Ethical Ways to Improve Your InfScore (No Bought Followers)

You improve InfScore by strengthening the public signals the versioned model already scores. Buying followers, running engagement pods, or standing up an empty second profile skips that work, and it often makes the heavier rows worse. On model 2026-08-18.1 the mix is audience quality 32%, engagement 27%, reach & momentum 23%, and platform strength 18%. Every factor has a public lever and a vanity move that stays out of the math. Instagram, TikTok, and YouTube are lookup surfaces of that one composite, not three formulas and not a separate score per app. This guide maps the levers, the blocked tactics, and how to read a later fetch of a live scorecard without inventing an uplift.

What "improve" means on a versioned score

InfScore is a public-signal rating. Search a creator on infscore.com and you get a free 0–100 composite, a tier label, four factor rows, and a scoring-coverage percentage. No account is required. The scorecard says the limit in one line: available factor weights are normalized when provider data is missing. The result is not a percentile, a price, or a campaign forecast.

The current public model id is 2026-08-18.1. If the formula changes later, the version string on the card changes with it. Two scores are comparable only when you also look at that id and at coverage.

"Improve," in this guide, means the public inputs behind a factor moved and a later fetch of the same profile picked that up. It is not a published point reward for a named tactic. The four-factor tour is How to Measure Influencer Performance. The displayed weights are the weights this model uses.

Factor	Weight	Ethical public lever	What does not count
Audience quality	32%	Credible public audience	Bought followers and rented traffic
Engagement	27%	Public rate on the 0–8% scale	Pods and bought comments
Reach & momentum	23%	Log-scale size, plus growth when present	Hollow spikes, kit totals
Platform strength	18%	Another observed public audience	Empty clones and reserved names

Each row is calculated on its own. A strong engagement factor cannot lend points to a missing audience-quality factor. The composite mixes only the rows that returned a number. If a provider signal is missing, the factor is a dash, not a quiet zero. That weight is held out. The remaining observed weights are rebalanced in their published proportions. Missing Data on Creator Metrics is the arithmetic. How to Read an Influencer Scorecard is the order: coverage, dashes, independent rows, then the 0–100 number. You cannot repair a dash from inside InfScore, and you cannot purchase a different mix. The work sits on the public profile a provider can already see.

Audience quality first (32%)

Audience quality is the heaviest factor because a large following without credibility is a noisy input. This row uses provider-measured audience credibility when that public signal exists. On the card it is a 0–100 factor score, or a dash. For the longer split between size and credibility, see Audience Quality vs Follower Count.

A high score means that credibility signal was strong. A lower score means it was weaker. A dash means the input did not arrive. You do not get to type a replacement.

The public work on this row is slow:

• Publish for people who already care about the subject. A rented crowd that leaves after a prize does not become credibility.
• Prefer collaborations whose audiences overlap yours. A mass shoutout from an unrelated account adds names.
• If you run a giveaway, treat it as a reach event. It is a weak way to build the audience this factor measures.
• Stop adding to an old follow-for-follow pile. Another month of real viewers beats a cleanup post.
• Keep the creator identity distinct from brand pages you also run, so the public graph stays one person.

Bought followers usually show up here, when the provider can see them, as a less credible audience. InfScore does not need a confession. If the public audience looks thin, the factor says so. There is no recovery table for a purchased graph.

A live reading on 2026-09-23 shows a present row. Emma Chamberlain on Instagram has audience quality 84.4, engagement 48, reach & momentum 83, and platform strength 45, with coverage at 100%. The composite is 67, Emerging. That 84.4 sits inside a 32% weight. It is not a number you can buy, and it is not a grade of a posting schedule.

The same day's other cards show the gap. Khaby Lame on TikTok and MrBeast on YouTube both have audience quality as a dash. Coverage on each is 68, because 27 + 23 + 18 = 68. A media kit or a logged-in screenshot does not fill that dash. The other rows stay real. They cannot stand in for the missing 32%.

Real engagement, not pods (27%)

Engagement strength converts a public engagement rate onto a fixed 0–8% scale. Eight percent is the top of that scale and maps to 100 on the factor. Values above 8% are capped, so one outlier burst cannot take over the composite. A 4% public rate maps to 50. A 2% rate maps to 25. The conversion is linear until the cap, then it stops. More detail lives in What a Good Engagement Rate Looks Like in 2026.

The scale is the same for every creator on this model. InfScore does not invent a niche benchmark, and it does not compare you with an unpublished peer set. The public percentage is the input.

Pods and bought comments can lift a raw ratio for a moment. They tend to leave audience quality worse, and that row is 32% of the model against engagement's 27%. There is no published point schedule for leaving a pod or for joining one.

Work that changes the public rate without manufacturing it:

• Ask for a reply that takes a sentence, then answer the people who write one.
• Publish some posts worth saving: a tight explainer, a template, a specific recommendation.
• Keep the month closer to even than to one spike. A single viral ratio does not define the row once the cap applies.
• Retire a format that never earns a conversation. Replace it with public work, not a purchased comment block.

The three live cards, checked 2026-09-23, show the scale without ranking who "won" engagement.

Live lookup	Public rate on the card	Engagement factor	How to read it
Emma Chamberlain, Instagram	3.88%	48	Near the published 4% → 50 anchor. One row, not the InfScore.
MrBeast, YouTube	1.95%	24	A modest public rate on the same 0–8% scale.
Khaby Lame, TikTok	8.76%	100	Above 8%, so the factor is capped. Extra rate does not keep adding.

Khaby's 100 is the cap, not a recipe. Emma's 48 and MrBeast's 24 are the same conversion on two other lookup surfaces. If engagement is missing, the factor is a dash and coverage falls by 27, to 73, when the other three factors ran (32 + 23 + 18). InfScore does not copy a rate from another platform to hide that gap. The TikTok comparison of this row against reach is TikTok Reach vs Engagement.

Reach and momentum from real audiences (23%)

Reach is logarithmic. On this model it runs from about 1,000 public followers (0 on that piece) through 100 million (100), which compresses the gap between a large account and a gigantic one.

When a public growth or consistency signal exists, the factor becomes a blend: 70% logarithmic reach and 30% momentum. If growth is missing, reach still scores on its own. Views and watch time stay out of the free composite. They appear in the paid report only when the provider sends them.

Public work that can change this row:

• Post on a cadence you can keep. Momentum comes from provider growth, not from a schedule you announced.
• Build where the work already compounds, then extend. An empty second profile adds nothing to the total.
• A jump of accounts that never return is a weak growth signal even when the headline count moved.
• Totals are summed from observed platform counts. A platform the provider did not return does not add.

YouTube Momentum Signals covers the growth blend on that surface. Subscriber count is the size input. It is not a separate momentum prize.

On the 2026-09-23 YouTube card, MrBeast's reach & momentum is 100, next to about 745.3 million total public followers. Emma's reach row is 83 beside about 14.1 million. Khaby's is 84. The log curve is why those factor gaps are narrower than the follower totals. Subtracting 100 − 83 does not give you a plan.

If the public audience total is missing or not usable, reach & momentum is a dash even when you can recite three follower counts from memory. Coverage then falls by 23, to 77, when the other factors ran (32 + 27 + 18).

Platform strength without empty clones (18%)

Platform strength is the smallest published weight, and it is the row people treat as a distribution strategy. It scores how many public audiences were observed, and whether those audiences are badly lopsided. It does not pick a best network. The dedicated piece is Cross-Platform Creators.

On model 2026-08-18.1 the coverage floors are:

• 1 observed platform → 45
• 2 observed platforms → 70
• 3 or more → 90, plus a small balance bonus when follower counts are not concentrated on a single network

A platform counts when the provider returns a public audience greater than zero. A reserved username, a zero, or a graph that never came back does not count. One observed audience gets the floor and no balance bonus.

Use the same public identity on the networks where you actually publish, so a provider can connect them. Reserving a name so the slot looks full is a different act. Opening three empty accounts will not walk you from 45 to 70 to 90.

Live floors, re-checked 2026-09-23, are three lookups of the same composite. They are not one creator measured three times, and they are not proof of uplift.

Live lookup	Platform strength	Floor reading
Emma Chamberlain, Instagram	45	One observed public audience. No balance bonus.
Khaby Lame, TikTok	70	Two observed audiences. The balance bonus did not lift the row.
MrBeast, YouTube	95	Three or more. The 90 floor plus a small balance bonus.

Emma's 45 means one public audience was observed. That is not a verdict on Instagram as a network, and it is not a nudge to stand up unused handles so the floor can move. The Instagram-facing map of the other rows is Instagram InfScore Factors.

Another observed audience can lift this factor off 45 toward 70, or off 70 toward 90 and a small balance bonus. That new score then takes its 18% place in the mix. It leaves audience quality, engagement, and any dash where they were.

If no usable public audience was observed, the row is a dash. Coverage falls by 18, to 82, when the other three factors ran (32 + 27 + 23).

What does not help (and often hurts)

Plenty of activity changes a number inside an app and still never enters this model. Some of it also makes the public graph worse.

Bought followers. They add a headline count and, when the provider can see the pattern, they weaken audience quality. A hollow jump is not the lever for the 32% row.

Engagement pods and bought comments. A manufactured ratio can flicker on the public rate. The 0–8% scale converts whatever the provider returns, including a rate that will not hold. Pods are not a second engagement formula.

Fake or empty second profiles. Reserved usernames and unused clones do not count as platform strength. The same caption on three empty accounts is not observed coverage.

Private analytics and media kits. The free score uses public creator signals. A logged-in screenshot, a pitch deck, or a field the provider did not return is not an input. Paying $0.99 does not open that dashboard, and it does not fill a dash with a number you pasted.

A homepage tile, or one green row. Example cards on the homepage are illustrative. Cite a live /score/ URL and model 2026-08-18.1, or cite nothing. A platform-strength score of 95 is still 18% when the row exists. An engagement factor of 100 is still 27%. The composite is the rebalanced mix of rows that returned a number.

There is no published recovery score for bought followers or pods. If the public signals later look different, a fresh fetch can show that. If they do not, the number should stay put. No percentage for the wait appears here either.

How to check progress on a live card

Read a later fetch the same way you read the first one: coverage first, then dashes, then each factor, then the 0–100 number and the model id. A changed headline without those facts is not evidence that your work moved the score.

1. Scoring coverage. Full coverage is 100, and the 32/27/23/18 mix applies in full. Audience quality dashed: 68. Engagement dashed: 73. Reach & momentum dashed: 77. Platform strength dashed: 82. A coverage change can move the composite even when the rows you still have did not move.
2. Dashes. A dash is a gap. It is not a zero, and it is not a silent 100. Missing signals stay missing.
3. Each factor on its own. Audience quality cannot borrow engagement. Reach cannot stand in for a missing credibility field. Platform strength cannot fill any of them.
4. The 0–100 composite, after the rows. Tiers are bands on that number: Elite 90–100, Rising 80–89, Emerging 65–79, Developing below 65. Crossing a band is the composite crossing a line. Look at which row moved.
5. Model id. Cite 2026-08-18.1. A comparison across model versions is not a progress report.
6. The same live URL. Do not paste a homepage tile. Do not subtract two different creators and call the difference your result.

A concrete pair of fetches shows why the headline is a poor progress meter. The cross-platform guide recorded a 2026-09-22 reading of MrBeast on YouTube: engagement 27, reach & momentum 100, platform strength 94, audience quality dashed, coverage 68%, composite 69. Re-opened for this guide on 2026-09-23, the same URL shows engagement 24, reach & momentum 100, platform strength 95, audience quality still a dash, coverage still 68%, composite still 69.

Both mixes round to the same composite. The 2026-09-22 mix is (27 × 27 + 100 × 23 + 94 × 18) ÷ 68 = 69.426, which rounds to 69. The 2026-09-23 mix is (24 × 27 + 100 × 23 + 95 × 18) ÷ 68 = 68.5, which rounds to 69. That pair is not a case study, and no cause is assigned here. A screenshot of 69 alone hides the two rows that moved.

The other two cards that day are different people, not a before-and-after. Emma, full coverage: (84.4 × 32 + 48 × 27 + 83 × 23 + 45 × 18) ÷ 100 = 67.158, which rounds to 67. Khaby, audience quality dashed: (100 × 27 + 84 × 23 + 70 × 18) ÷ 68 = 86.647, which rounds to 87. The handle on that URL is khaby.lame. Coverage differs, dashes differ, and the creators differ, so 67, 87, and 69 are not a ranking.

A separate case from the model tests, not from those live cards, uses round numbers. Audience quality missing, engagement at a 4% public rate (50 on the factor), reach & momentum 40, platform strength 45. Coverage is 68. The composite is (50 × 27 + 40 × 23 + 45 × 18) ÷ 68, which rounds to 45. The one-platform floor is not the InfScore. Paying does not fill the audience-quality dash, and an empty second account does not turn that 45 into a reward.

The public InfScore stays free. No signup. Pricing is one report, one dollar, no recurring fee. The $0.99 report unlocks denser provider modules when they exist, including audience, growth, and a branded PDF. Same weights. Paying does not change a factor or turn a dash into a number.

After that order, search yourself on InfScore the way a stranger would. Write down the composite, the coverage, the model id, and any dashes. Change the public behavior behind a real row: who the audience is, what earns a reply, the cadence you can keep, the networks you actually post on. Re-check after a normal month of that work, not after one post and not after a purchase.

This versioned composite uses only available public creator signals. Missing factors stay unavailable and their weight is rebalanced across observed factors. It is not a percentile, a price estimate, or a guarantee of campaign results. Search a public profile on infscore.com. If you want the modules behind a specific scorecard, the full report is there for $0.99, and only when you actually need it.