Advertising budget: how much you need and how to split it
The question «how much do we need for advertising» sounds simple, but the answer cannot be borrowed from someone else's experience. It is calculated from your own economics. Here is how to work it out and manage it afterwards.
Calculate from the goal, not from what you have: how many clients you need, your conversion from request, what a request costs — multiply and you have the budget. The floor is 30–50 clicks a week on your key queries, below which no usable statistics accumulate. Increase gradually, by 20–30% at a time.
In this article8 sections
Calculate from the goal, not from what you have
The correct order is the opposite of the habitual one: not «I have $500, what can it buy» but «I need 20 clients, what does that cost».
- How many clients you needTake it from the sales plan. Say, 20 a month.
- What your conversion from request isIf one in four buys, you need 80 requests.
- What a request costsFrom your own statistics or from a test launch. Say, $15.
- Multiply80 × $15 = $1200 a month in advertising budget.
- Check against profitIf 20 clients bring in less than $1200 plus your costs, either the goal or the price needs revisiting.
If you do not know your cost per request, you need to measure it. That means a test run on a minimum budget over two to three weeks. Without that figure the whole calculation is guesswork.
The minimum budget to launch
There is a floor below which advertising fails not because the money is too little but because no statistics accumulate.
- You need at least 30–50 clicks a week on your key queries to understand what is happening.
- Multiply that by the cost per click in your niche — that is your minimum weekly budget.
- The first two to three weeks count as a learning period: results then are worse than they will be later.
A budget too small and spread across broad queries is the most common way to waste money and conclude that «advertising does not work».
How to split between channels
| Stage | Split | The logic |
|---|---|---|
| First launch | 100% into one channel | Find out whether it works at all |
| After the first data | 70% proven, 30% testing | Grow what works, look for what is next |
| Steady operation | 60/25/15 across channels | Diversification and lower risk |
The rule is simple: never test more than one new channel at a time. Otherwise you cannot tell what produced the result.
When to increase the budget
- Cost per request has been stable and acceptable for several weeks running.
- Requests are handled on time — more volume will not mean more lost leads.
- There is headroom: your ads are not already reaching everyone who searches.
- Sales confirm that requests turn into money.
Increase gradually, by 20–30% at a time, and watch whether the cost per request rises. A sharp increase often makes the system start buying more expensive, lower-quality traffic.
When to cut back
- Cost per client has crept close to what a client brings in.
- Requests are no longer handled on time — fix the process first, grow the flow after.
- A seasonal dip: paying full rate in a dead season is rarely worth it.
- Organic traffic has started bringing the same requests for free — advertising on those queries now partly duplicates it.
Switching advertising off entirely «for a month, to save money» usually costs more than it saves: accumulated statistics lose value and automated strategies have to relearn from scratch.
What to account for beyond the budget itself
A planning mistake is counting only the money spent on clicks. The full cost of a channel includes:
- The specialist's work managing it.
- Creating and refreshing creatives.
- Landing pages for the campaigns.
- Staff time spent handling requests.
When you calculate cost per client, all of that belongs in the numerator. Otherwise the figure will be pretty and wrong.
Seasonality and how to plan for it
Almost every niche has months when demand is high and months when there is barely any. A flat budget across the year is nearly always a mistake.
- In season, competition and click prices rise — but so does conversion, so the spend pays back better.
- Off season it makes sense to narrow reach to the most precise queries rather than switch off entirely.
- Preparation for the season starts one to two months ahead: pages, ads, audiences.
- Last year's data is the best reference. Without it, your first year works as reconnaissance.
How to tell the budget is being spent badly
| Sign | What it usually means |
|---|---|
| Many clicks, few requests | Irrelevant queries or a weak landing page |
| Requests but no sales | The wrong audience, or a handling problem |
| Cost per request rising month by month | Ad burnout or growing competition |
| Spending is uneven | Settings too broad and automated strategies without caps |
| The budget runs out by lunchtime | Reach far too wide for a budget that size |
Frequent questions
What percentage of revenue should go to advertising?
There is no universal number. Work from an acceptable cost per client: what you are willing to pay for one and how many you need. The percentage is a consequence, not a starting point.
What if the budget only covers one channel?
Choose the one where your audience is already looking for a solution. For most local services that is search and maps, not social media.
Can we start with a very small budget?
You can, with one caveat: a small budget yields little data and stretches out the uncertain period. Better to launch narrowly on a few queries with a sufficient budget than broadly with a trickle.
Is the advertising budget paid to the contractor?
No. It should go directly to the ad platform from your own account. The contractor is paid separately for their work, and access to the account stays with you.
What if the budget runs out mid-month?
Do not top it up blindly — look at where it went. The cause is usually settings that are too broad: the system is spending on queries you never meant to buy.
Can we plan a budget a year ahead?
In broad figures yes, in detail no. Click prices and competition change, so the plan is revised quarterly against actual data.
The materials answer general questions. We will look at your specific case — free and without obligation.