation | Confirming the winners from testing, on a slightly larger budget and over a longer period | The cost per result stays stable as the budget grows, not just while the budget is small |
| Scaling | Increasing the budget on validated combinations, with changes made one at a time, not all at once | The cost per result rises at an acceptable rate relative to the extra volume of sales |
What volume of creative a real test requires
Most social media campaigns fail not from a lack of budget, but from a lack of tested variants. A test with a single ad isn't a test, it's a guess with a budget behind it. You need enough variants to be able to compare, but not so many that the budget gets split into slices too small to mean anything.
- 01Start with 3-4 different message angles — not 3-4 variations of the same angle — so you're testing ideas, not just background colours.
- 02For each angle, use at least two formats: a short video and a static image or a carousel.
- 03Let each variant collect enough impressions or clicks before you draw a conclusion — one good day or one bad day isn't a result.
- 04Stop the weak variants and move the budget to the 2-3 that clearly stand out, instead of keeping everything you launched active indefinitely.
- 05Refresh your creative before audience fatigue shows up in the cost per result, not after it has already risen visibly.
Cost per lead, cost per result: what a good number actually means
Cost per result is the simplest figure in a campaign: the budget spent divided by the number of results in the chosen objective. The problem comes up when someone asks for "a good cost per lead", as if a universal number existed. It doesn't: a cost per lead of 50 lei can be excellent for a 500-lei-a-month subscription and a guaranteed loss for a product with a 30-lei margin. The figure that actually matters is the cost per paying customer — and between the two sits the qualification rate, which only the company knows.
Without a reporting loop back from sales — how many leads became customers, and at what value — no agency can say whether a campaign was actually profitable. The figures in Ads Manager show the cost up to the lead or click; what happens after that lives in the company's CRM, not in the ad platform.
Why two reports show different numbers for the same campaign
The same campaign can show "47 leads" in Meta Ads Manager and a visibly smaller number in Google Analytics, without either figure being wrong — they measure different time windows. The attribution window is the interval in which a platform gives an ad credit for a conversion. By default, Meta counts conversions within 1 or 7 days of a click and within 1 day of an ad impression; the windows can be changed in Ads Manager, but that's where they start from. Google Ads starts from a 30-day window for click-through conversions and a one-day window for view-through conversions, unless you customise the setting.
In practice, a customer who sees an ad on Monday and buys ten days later shows up as a "result" in Google Ads, but not in Meta, if the click-through window is 7 days. Before you compare the cost per result between platforms, check whether the attribution windows are set similarly — otherwise you're comparing two different counting methods, not two different campaigns.
What to ask an agency before you sign
Before you hand your ad budget to an agency, a few simple questions separate a serious partner from one that just runs campaigns without being accountable for the result.
- 01Full access to ad accounts, in your company's name The Meta, TikTok or LinkedIn Ads account must belong to your company, not the agency — otherwise the history, data and audiences stay locked there if you switch providers.
- 02A written testing plan, not just "let's launch and see" How many message angles, how many formats, how much budget per variant, and how long the test runs, decided before launch.
- 03Reporting on cost per result and cost per customer, not just reach and likes Vanity metrics are easy to make look good; the ones that matter are tied directly to sales.
- 04Clarity about which attribution window is used, and why So you can compare the figures the agency reports with what you see in your CRM, not just what Ads Manager shows.
The list holds regardless of which agency you choose — from a freelancer to studios like The Niche Society — because it separates campaigns run with accountability from ones done just to tick the social media presence box.
When social media advertising is the wrong channel
Part of the marketing budget ends up in social media advertising out of reflex, not strategy — because "everyone does it", not because the target audience actually has buying intent there. It works very well for creating demand, but responds more weakly to demand that already exists. If people are already actively searching for what you sell — "emergency plumber", "accountant for sole traders" — paid or organic search captures that intent more cheaply than an ad that first has to stop the scroll.
Other signs that the timing isn't right: you don't yet have a pixel or a conversions API installed correctly; the sale requires more internal discussion on the customer's side, not a decision made in five minutes on the phone; or the company doesn't yet have the operational capacity to respond to a sudden wave of new requests.
A good campaign doesn't create demand out of nothing — it speeds up a decision the audience would have made anyway, sooner or later. If there isn't already a real need to solve, no testing budget will invent one.
Sources and further reading.
- 01Meta Business Help Center — About attribution models and attribution settings
- 02TikTok for Business Help Center — About daily budgets
- 03LinkedIn Marketing Solutions — Making the Most of Your Budget
- 04Google Ads Help — About conversion windows
- 05Eurostat — Social media use by type, internet advertising and size class of enterprise (isoc_cismt)
Frequently asked questions
What's the minimum cost to start social media campaigns?
Platforms impose technical thresholds, not recommended budgets: TikTok requires a minimum of 20 dollars a day per ad group and 50 dollars a day or in total per campaign, while LinkedIn requires a minimum of 10 dollars a day and 100 dollars total budget for a new campaign. Below these numbers, delivery is unstable. For a real test, with several variants running in parallel, the useful budget is usually several times higher than the platform's technical minimum.
What's the difference between reach, traffic, leads and sales as campaign objectives?
Reach builds a brand's presence, without producing, by itself, anything measurable in sales. Traffic brings clicks to a site, useful for retargeting, but a click isn't buying intent. Leads capture a direct contact — form, message, phone — with a cost per result that's comparable from one month to the next. Sales, optimised for the purchase event, require enough conversion history for the platform to have something to learn from.
How do I know if a cost per lead is good?
There's no universal good or bad cost per lead — it depends on the product's margin and the qualification rate, meaning how many leads it actually takes to produce a paying customer. The figure that really matters is the cost per customer, calculated from sales data, not the cost per lead in Ads Manager.
Does social media advertising work for any business?
No. It works best where the audience isn't already actively searching for the solution, so it can create new demand. If people are already searching for what you sell, paid or organic search captures that intent more cheaply. It also makes less sense when there's no proper conversion measurement in place, when the sale requires more internal discussion on the customer's side, or when the company can't, for the moment, respond quickly to a wave of new leads.
Let's see what can be automated in your business.
A free 30-minute session: we'll tell you what can be automated, how long it takes and what it costs, with a fixed price after discovery.