Search Ads at Scale: Bids, Targets and Profit
See why the most profitable bid is far below break-even, set a target cost per order with profit built in, and learn how to direct automated bidding, with a worked toy model.
In this lesson you will
- Find the profit-maximising bid in a model
- Set a target cost per order with profit built in
- Give automated bidding the right goal
- Know what data it needs
About 25 minutes
More clicks is not the goal
Noor increases her search bid because she wants more traffic. Clicks go up, and so does the bill, and at the end of the month the profit has gone. In lesson 7 she learned the break-even cost per click. This lesson goes a step further. Her platform may offer automated bidding that sets prices for her, and she needs to know what to ask it to do. The answer is rarely ‘get as many clicks as possible’. It is the most profit, or a target cost per order that leaves room for profit.
A model of bidding
To see the shape of the problem, the program below uses an invented model. As the bid rises, clicks rise at first and then flatten, because there is only so much demand. Each click costs the bid, a simplification, since real auctions charge according to their own rules. Each click is worth the margin times the conversion rate. The program tries bids from 10p to 3 pounds and finds the one with the most profit. This is a teaching toy, not how any platform sets prices, but the pattern it shows is real.
import math
# MADE-UP search ad model, pounds. Clicks rise with the bid but flatten out.
margin, conv = 14.0, 0.03
def clicks(bid):
return 800 * (1 - math.exp(-bid / 0.45))
best = None
for cents in range(10, 301, 10):
bid = cents / 100
c = clicks(bid)
cost = c * bid
profit = c * conv * margin - cost
if best is None or profit > best[1]:
best = (bid, profit, c)
if cents in (10, 20, 30, 40, 60, 100, 200):
print(f"bid {bid:.2f}: profit {profit:.1f}")
print("break-even bid:", round(margin * conv, 2))
print("best bid:", best[0])
print("best profit:", round(best[1], 1))
print("clicks at best bid:", round(best[2]))
top = clicks(3.0)
print("clicks at 3.00:", round(top))
print("profit at 3.00:", round(top * conv * margin - top * 3.0, 1))
print("target cost per order at 20% profit:", round(margin * 0.8, 2))
bid 0.10: profit 51.0
bid 0.20: profit 63.2
bid 0.30: profit 46.7
bid 0.40: profit 9.4
bid 0.60: profit -106.0
bid 1.00: profit -413.7
bid 2.00: profit -1249.2
break-even bid: 0.42
best bid: 0.2
best profit: 63.2
clicks at best bid: 287
clicks at 3.00: 799
profit at 3.00: -2061.4
target cost per order at 20% profit: 11.2
At a bid of 10p the profit is 51, at 20p it peaks, and then it falls. The best bid is 0.2 pounds, giving 287 clicks and 63.2 pounds of profit. At 40p the profit is only 9.4, even though the break-even bid is 0.42 pounds: you are close to the line where each extra click earns almost nothing. At a bid of 1 pound it loses 413.7 pounds. At 3 pounds the model buys 799 clicks, hardly more than at a lower bid, and loses 2061.4 pounds. The lesson is that the most profitable bid sits well below the break-even bid, because each extra click costs more than the one before and brings little extra. The last line gives another way to set a limit: if you want a 20 per cent profit on each order, your cost per order must stay under 11.2 pounds.
Working with automated bidding
- Choose the goal that matches your business. Platforms may offer goals such as maximising clicks, conversions or conversion value, or aiming for a target cost per order or return. Names and options change, so read the current help pages.
- Give it good data. Automation learns from conversions. Track real orders, not just clicks (lessons 9 and 17).
- Set a sensible target. If you set a target cost per order below what the market allows, you may get almost no traffic.
- Keep a budget limit. Cap daily spend, and review often at first.
- Check quality. Better ads and pages can lower costs, as the platform’s own help describes.
- Work out margin per order and your conversion rate.
- Compute the break-even cost per click, and a target cost per order with a profit built in.
- Start with a modest budget and a target slightly easier than the one you want.
- After enough orders to judge, tighten the target gradually.
- Stop if the sums stop working, even if the platform says it is optimising.
Maximising clicks or volume with a margin that cannot support it. Automation will happily spend your budget on clicks that do not pay. Tell it what profit looks like.
The click curve here is invented, and real curves differ by market and season. Use the method, not the numbers. Always read the platform’s current rules, and do not break its policies on what you can advertise.
Practice for this lesson
Check what you learned
Five quick questions. Pick an answer and you will see straight away whether it is right and why.
What is the better goal than maximum clicks?
- The most likes
- The highest bid
- The most profit, or a target cost per order
- The most pages
Clicks only help if they pay for themselves.
Where did profit peak in the model?
- At a low bid, well below break-even
- At the break-even bid
- At the highest bid
- At zero
The best bid was 20p, against a 42p break-even.
What happens to clicks at very high bids?
- They double
- They barely increase
- They fall to zero
- They stay fixed
Demand flattens, so extra bid buys little.
What does automated bidding learn from?
- Your logo
- Page colour
- Your followers
- Conversions you track
It needs good conversion data to optimise.
What if the target cost per order is too low?
- You get unlimited traffic
- You may get almost no traffic
- The ads cost more
- Nothing changes
An impossible target starves delivery.
Summary
- Aim for profit, not clicks.
- The most profitable bid sits below break-even.
- Set a target cost per order that leaves a profit.
- Automation needs good conversion data and limits.
Questions
Is the click curve real?
No. It is invented to show the shape. Real curves differ by market, season and competition.
Should I use automated bidding?
It can help once you have enough tracked conversions. Start with a small budget and a limit, and check the platform's current guidance.
Are the figures real?
No. They are made-up numbers in pounds to show the method.
