Industry averages are the wrong planning input for a business running exactly one campaign

A small business planning its first media budget reaches for an industry average, and the published tables oblige. What the tables also show, read carefully, is how little an average constrains. Cost per lead runs from $75.19 in industrial and commercial up to $131.63 in the most expensive vertical measured across twenty-three industries. Inside a single sector the spread is wider still: home services converts paid search at a 7.33% average while general contracting, which most people would place in the same sector, converts at 2.61%. A business running one campaign in one geography does not experience an average of those. It experiences one number, and it will not know that number until it has bought it.

Home services converts at 7.33% and general contracting at 2.61%. A business planning against a sector average is planning against a figure that describes neither of them.

1. 01

   The published spread is wider than any optimisation a small budget can buy

   Between two of the measured verticals the cost per lead differs by roughly three quarters again. No amount of match-type tightening or negative-keyword work on a small account moves a number that far. The planning figure therefore has to come from the business rather than from the table, which means the first month is an instrument rather than a campaign.

2. 02

   Conversion rate decides whether a small budget works, not cost per click

   A 2.61% rate and a 7.33% rate on the same click price produce entirely different businesses. Cost per click is the number small advertisers watch because it updates daily; conversion rate is the number that determines whether the account is viable at all.

3. 03

   A small budget supports one variable at a time

   Two channels, three audiences and four landing pages on a small spend produces no cell with enough volume to read. The model runs one service, one geography and one offer until a baseline exists, then changes one thing.

4. 04

   Search and social answer different questions and the choice is not a split

   Search captures demand that already exists; social creates awareness of demand that does not yet. At small scale, splitting a budget across both usually funds two campaigns that are each too small to learn from. The model reads the published Facebook benchmark set for the same vertical alongside the search set and commits to one, rather than dividing.

Published benchmark tables span a range wider than any optimisation a single small budget can buy; for local SMBs the binding constraint is the review gate (**57% only use businesses rated 4★+**, BrightLocal).

The model starts by narrowing rather than by spending: one service, one geography, one offer, and a single channel chosen by comparing the published search and social benchmark sets for that vertical against whether the demand is already being searched for. Conversion tracking and call tracking go in before the first impression, because a month of untracked spend produces no baseline and cannot be repeated deliberately. Weeks one to four buy the baseline: the account's own cost per click, conversion rate and cost per lead, at whatever volume the budget supports. From week five the model changes one variable at a time against that baseline, and reports cost per booked job rather than cost per form fill, since a form fill is the metric that looks best while meaning least at this size.

1. 1

   Week 1

   Narrow the scope

   One service, one geography, one offer agreed and written down

   Owner: OmniFlow + owner

2. 2

   Week 1

   Channel decision

   Published search and social benchmarks for the vertical compared, one channel committed to

   Owner: OmniFlow

3. 3

   Week 2

   Instrument

   Conversion and call tracking live before the first impression

   Owner: OmniFlow

4. 4

   Weeks 2–6

   Buy the baseline

   Own cost per click, conversion rate and cost per lead established at whatever volume the budget allows

   Owner: OmniFlow

5. 5

   Weeks 6–12

   Single-variable iteration

   One change at a time against the measured baseline

   Owner: OmniFlow

6. 6

   Week 12

   Review

   Cost per booked job against the baseline, not against the published table

   Owner: OmniFlow + owner

The model would expect the business's own first-month figures to sit outside the published range in at least one direction, and treats that as the expected result rather than as a failure of the plan. What it targets is a baseline that is real, measured on the account itself, and stable enough that a single change can be read against it. Every published figure quoted here is a framing device for the first four weeks and is discarded the moment the account produces its own. Nothing above is a forecast of what any particular account will produce.

Modelled. Inputs: WordStream by LocaliQ, Google Ads Benchmarks 2026 (13,474 US search campaigns, April 2025 – March 2026); WordStream by LocaliQ, Facebook Ads Benchmarks 2025 (data April 2024 – June 2025); LocaliQ, 2025 Search Ad Benchmarks for Home Services (3,211 US search campaigns, April 2024 – March 2025). Modelled outputs are not a forecast or a guarantee of results. Cited: WordStream/LocaliQ 2026 (range width); BrightLocal *Local Consumer Review Survey*.

Every figure below is published research, not a client result. They are the inputs to the arithmetic above, listed so it can be checked rather than taken on trust. The bracketed number points to the full citation at the end of this page.

These are the surfaces this engagement is run and measured from, shown with representative figures built around the benchmarks cited on this page. Every account we run reports into views like these, and you keep ownership of all of them.

Each figure on this page, the system it is read from, and the definition and window it is measured over.

Not applicable — this is a modelled engagement. Its weakest input is selection: published benchmarks are drawn from accounts already running managed campaigns at meaningful scale, with conversion history and machine-learning signal behind them. A new account with neither should expect to perform worse than the table before it performs better, and the baseline phase exists so that expectation is set by the account's own data rather than by a number borrowed from advertisers who have been running for years.

A modelled capability, not a client account. Figures illustrate what the model produces and are labelled as modelled wherever they appear. See all our work

Full citations for everything cited on this page, with the sample and period each study covers, so you can go and read the original.