Methodology
Every number on this site comes from one place and is calculated one way. This page documents both, including what the dataset does not contain, so you can judge whether it fits your use case before relying on it.
What happened in July 2026
This page previously described a dataset of 216 keywords across three advertising platforms. That count matches the legacy production export: 107 Google Search rows, 57 Meta rows and 52 TikTok rows. The source seed files contain only 214 rows, so even the legacy sources disagreed with production. More importantly, none of those 216 CPC values was an observed advertiser-account cost. We found this by auditing our own figures and we are documenting it here rather than quietly replacing them.
Two problems, both now fixed:
- The Meta Ads and TikTok Ads figures were never collected. They were produced by multiplying the Google CPC by a fixed ratio. Any cross-platform comparison drawn from them described the ratio we picked, not the market. Those 109 rows have been removed.
- The Google figures were not from an API either. In the previous dataset, 58% of CPC values ended in a round number of cents. In a genuine API response the last digit is effectively random and roughly 10% land on zero. That distribution is the signature of hand-written figures.
Everything below describes the current dataset, captured in July 2026 and traceable to DataForSEO's API. It is smaller than what came before. It is the first version with a verifiable source.
Where the data comes from
Figures are pulled from DataForSEO (affiliate link), specifically the keywords_data/google_ads/search_volume endpoint, which provides Google Ads planning data. Every refresh is stored as a dated snapshot and never overwritten, so the collection date shown on a page always matches the data on it.
These CPC and bid figures are planning estimates, not observed prices from advertiser accounts. We do not operate the ad accounts they describe. This is third-party planning data, not first-party spend.
What is in the database
The current snapshot holds 98 keywords across 14 industries, all on Google Search at the national United States level. Per-industry coverage ranges from 3 to 10 keywords. That is a small sample and we label it as such on every page where it matters:
| Industry | Keywords | Median CPC | Typical range (p25–p75) |
|---|---|---|---|
| Legal Services | 9 | $89.55 | $44.64 – $163.25 |
| Insurance | 8 | $50.36 | $24.21 – $105.21 |
| B2B / Lead Gen | 5 * | $49.92 | $48.70 – $60.59 |
| SaaS / Software | 10 | $45.30 | $41.12 – $76.36 |
| Education | 7 | $28.48 | $16.49 – $41.23 |
| Home Services | 8 | $23.39 | $13.08 – $32.79 |
| Healthcare | 5 * | $9.51 | $5.09 – $11.01 |
| Financial Services | 10 | $6.93 | $4.72 – $21.20 |
| Automotive | 7 | $5.20 | $3.13 – $6.27 |
| Beauty & Cosmetics | 3 * | $3.15 | $2.44 – $3.19 |
| E-commerce | 4 * | $2.57 | $2.45 – $2.76 |
| Travel & Tourism | 8 | $2.17 | $1.77 – $2.55 |
| Real Estate | 7 | $1.36 | $1.22 – $6.22 |
| Gaming | 7 | $1.26 | $0.65 – $1.68 |
How the figures are calculated
Median, never mean. CPC distributions have long right tails: a handful of terms sit an order of magnitude above the rest, and an average would describe none of the keywords in the set.
Typical range. The range beside each median is the interquartile range, from the 25th to the 75th percentile. Half the keywords in that industry fall inside it, and it describes what to expect far better than a single number.
Top-of-page bid. Returned by the API as a low and high planning estimate. It is not a bid observed in your account and not what you will necessarily pay per click.
Opportunity score. Derived, not measured: (volume × (1 − competition)) ÷ CPC. It ranks reachable demand per dollar. It is our formula, published so you can disagree with it.
Volume trend. The API returns twelve months of monthly search volume per keyword. We sum those across each industry and report the change from the first month to the last. Unlike CPC history, this is genuinely retroactive.
What we do not have
- No Meta or TikTok data. The DataForSEO endpoint used for this snapshot provides Google Ads planning data, not equivalent CPC figures for those platforms, so we cannot publish supported figures for them.
- No CPM or CTR. Not returned by this endpoint. Every figure here is a cost per click.
- No geographic segmentation. National-level United States. Metro costs can differ substantially.
- No cost history yet. Historical CPC cannot be retrieved retroactively. Our series begins with the July 2026 snapshot and grows from there.
- No account-level factors. Quality Score, ad relevance and account history all move your real cost. No benchmark can account for them.
- Small samples. Four industries have fewer than six keywords. Their medians are indicative, not robust, and those pages say so.
How to use these numbers
Benchmarks are useful for setting planning ranges, sanity-checking a media plan, and spotting when your own costs sit far outside the norm for your category. They are not a forecast of what you will pay.
If your measured CPC falls inside the interquartile range for your industry, your costs are unremarkable and your effort is better spent on conversion. If it sits well above the 75th percentile, that gap is worth investigating before you increase budget.
Update cadence
Each refresh is captured as a dated snapshot. Pages state their collection date rather than a quarter label, so a page can never claim to be current when the data behind it is not.
Disclosure
Links to DataForSEO on this site are affiliate links. If you open an account through one and make a payment, we receive a commission. It costs you nothing extra and it does not change what we publish: we used DataForSEO before there was any commission, and the audit above documents us contradicting our own previously published figures. We link them because they are the source of the data, and we would name the source either way.
We do not accept payment for coverage, and no figure on this site is sponsored.
Corrections
If a figure looks wrong to you, we want to know. Email support@cpctools.com with the keyword and what you measured. The audit above started exactly that way, with a number that did not look right.